Red Hat Enterprise Linux 6

Virtualization Administration Guide

Managing your virtual environment

Jiri Herrmann
Scott Radvan

Yehuda Zimmerman
Dayle Parker

Laura Novich

Red Hat Enterprise Linux 6 Virtualization Administration Guide

Managing your virtual environment

Jiri Herrmann
Red Hat Custo mer Co ntent Services
[email protected] m
Yehuda Zimmerman
Red Hat Custo mer Co ntent Services
[email protected] m
Laura No vich
Red Hat Custo mer Co ntent Services
Sco tt Radvan
Red Hat Custo mer Co ntent Services
Dayle Parker
Red Hat Custo mer Co ntent Services

Legal Notice
Co pyright 20 16 Red Hat, Inc.
This do cument is licensed by Red Hat under the Creative Co mmo ns Attributio n-ShareAlike 3.0
Unpo rted License. If yo u distribute this do cument, o r a mo dified versio n o f it, yo u must pro vide
attributio n to Red Hat, Inc. and pro vide a link to the o riginal. If the do cument is mo dified, all Red
Hat trademarks must be remo ved.
Red Hat, as the licenso r o f this do cument, waives the right to enfo rce, and agrees no t to assert,
Sectio n 4 d o f CC-BY-SA to the fullest extent permitted by applicable law.
Red Hat, Red Hat Enterprise Linux, the Shado wman lo go , JBo ss, OpenShift, Fedo ra, the Infinity
lo go , and RHCE are trademarks o f Red Hat, Inc., registered in the United States and o ther
co untries.
Linux is the registered trademark o f Linus To rvalds in the United States and o ther co untries.
Java is a registered trademark o f Oracle and/o r its affiliates.
XFS is a trademark o f Silico n Graphics Internatio nal Co rp. o r its subsidiaries in the United
States and/o r o ther co untries.
MySQL is a registered trademark o f MySQL AB in the United States, the Euro pean Unio n and
o ther co untries.
No de.js is an o fficial trademark o f Jo yent. Red Hat So ftware Co llectio ns is no t fo rmally
related to o r endo rsed by the o fficial Jo yent No de.js o pen so urce o r co mmercial pro ject.
The OpenStack Wo rd Mark and OpenStack lo go are either registered trademarks/service
marks o r trademarks/service marks o f the OpenStack Fo undatio n, in the United States and o ther
co untries and are used with the OpenStack Fo undatio n's permissio n. We are no t affiliated with,
endo rsed o r spo nso red by the OpenStack Fo undatio n, o r the OpenStack co mmunity.
All o ther trademarks are the pro perty o f their respective o wners.

Abstract
The Virtualizatio n Administratio n Guide co vers administratio n o f ho st physical machines,
netwo rking, sto rage, device and guest virtual machine management, and tro ublesho o ting. No te:
This do cument is under develo pment, is subject to substantial change, and is pro vided o nly as
a preview. The included info rmatio n and instructio ns sho uld no t be co nsidered co mplete, and
sho uld be used with cautio n. To expand yo ur expertise, yo u might also be interested in the Red
Hat Enterprise Virtualizatio n (RH318) training co urse.

T able of Cont ent s

T able of Contents
. .hapt
C
. . . .er
. .1. .. Server
. . . . . . Best
. . . . .Pract
. . . . ices
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1. 2. . . . . . . . . .
. .hapt
C
. . . .er
. .2. .. Securit
......y
. .for
. . .Virt
. . . ualiz
. . . . at
. . ion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. 3. . . . . . . . . .
2 .1. Sto rag e Sec urity Is s ues
13
2 .2. SELinux and Virtualiz atio n
13
2 .3. SELinux
15
2 .4. Virtualiz atio n Firewall Info rmatio n
15
. .hapt
C
. . . .er
. .3.
. .sVirt
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1. 7. . . . . . . . . .
3 .1. Sec urity and Virtualiz atio n
18
3 .2. s Virt Lab eling
19
. .hapt
C
. . . .er
. .4. .. KVM
. . . . Live
. . . . Migrat
. . . . . . ion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2. 0. . . . . . . . . .
4 .1. Live Mig ratio n Req uirements
20
4 .2. Live Mig ratio n and Red Hat Enterp ris e Linux Vers io n Co mp atib ility
22
4 .3. Shared Sto rag e Examp le: NFS fo r a Simp le Mig ratio n
4 .4. Live KVM Mig ratio n with virs h
4 .4.1. Ad d itio nal Tip s fo r Mig ratio n with virs h

23
24
26

.4.2. Ad d itio nal O p tio ns fo r the virs h mig rate Co mmand

4
4 .5. Mig rating with virt-manag er

28
29

. .hapt
C
. . . .er
. .5.
. .Remot
. . . . . .e. Management
. . . . . . . . . . . . of
..G
. .uest
. . . .s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
...........
5 .1. Remo te Manag ement with SSH
5 .2. Remo te Manag ement O ver TLS and SSL

36
39

5 .3. Trans p o rt Mo d es

41

. .hapt
C
. . . .er
. .6. .. O
. .vercommit
. . . . . . . . .t .ing
. . .wit
. . .h. KVM
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. 6. . . . . . . . . .
6 .1. Intro d uc tio n
46
6 .2. O verc o mmitting Virtualiz ed CPUs

47

. .hapt
C
. . . .er
. .7. .. KSM
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4. 9. . . . . . . . . .
. .hapt
C
. . . .er
. .8. .. Advanced
. . . . . . . . .G
. .uest
. . . . Virt
. . . ual
. . . .Machine
. . . . . . . Administ
. . . . . . . . rat
. . .ion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
...........
8 .1. Co ntro l G ro up s (c g ro up s )
53
8 .2. Hug e Pag e Sup p o rt
8 .3. Running Red Hat Enterp ris e Linux as a G ues t Virtual Mac hine o n a Hyp er-V Hyp ervis o r

53
54

8 .4. G ues t Virtual Mac hine Memo ry Allo c atio n

8 .5. Auto matic ally Starting G ues t Virtual Mac hines
8 .6 . Dis ab le SMART Dis k Mo nito ring fo r G ues t Virtual Mac hines
8 .7. Co nfig uring a VNC Server

54
55
56
56

8 .8 . G enerating a New Uniq ue MAC Ad d res s

8 .8 .1. Ano ther Metho d to G enerate a New MAC fo r Yo ur G ues t Virtual Mac hine
8 .9 . Imp ro ving G ues t Virtual Mac hine Res p o ns e Time
8 .10 . Virtual Mac hine Timer Manag ement with lib virt

56
57
57
58

8 .10 .1. timer Child Element fo r c lo c k

8 .10 .2. trac k
8 .10 .3. tic kp o lic y
8 .10 .4. freq uenc y, mo d e, and p res ent
8 .10 .5. Examp les Us ing Clo c k Sync hro niz atio n
8 .11. Us ing PMU to Mo nito r G ues t Virtual Mac hine Perfo rmanc e
8 .12. G ues t Virtual Mac hine Po wer Manag ement

59
60
60
60
61
61
62

. .hapt
C
. . . .er
. .9. .. G
. .uest
. . . . virt
. . . ual
. . . machine
. . . . . . . . device
. . . . . . configurat
. . . . . . . . . .ion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. 3. . . . . . . . . .
9 .1. PCI Devic es
64

Virt ualiz at ion Administ rat ion G uide

9 .1.1. As s ig ning a PCI Devic e with virs h

65

9 .1.2.
9 .1.3.
9 .1.4.
9 .1.5.

69
71
74
75

As s ig ning a PCI Devic e with virt-manag er

PCI Devic e As s ig nment with virt-ins tall
Detac hing an As s ig ned PCI Devic e
Creating PCI Brid g es

9 .1.6 . PCI Pas s thro ug h

9 .1.7. Co nfig uring PCI As s ig nment (Pas s thro ug h) with SR-IO V Devic es
9 .1.8 . Setting PCI Devic e As s ig nment fro m a Po o l o f SR-IO V Virtual Func tio ns
9 .2. USB Devic es
9 .2.1. As s ig ning USB Devic es to G ues t Virtual Mac hines

75
76
78
81
81

.2.2. Setting a Limit o n USB Devic e Red irec tio n

9
9 .3. Co nfig uring Devic e Co ntro llers
9 .4. Setting Ad d res s es fo r Devic es
9 .5. Manag ing Sto rag e Co ntro llers in a G ues t Virtual Mac hine

81
82
86
87

9 .6 . Rand o m Numb er G enerato r (RNG ) Devic e

88

. .hapt
C
. . . .er
. .1. 0. .. Q
. .EMU. . . . .img
. . . and
. . . .Q
. .EMU
. . . .G
. .uest
. . . . Agent
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9. 0. . . . . . . . . .
10 .1. Us ing q emu-img
90
10 .2. Q EMU G ues t Ag ent
10 .2.1. Ins tall and Enab le the G ues t Ag ent

95
95

10 .2.2. Setting up Co mmunic atio n b etween G ues t Ag ent and Ho s t

10 .2.3. Us ing the Q EMU G ues t Ag ent

95
96

10 .2.4. Us ing the Q EMU G ues t Ag ent with lib virt

97

10 .2.5. Creating a G ues t Virtual Mac hine Dis k Bac kup

10 .3. Running the Q EMU G ues t Ag ent o n a Wind o ws G ues t

97
98

0 .3.1. Us ing lib virt Co mmand s with the Q EMU G ues t Ag ent o n Wind o ws G ues ts
1
10 .4. Setting a Limit o n Devic e Red irec tio n

10 1
10 1

10 .5. Dynamic ally Chang ing a Ho s t Phys ic al Mac hine o r a Netwo rk Brid g e that is Attac hed to a Virtual
NIC
10 2
. .hapt
C
. . . .er
. .1. 1. .. St
. . orage
. . . . . .Concept
. . . . . . . s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0. 4. . . . . . . . . .
11.1. Sto rag e Po o ls
11.2. Vo lumes

10 4
10 5

. .hapt
C
. . . .er
. .1. 2. .. St
. . orage
. . . . . .Pools
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.0. 7. . . . . . . . . .
12.1. Dis k-b as ed Sto rag e Po o ls
12.1.1. Creating a Dis k-b as ed Sto rag e Po o l Us ing virs h
12.1.2. Deleting a Sto rag e Po o l Us ing virs h
12.2. Partitio n-b as ed Sto rag e Po o ls
12.2.1. Creating a Partitio n-b as ed Sto rag e Po o l Us ing virt-manag er

110
111
111

12.2.2. Deleting a Sto rag e Po o l Us ing virt-manag er

114

12.2.3. Creating a Partitio n-b as ed Sto rag e Po o l Us ing virs h

12.2.4. Deleting a Sto rag e Po o l Us ing virs h

115
117

12.3. Direc to ry-b as ed Sto rag e Po o ls

12.3.1. Creating a Direc to ry-b as ed Sto rag e Po o l with virt-manag er

118
118

12.3.2. Deleting a Sto rag e Po o l Us ing virt-manag er

121

12.3.3. Creating a Direc to ry-b as ed Sto rag e Po o l with virs h

12.3.4. Deleting a Sto rag e Po o l Us ing virs h

122
124

12.4. LVM-b as ed Sto rag e Po o ls

12.4.1. Creating an LVM-b as ed Sto rag e Po o l with virt-manag er

124
125

12.4.2. Deleting a Sto rag e Po o l Us ing virt-manag er

130

12.4.3. Creating an LVM-b as ed Sto rag e Po o l with virs h

12.4.4. Deleting a Sto rag e Po o l Us ing virs h

131
133

12.5. iSCSI-b as ed Sto rag e Po o ls

10 7
10 8

133

T able of Cont ent s

12.5.1. Co nfig uring a So ftware iSCSI Targ et

133

12.5.2. Ad d ing an iSCSI Targ et to virt-manag er

137

12.5.3. Deleting a Sto rag e Po o l Us ing virt-manag er

12.5.4. Creating an iSCSI-b as ed Sto rag e Po o l with virs h

140
141

12.5.5. Deleting a Sto rag e Po o l Us ing virs h

143

12.6 . NFS-b as ed Sto rag e Po o ls

12.6 .1. Creating an NFS-b as ed Sto rag e Po o l with virt-manag er

143
143

2.6 .2. Deleting a Sto rag e Po o l Us ing virt-manag er

1
12.7. G lus terFS Sto rag e Po o ls

146
147

12.8 . Us ing an NPIV Virtual Ad ap ter (vHBA) with SCSI Devic es

147

12.8 .1. Creating a vHBA

12.8 .2. Creating a Sto rag e Po o l Us ing the vHBA

148
150

12.8 .3. Co nfig uring the Virtual Mac hine to Us e a vHBA LUN
12.8 .4. Des tro ying the vHBA Sto rag e Po o l

151
152

. .hapt
C
. . . .er
. .1. 3.
. . .Volumes
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. 53
...........
13.1. Creating Vo lumes

153

13.2. Clo ning Vo lumes

13.3. Ad d ing Sto rag e Devic es to G ues ts

153
154

13.3.1. Ad d ing File-b as ed Sto rag e to a G ues t

13.3.2. Ad d ing Hard Drives and O ther Blo c k Devic es to a G ues t
13.4. Deleting and Remo ving Vo lumes

154
157
158

. .hapt
C
. . . .er
. .1. 4. .. Managing
. . . . . . . . . guest
. . . . . . virt
. . . ual
. . . machines
. . . . . . . . . wit
. . .h. virsh
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.6. 0. . . . . . . . . .
14.1. G eneric Co mmand s
14.1.1. help

16 0
16 0

14.1.2. q uit and exit

16 1

14.1.3. vers io n
14.1.4. Arg ument Dis p lay

16 1
16 1

14.1.5. c o nnec t
14.1.6 . Dis p laying Bas ic Info rmatio n

16 1
16 2

14.1.7. Injec ting NMI

16 2

14.2. Attac hing and Up d ating a Devic e with virs h

14.3. Attac hing Interfac e Devic es

16 2
16 3

14.4. Chang ing the Med ia o f a CDRO M

14.5. Do main Co mmand s

16 4
16 4

14.5.1. Co nfig uring a Do main to b e Started Auto matic ally at Bo o t

14.5.2. Co nnec ting the Serial Co ns o le fo r the G ues t Virtual Mac hine

16 4
16 4

14.5.3. Defining a Do main with an XML File

16 5

14.5.4. Ed iting and Dis p laying a Des c rip tio n and Title o f a Do main

16 5

14.5.5. Dis p laying Devic e Blo c k Statis tic s

16 5

14.5.6 . Retrieving Netwo rk Statis tic s

14.5.7. Mo d ifying the Link State o f a Do main' s Virtual Interfac e

16 6
16 6

14.5.8 . Lis ting the Link State o f a Do main' s Virtual Interfac e

16 6

14.5.9 . Setting Netwo rk Interfac e Band wid th Parameters

16 6

14.5.10 . Retrieving Memo ry Statis tic s fo r a Running Do main

16 7

14.5.11. Dis p laying Erro rs o n Blo c k Devic es

14.5.12. Dis p laying the Blo c k Devic e Siz e

16 7
16 7

14.5.13. Dis p laying the Blo c k Devic es As s o c iated with a Do main

16 7

14.5.14. Dis p laying Virtual Interfac es As s o c iated with a Do main

16 8

14.5.15. Us ing b lo c kc o mmit to Sho rten a Bac king Chain

16 8

14.5.16 . Us ing b lo c kp ull to Sho rten a Bac king Chain

14.5.17. Us ing b lo c kres iz e to Chang e the Siz e o f a Do main Path

16 8
170

14.5.18 . Dis k Imag e Manag ement with Live Blo c k Co p y

170

Virt ualiz at ion Administ rat ion G uide

14.5.18 . Dis k Imag e Manag ement with Live Blo c k Co p y

170

14.5.19 . Dis p laying a URI fo r Co nnec tio n to a G rap hic al Dis p lay

172

14.5.20 . Do main Retrieval Co mmand s

172

14.5.21. Co nverting Q EMU Arg uments to Do main XML

14.5.22. Creating a Dump File o f a Do main' s Co re

173
174

14.5.23. Creating a Virtual Mac hine XML Dump (Co nfig uratio n File)

174

14.5.24. Creating a G ues t Virtual Mac hine fro m a Co nfig uratio n File

176

14.6 . Ed iting a G ues t Virtual Mac hine' s c o nfig uratio n file

176

14.6 .1. Ad d ing Multifunc tio n PCI Devic es to KVM G ues t Virtual Mac hines

176

14.6 .2. Sto p p ing a Running Do main to Res tart It Later

14.6 .3. Dis p laying CPU Statis tic s fo r a Sp ec ified Do main

177
177

14.6 .4. Saving a Sc reens ho t

178

14.6 .5. Send ing a Keys tro ke Co mb inatio n to a Sp ec ified Do main

178

14.6 .6 . Send ing Pro c es s Sig nal Names to Virtual Pro c es s es

179

4.6 .7. Dis p laying the IP Ad d res s and Po rt Numb er fo r the VNC Dis p lay
1
14.7. NUMA No d e Manag ement

179
179

14.7.1. Dis p laying No d e Info rmatio n

179

14.7.2. Setting NUMA Parameters

179

14.7.3. Dis p laying the Amo unt o f Free Memo ry in a NUMA Cell

18 0

14.7.4. Dis p laying a CPU Lis t

14.7.5. Dis p laying CPU Statis tic s

18 0
18 0

14.7.6 . Sus p end ing the Ho s t Phys ic al Mac hine

18 1

14.7.7. Setting and Dis p laying the No d e Memo ry Parameters

18 1

14.7.8 . Creating Devic es o n Ho s t No d es

18 1

14.7.9 . Detac hing a No d e Devic e

14.7.10 . Retrieving a Devic e' s Co nfig uratio n Setting s

18 1
18 1

14.7.11. Lis ting Devic es o n a No d e

18 2

14.7.12. Trig g ering a Res et fo r a No d e

18 2

14.8 . Starting , Sus p end ing , Res uming , Saving , and Res to ring a G ues t Virtual Mac hine

18 2

14.8 .1. Starting a Defined Do main

14.8 .2. Sus p end ing a G ues t Virtual Mac hine

18 2
18 3

14.8 .3. Sus p end ing a Running Do main

18 3

14.8 .4. Waking Up a Do main fro m a p ms us p end State

18 3

14.8 .5. Und efining a Do main

18 3

14.8 .6 . Res uming a G ues t Virtual Mac hine

14.8 .7. Save a G ues t Virtual Mac hine

18 4
18 4

14.8 .8 . Up d ating the Do main XML File that will b e Us ed fo r Res to ring the G ues t

18 5

14.8 .9 . Extrac ting the Do main XML File

18 5

14.8 .10 . Ed it Do main XML Co nfig uratio n Files

18 5

14.8 .11. Res to re a G ues t Virtual Mac hine

18 5

14.9 . Shutting Do wn, Reb o o ting , and Fo rc ing Shutd o wn o f a G ues t Virtual Mac hine
14.9 .1. Shutting Do wn a G ues t Virtual Mac hine

18 6
18 6

14.9 .2. Shutting Do wn Red Hat Enterp ris e Linux 6 G ues ts o n a Red Hat Enterp ris e Linux 7 Ho s t
14.9 .3. Manip ulating the lib virt-g ues ts Co nfig uratio n Setting s

18 6

14.9 .4. Reb o o ting a G ues t Virtual Mac hine

19 0

14.9 .5. Fo rc ing a G ues t Virtual Mac hine to Sto p

14.9 .6 . Res etting a Virtual Mac hine

19 1
19 1

14.10 . Retrieving G ues t Virtual Mac hine Info rmatio n

19 1

14.10 .1. G etting the Do main ID o f a G ues t Virtual Mac hine

19 1

14.10 .2. G etting the Do main Name o f a G ues t Virtual Mac hine

19 1

14.10 .3. G etting the UUID o f a G ues t Virtual Mac hine

14.10 .4. Dis p laying G ues t Virtual Mac hine Info rmatio n

19 1
19 2

14.11. Sto rag e Po o l Co mmand s

14.11.1. Searc hing fo r a Sto rag e Po o l XML

18 8

19 2
19 2

T able of Cont ent s

14.11.1. Searc hing fo r a Sto rag e Po o l XML

19 2

14.11.2. Creating , Defining , and Starting Sto rag e Po o ls

19 3

14.11.2.1. Build ing a s to rag e p o o l

14.11.2.2. Creating and d efining a s to rag e p o o l fro m an XML file

19 3
19 3

14.11.2.3. Creating and s tarting a s to rag e p o o l fro m raw p arameters

19 4

14.11.2.4. Auto -s tarting a s to rag e p o o l

19 4

14.11.3. Sto p p ing and Deleting Sto rag e Po o ls

19 4

14.11.4. Creating an XML Dump File fo r a Sto rag e Po o l

14.11.5. Ed iting the Sto rag e Po o l' s Co nfig uratio n File

19 4
19 4

14.11.6 . Co nverting Sto rag e Po o ls

19 5

14.12. Sto rag e Vo lume Co mmand s

14.12.1. Creating Sto rag e Vo lumes
14.12.1.1. Creating a s to rag e vo lume fro m an XML file
14.12.1.2. Clo ning a s to rag e vo lume

19 5
19 5
19 5
19 6

14.12.2. Deleting Sto rag e Vo lumes

19 6

14.12.3. Dump ing Sto rag e Vo lume Info rmatio n to an XML File

19 7

14.12.4. Lis ting Vo lume Info rmatio n

19 7

14.12.5. Retrieving Sto rag e Vo lume Info rmatio n

14.12.6 . Up lo ad ing and Do wnlo ad ing Sto rag e Vo lumes

19 7
19 7

14.12.6 .1. Up lo ad ing c o ntents to a s to rag e vo lume

19 7

14.12.6 .2. Do wnlo ad ing the c o ntents fro m a s to rag e vo lume

19 8

14.12.7. Re-s iz ing Sto rag e Vo lumes

14.13. Dis p laying Per-g ues t Virtual Mac hine Info rmatio n

19 8
19 8

14.13.1. Dis p laying the G ues t Virtual Mac hines

14.13.2. Dis p laying Virtual CPU Info rmatio n

19 8
20 0

14.13.3. Co nfig uring Virtual CPU Affinity

20 0

14.13.4. Dis p laying Info rmatio n ab o ut the Virtual CPU Co unts o f a Do main

20 1

14.13.5. Co nfig uring Virtual CPU Affinity

20 1

14.13.6 . Co nfig uring Virtual CPU Co unt

14.13.7. Co nfig uring Memo ry Allo c atio n

20 1
20 3

14.13.8 . Chang ing the Memo ry Allo c atio n fo r the Do main

20 4

14.13.9 . Dis p laying G ues t Virtual Mac hine Blo c k Devic e Info rmatio n

20 4

14.13.10 . Dis p laying G ues t Virtual Mac hine Netwo rk Devic e Info rmatio n

20 4

14.14. Manag ing Virtual Netwo rks

14.15. Mig rating G ues t Virtual Mac hines with virs h
14.15.1. Interfac e Co mmand s

20 4
20 5
20 5

14.15.1.1. Defining and s tarting a ho s t p hys ic al mac hine interfac e via an XML file

20 6

14.15.1.2. Ed iting the XML c o nfig uratio n file fo r the ho s t interfac e

20 6

14.15.1.3. Lis ting ac tive ho s t interfac es

14.15.1.4. Co nverting a MAC ad d res s into an interfac e name

20 6
20 6

14.15.1.5. Sto p p ing a s p ec ific ho s t p hys ic al mac hine interfac e

20 6

14.15.1.6 . Dis p laying the ho s t c o nfig uratio n file

20 6

14.15.1.7. Creating b rid g e d evic es

20 6

14.15.1.8 . Tearing d o wn a b rid g e d evic e

14.15.1.9 . Manip ulating interfac e s nap s ho ts

20 7
20 7

14.15.2. Manag ing Snap s ho ts

20 7

14.15.2.1. Creating Snap s ho ts

20 7

14.15.2.2. Creating a s nap s ho t fo r the c urrent d o main

20 8

14.15.2.3. Taking a s nap s ho t o f the c urrent d o main

14.15.2.4. s nap s ho t-ed it-d o main

20 9
20 9

14.15.2.5. s nap s ho t-info -d o main

20 9

14.15.2.6 . s nap s ho t-lis t-d o main

20 9

14.15.2.7. s nap s ho t-d ump xml d o main s nap s ho t

210

14.15.2.8 . s nap s ho t-p arent d o main

210

Virt ualiz at ion Administ rat ion G uide

14.15.2.8 . s nap s ho t-p arent d o main

210

14.15.2.9 . s nap s ho t-revert d o main

14.15.2.10 . s nap s ho t-d elete d o main

211
211

14.16 . G ues t Virtual Mac hine CPU Mo d el Co nfig uratio n

212

14.16 .1. Intro d uc tio n

212

14.16 .2. Learning ab o ut the Ho s t Phys ic al Mac hine CPU Mo d el

212

4.16 .3. Determining a Co mp atib le CPU Mo d el to Suit a Po o l o f Ho s t Phys ic al Mac hines

1
14.17. Co nfig uring the G ues t Virtual Mac hine CPU Mo d el

212
215

14.18 . Manag ing Res o urc es fo r G ues t Virtual Mac hines

215

14.19 . Setting Sc hed ule Parameters

216

14.20 . Dis p lay o r Set Blo c k I/O Parameters

217

14.21. Co nfig uring Memo ry Tuning

14.22. Virtual Netwo rking Co mmand s

217
217

14.22.1. Auto s tarting a Virtual Netwo rk

217

14.22.2. Creating a Virtual Netwo rk fro m an XML File

218

14.22.3. Defining a Virtual Netwo rk fro m an XML File

218

14.22.4. Sto p p ing a Virtual Netwo rk

14.22.5. Creating a Dump File

218
218

14.22.6 . Ed iting a Virtual Netwo rk' s XML Co nfig uratio n File

218

14.22.7. G etting Info rmatio n ab o ut a Virtual Netwo rk

219

14.22.8 . Lis ting Info rmatio n ab o ut a Virtual Netwo rk

14.22.9 . Co nverting a Netwo rk UUID to Netwo rk Name

219
219

14.22.10 . Starting a (Previo us ly Defined ) Inac tive Netwo rk

14.22.11. Und efining the Co nfig uratio n fo r an Inac tive Netwo rk

219
219

14.22.12. Co nverting a Netwo rk Name to Netwo rk UUID

14.22.13. Up d ating an Exis ting Netwo rk Definitio n File

219
219

. .hapt
C
. . . .er
. .1. 5.
. . Managing
. . . . . . . . .G
. .uest
. . . .s. wit
. . .h. .t he
. . . Virt
. . . ual
. . . Machine
. . . . . . . . Manager
. . . . . . . . (virt
. . . .- .manager)
. . . . . . . . . . . . . . . . . . . . .2.2. 1. . . . . . . . . .
15.1. Starting virt-manag er
221
15.2. The Virtual Mac hine Manag er Main Wind o w
15.3. The Virtual Hard ware Details Wind o w

222
222

15.3.1. Attac hing USB Devic es to a G ues t Virtual Mac hine

15.4. Virtual Mac hine G rap hic al Co ns o le
15.5. Ad d ing a Remo te Co nnec tio n

224
226
228

15.6 . Dis p laying G ues t Details

15.7. Perfo rmanc e Mo nito ring

229
236

15.8 . Dis p laying CPU Us ag e fo r G ues ts

15.9 . Dis p laying CPU Us ag e fo r Ho s ts
15.10 . Dis p laying Dis k I/O

237
238
239

15.11. Dis p laying Netwo rk I/O

240

. .hapt
C
. . . .er
. .1. 6. .. G
. .uest
. . . . Virt
. . . ual
. . . .Machine
. . . . . . . Disk
. . . . .Access
. . . . . . wit
. . .h. .O. ffline
. . . . . T. ools
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.4. 4. . . . . . . . . .
16 .1. Intro d uc tio n
244
16 .2. Termino lo g y
16 .3. Ins tallatio n
16 .4. The g ues tfis h Shell

245
246
246

16 .4.1. Viewing File Sys tems with g ues tfis h

16 .4.1.1. Manual Lis ting and Viewing

247
247

16 .4.1.2. Via g ues tfis h ins p ec tio n

16 .4.1.3. Ac c es s ing a g ues t virtual mac hine b y name
16 .4.2. Mo d ifying Files with g ues tfis h

248
249
249

16 .4.3. O ther Ac tio ns with g ues tfis h

16 .4.4. Shell Sc rip ting with g ues tfis h

249
249

16 .4.5. Aug eas and lib g ues tfs Sc rip ting

250

T able of Cont ent s

16 .5. O ther Co mmand s
16 .6 . virt-res c ue: The Res c ue Shell
16 .6 .1. Intro d uc tio n
16 .6 .2. Running virt-res c ue

251
251
251
252

16 .7. virt-d f: Mo nito ring Dis k Us ag e

16 .7.1. Intro d uc tio n
16 .7.2. Running virt-d f

253
253
253

16 .8 . virt-res iz e: Res iz ing G ues t Virtual Mac hines O ffline

16 .8 .1. Intro d uc tio n

254
254

6 .8 .2. Exp and ing a Dis k Imag e

1
16 .9 . virt-ins p ec to r: Ins p ec ting G ues t Virtual Mac hines
16 .9 .1. Intro d uc tio n

254
256
256

16 .9 .2. Ins tallatio n

16 .9 .3. Running virt-ins p ec to r

256
256

16 .10 . virt-win-reg : Read ing and Ed iting the Wind o ws Reg is try
16 .10 .1. Intro d uc tio n
16 .10 .2. Ins tallatio n

258
258
258

6 .10 .3. Us ing virt-win-reg

1
16 .11. Us ing the API fro m Pro g ramming Lang uag es

258
259

6 .11.1. Interac tio n with the API via a C Pro g ram

1
16 .12. virt-s ys p rep : Res etting Virtual Mac hine Setting s
16 .13. Tro ub les ho o ting

26 0
26 4
26 7

16 .14. Where to Find Further Do c umentatio n

26 7

. .hapt
C
. . . .er
. .1. 7. .. Using
. . . . . .Simple
. . . . . .T
. .ools
. . . .for
. . .G. uest
. . . . .Virt
. . . ual
. . . Machine
. . . . . . . . Management
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.6. 8. . . . . . . . . .
17.1. Us ing virt-viewer
26 8
17.2. remo te-viewer

26 9

. .hapt
C
. . . .er
. .1. 8. .. Virt
. . . ual
. . . .Net
. . .working
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.7. 1. . . . . . . . . .
18 .1. Virtual Netwo rk Switc hes
271
18 .2. Brid g ed Mo d e
18 .3. Netwo rk Ad d res s Trans latio n Mo d e
18 .3.1. DNS and DHCP

272
273
274

18 .4. Ro uted Mo d e
18 .5. Is o lated Mo d e

275
276

18 .6 . The Default Co nfig uratio n

18 .7. Examp les o f Co mmo n Sc enario s
18 .7.1. Brid g ed Mo d e

277
278
278

18 .7.2. Ro uted Mo d e
18 .7.3. NAT Mo d e

279
28 0

8 .7.4. Is o lated Mo d e
1
18 .8 . Manag ing a Virtual Netwo rk
18 .9 . Creating a Virtual Netwo rk

28 1
28 1
28 2

18 .10 . Attac hing a Virtual Netwo rk to a G ues t

18 .11. Attac hing Direc tly to a Phys ic al Interfac e

28 9
29 3

18 .12. Ap p lying Netwo rk Filtering

18 .12.1. Intro d uc tio n
18 .12.2. Filtering Chains

29 5
29 5
29 6

18 .12.3. Filtering Chain Prio rities

18 .12.4. Us ag e o f Variab les in Filters

29 8
29 8

18 .12.5. Auto matic IP Ad d res s Detec tio n and DHCP Sno o p ing
18 .12.5.1. Intro d uc tio n
18 .12.5.2. DHCP Sno o p ing

30 0
30 0
30 1

18 .12.6 . Res erved Variab les

30 2

Virt ualiz at ion Administ rat ion G uide

18 .12.6 . Res erved Variab les
18 .12.7. Element and Attrib ute O verview

30 2
30 2

18 .12.8 . Referenc es to O ther Filters

18 .12.9 . Filter Rules
18 .12.10 . Sup p o rted Pro to c o ls

30 2
30 3
30 4

18 .12.10 .1. MAC (Ethernet)

18 .12.10 .2. VLAN (8 0 2.1Q )

30 5
30 5

18 .12.10 .3. STP (Sp anning Tree Pro to c o l)

18 .12.10 .4. ARP/RARP
18 .12.10 .5. IPv4

30 5
30 6
30 7

18 .12.10 .6 . IPv6
18 .12.10 .7. TCP/UDP/SCTP

30 8
30 8

18 .12.10 .8 . ICMP
18 .12.10 .9 . IG MP, ESP, AH, UDPLITE, ' ALL'
18 .12.10 .10 . TCP/UDP/SCTP o ver IPV6

30 9
310
311

18 .12.10 .11. ICMPv6

18 .12.10 .12. IG MP, ESP, AH, UDPLITE, ' ALL' o ver IPv6

312
312

18 .12.11. Ad vanc ed Filter Co nfig uratio n To p ic s

18 .12.11.1. Co nnec tio n trac king
18 .12.11.2. Limiting Numb er o f Co nnec tio ns
18 .12.11.3. Co mmand line to o ls
18 .12.11.4. Pre-exis ting netwo rk filters
18 .12.11.5. Writing yo ur o wn filters
18 .12.11.6 . Samp le c us to m filter
18 .12.12. Limitatio ns

313
313
314
315
315
316
318
321

18 .13. Creating Tunnels

18 .13.1. Creating Multic as t Tunnels

322
322

18 .13.2. Creating TCP Tunnels

18 .14. Setting vLAN Tag s
18 .15. Ap p lying Q o S to Yo ur Virtual Netwo rk

322
323
324

. .hapt
C
. . . .er
. .1. 9. .. qemu. . . . . .kvm
. . . .Commands,
. . . . . . . . . . Flags,
. . . . . . and
. . . .Argument
. . . . . . . . .s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
. . 5. . . . . . . . . .
19 .1. Intro d uc tio n
Whitelis t Fo rmat

325
325

19 .2. Bas ic O p tio ns

E mulated Mac hine
P ro c es s o r Typ e

325
325
325

P ro c es s o r To p o lo g y
N UMA Sys tem

326
326

M emo ry Siz e
K eyb o ard Layo ut

326
326

G ues t Name
G ues t UUID
19 .3. Dis k O p tio ns

326
326
326

G eneric Drive
B o o t O p tio n

326
327

nap s ho t Mo d e
S
19 .4. Dis p lay O p tio ns
D is ab le G rap hic s
V G A Card Emulatio n
V NC Dis p lay
p ic e Des kto p
S
19 .5. Netwo rk O p tio ns
T AP netwo rk

328
328
328
328
328
329
329
329

T able of Cont ent s

19 .6 . Devic e O p tio ns
G eneral Devic e

330
330

G lo b al Devic e Setting
C harac ter Devic e
E nab le USB

337
337
337

19 .7. Linux/Multib o o t Bo o t
K ernel File

337
337

R am Dis k
C o mmand Line Parameter
19 .8 . Exp ert O p tio ns

338
338
338

K VM Virtualiz atio n
D is ab le Kernel Mo d e PIT Reinjec tio n

338
338

N o Shutd o wn
N o Reb o o t
S erial Po rt, Mo nito r, Q MP

338
338
338

M o nito r Red irec t

M anual CPU Start

339
339

R TC
Watc hd o g
Watc hd o g Reac tio n

339
339
339

G ues t Memo ry Bac king

S MBIO S Entry

339
339

19 .9 . Help and Info rmatio n O p tio ns

H elp
V ers io n

339
339
339

ud io Help
A
19 .10 . Mis c ellaneo us O p tio ns

339
340

M ig ratio n
N o Default Co nfig uratio n
D evic e Co nfig uratio n File

340
340
340

L o ad ed Saved State

340

. .hapt
C
. . . .er
. .2. 0. .. Manipulat
. . . . . . . . . ing
. . . t. he
. . .Domain
. . . . . . . XML
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
. . 1. . . . . . . . . .
2 0 .1. G eneral Info rmatio n and Metad ata
341
2 0 .2. O p erating Sys tem Bo o ting
2 0 .2.1. BIO S Bo o tlo ad er
2 0 .2.2. Ho s t Phys ic al Mac hine Bo o tlo ad er

342
342
344

0 .2.3. Direc t kernel b o o t

2
2 0 .3. SMBIO S Sys tem Info rmatio n

344
345

2 0 .4. CPU Allo c atio n

2 0 .5. CPU Tuning
2 0 .6 . Memo ry Bac king

346
346
348

2 0 .7. Memo ry tuning

2 0 .8 . NUMA No d e Tuning

349
350

2 0 .9 . Blo c k I/O tuning

2 0 .10 . Res o urc e Partitio ning
2 0 .11. CPU Mo d el and To p o lo g y

350
351
352

0 .11.1. G ues t virtual mac hine NUMA to p o lo g y

2
2 0 .12. Events Co nfig uratio n

356
356

2 0 .13. Po wer Manag ement

2 0 .14. Hyp ervis o r Features

358
359

2 0 .15. Timekeep ing

2 0 .16 . Devic es
2 0 .16 .1. Hard Drives , Flo p p y Dis ks , CDRO Ms

36 0
36 2
36 3

Virt ualiz at ion Administ rat ion G uide

2 0 .16 .1. Hard Drives , Flo p p y Dis ks , CDRO Ms

36 3

2 0 .16 .1.1. Dis k element

2 0 .16 .1.2. So urc e element

36 4
36 5

2 0 .16 .1.3. Mirro r element

2 0 .16 .1.4. Targ et element
2 0 .16 .1.5. io tune

36 5
36 5
36 6

2 0 .16 .1.6 . d river

2 0 .16 .1.7. Ad d itio nal d evic e elements

36 6
36 7

2 0 .16 .2. Files ys tems

2 0 .16 .3. Devic e Ad d res s es
2 0 .16 .4. Co ntro llers

36 8
370
371

2 0 .16 .5. Devic e Leas es

2 0 .16 .6 . Ho s t Phys ic al Mac hine Devic e As s ig nment

372
373

2 0 .16 .6 .1. USB / PCI Devic es

2 0 .16 .6 .2. Blo c k / c harac ter d evic es
2 0 .16 .7. Red irec ted Devic es

373
375
377

2 0 .16 .8 . Smartc ard Devic es

2 0 .16 .9 . Netwo rk Interfac es

378
379

2 0 .16 .9 .1. Virtual netwo rks

2 0 .16 .9 .2. Brid g e to LAN
2 0 .16 .9 .3. Setting a p o rt mas q uerad ing rang e

38 0
38 1
38 2

2 0 .16 .9 .4. Us ers p ac e SLIRP s tac k

2 0 .16 .9 .5. G eneric Ethernet c o nnec tio n

38 2
38 3

2 0 .16 .9 .6 . Direc t attac hment to p hys ic al interfac es

2 0 .16 .9 .7. PCI p as s thro ug h
2 0 .16 .9 .8 . Multic as t tunnel

38 3
38 6
38 7

2 0 .16 .9 .9 . TCP tunnel

2 0 .16 .9 .10 . Setting NIC d river-s p ec ific o p tio ns

38 7
38 8

2 0 .16 .9 .11. O verrid ing the targ et element

2 0 .16 .9 .12. Sp ec ifying b o o t o rd er
2 0 .16 .9 .13. Interfac e RO M BIO S c o nfig uratio n

38 9
39 0
39 0

2 0 .16 .9 .14. Q uality o f s ervic e

2 0 .16 .9 .15. Setting VLAN tag (o n s up p o rted netwo rk typ es o nly)

39 1
39 2

0 .16 .9 .16 . Mo d ifying virtual link s tate

2
2 0 .16 .10 . Inp ut Devic es
2 0 .16 .11. Hub Devic es

39 2
39 3
39 3

2 0 .16 .12. G rap hic al frameb uffers

2 0 .16 .13. Vid eo Devic es

39 4
39 7

2 0 .16 .14. Co ns o les , Serial, Parallel, and Channel Devic es

2 0 .16 .15. G ues t Virtual Mac hine Interfac es
2 0 .16 .16 . Channel

39 8
39 9
40 1

0 .16 .17. Ho s t Phys ic al Mac hine Interfac e

2
2 0 .17. So und Devic es

40 2
40 6

2 0 .18 . Watc hd o g Devic e

2 0 .19 . Memo ry Ballo o n Devic e
2 0 .20 . Sec urity Lab el

40 7
40 8
40 9

2 0 .21. Examp le Do main XML Co nfig uratio n

410

. .hapt
C
. . . .er
. .2. 1. .. T
. .roubleshoot
. . . . . . . . . . .ing
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4.1. 2. . . . . . . . . .
2 1.1. Deb ug g ing and Tro ub les ho o ting To o ls
412

10

2 1.2. Prep aring fo r Dis as ter Rec o very

2 1.3. Creating virs h Dump Files
2 1.4. kvm_s tat

413
414
415

2 1.5. G ues t Virtual Mac hine Fails to Shutd o wn

418

T able of Cont ent s

2 1.5. G ues t Virtual Mac hine Fails to Shutd o wn
2 1.6 . Tro ub les ho o ting with Serial Co ns o les

418
419

2 1.7. Virtualiz atio n Lo g Files

2 1.8 . Lo o p Devic e Erro rs
2 1.9 . Live Mig ratio n Erro rs

420
420
420

2 1.10 . Enab ling Intel VT-x and AMD-V Virtualiz atio n Hard ware Extens io ns in BIO S
2 1.11. KVM Netwo rking Perfo rmanc e

420
421

2 1.12. Wo rkaro und fo r Creating External Snap s ho ts with lib virt

2 1.13. Mis s ing Charac ters o n G ues t Co ns o le with Jap anes e Keyb o ard
2 1.14. Verifying Virtualiz atio n Extens io ns

423
423
424

. .ppendix
A
. . . . . . . A.
..T
. .he
. . Virt
. . . .ual
. . .Host
. . . . .Met
. . .rics
. . . .Daemon
. . . . . . . (vhost
. . . . . .md)
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4.2. 6. . . . . . . . . .
. .ppendix
A
. . . . . . . B.
. . .Addit
. . . . ional
. . . . . Resources
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4.2. 7. . . . . . . . . .
B .1. O nline Res o urc es
427
B .2. Ins talled Do c umentatio n
427
. .ppendix
A
. . . . . . . C.
. . Revision
. . . . . . . . .Hist
. . . ory
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4.2. 8. . . . . . . . . .

11

Virt ualiz at ion Administ rat ion G uide

Chapter 1. Server Best Practices

The following tasks and tips can assist you with increasing the performance of your Red Hat
Enterprise Linux host. Additional tips can be found in the Red Hat Enterprise Linux Virtualization Tuning
and Optimization Guide
Run SELinux in enforcing mode. Set SELinux to run in enforcing mode with the setenfo rce
command.
# setenforce 1
Remove or disable any unnecessary services such as Auto FS, NFS, FT P , HT T P , NIS, tel netd ,
send mai l and so on.
Only add the minimum number of user accounts needed for platform management on the server
and remove unnecessary user accounts.
Avoid running any unessential applications on your host. Running applications on the host may
impact virtual machine performance and can affect server stability. Any application which may
crash the server will also cause all virtual machines on the server to go down.
Use a central location for virtual machine installations and images. Virtual machine images
should be stored under /var/l i b/l i bvi rt/i mag es/. If you are using a different directory for
your virtual machine images make sure you add the directory to your SELinux policy and relabel
it before starting the installation. Use of shareable, network storage in a central location is highly
recommended.

12

Chapt er 2 . Securit y for Virt ualiz at ion

Chapter 2. Security for Virtualization

When deploying virtualization technologies, you must ensure that the host physical machine and its
operating system cannot be compromised. In this case the host is a Red Hat Enterprise Linux system
that manages the system, devices, memory and networks as well as all guest virtual machines. If the
host physical machine is insecure, all guest virtual machines in the system are vulnerable. There are
several ways to enhance security on systems using virtualization. You or your organization should
create a Deployment Plan. This plan needs to contain the following:
Operating specifications
Specifies which services are needed on your guest virtual machines
Specifies the host physical servers as well as what support is required for these services
Here are a few security issues to consider while developing a deployment plan:
Run only necessary services on host physical machines. The fewer processes and services
running on the host physical machine, the higher the level of security and performance.
Enable SELinux on the hypervisor. Read Section 2.2, SELinux and Virtualization for more
information on using SELinux and virtualization. Additional security tips are located in the Red
Hat Enterprise Linux Virtualization Security Guide.
Use a firewall to restrict traffic to the host physical machine. You can setup a firewall with defaultreject rules that will help secure the host physical machine from attacks. It is also important to limit
network-facing services.
D o not allow normal users to access the host operating system. If the host operating system is
privileged, granting access to unprivileged accounts may compromise the level of security.

2.1. St orage Securit y Issues

Keeping in mind that any user with administrative security privileges for guest virtual machines can
potentially change the partitions in the host physical machine, it is imperative that only actual system
administrators are granted this level of security. In addition, the following should be considered:
The host physical machine should not use disk labels to identify file systems in the fstab file, the
i ni trd file which are accessed by the command line. If less privileged users, especially users of
guest virtual machines have write access to whole partitions or LVM volumes, then they can be
accidentally deleted and this mistake will impact all other guest virtual machines that are using
the same storage.
Users of guest virtual machines should not be given write access to entire disks or block devices
(for example, /d ev/sd b). To avoid this, use partitions such as /d ev/sd b1 or LVM volumes.

2.2. SELinux and Virt ualiz at ion

Security Enhanced Linux was developed by the NSA with assistance from the Linux community to
provide stronger security for Linux. SELinux limits an attacker's abilities and works to prevent many
common security exploits such as buffer overflow attacks and privilege escalation. It is because of
these benefits that all Red Hat Enterprise Linux systems should run with SELinux enabled and in
enforcing mode.
Pro ced u re 2.1. C reat in g an d mo u n t in g a lo g ical vo lu me o n a g u est virt u al mach in e wit h
SELin u x en ab led

13

Virt ualiz at ion Administ rat ion G uide

1. Create a logical volume. This example creates a 5 gigabyte logical volume named
NewVolumeName on the volume group named volumegroup. This example also assumes
that there is enough disk space. You may have to create additional storage on a network
device and give the guest access to it. Refer to Chapter 13, Volumes for more information.
# lvcreate -n NewVolumeName -L 5G volumegroup
2. Format the NewVolumeName logical volume with a file system that supports extended
attributes, such as ext3.
# mke2fs -j /dev/volumegroup/NewVolumeName
3. Create a new directory for mounting the new logical volume. This directory can be anywhere
on your file system. It is advised not to put it in important system directories (/etc, /var,
/sys) or in home directories (/ho me or /ro o t). This example uses a directory called
/vi rtsto rag e
# mkdir /virtstorage
4. Mount the logical volume.
# mount /dev/volumegroup/NewVolumeName /virtstorage
5. Set the SELinux type for the folder you just created.
# semanage fcontext -a -t virt_image_t "/virtstorage(/.*)?"
If the targeted policy is used (targeted is the default policy) the command appends a line to
the /etc/sel i nux/targ eted /co ntexts/fi l es/fi l e_co ntexts. l o cal file which
makes the change persistent. The appended line may resemble this:
/virtstorage(/.*)?

system_u:object_r:virt_image_t:s0

6. Run the command to change the type of the mount point (/vi rtsto rag e) and all files under
it to vi rt_i mag e_t (the resto reco n and setfi l es commands read the files in
/etc/sel i nux/targ eted /co ntexts/fi l es/).
# restorecon -R -v /virtstorage

14

Chapt er 2 . Securit y for Virt ualiz at ion

Note
Create a new file (using the to uch command) on the file system.
# touch /virtstorage/newfile
Verify the file has been relabeled using the following command:
# sudo ls -Z /virtstorage
-rw-------. root root system_u:object_r:virt_image_t:s0 newfile
The output shows that the new file has the correct attribute, vi rt_i mag e_t.

2.3. SELinux
This section contains topics to consider when using SELinux with your virtualization deployment.
When you deploy system changes or add devices, you must update your SELinux policy
accordingly. To configure an LVM volume for a guest virtual machine, you must modify the SELinux
context for the respective underlying block device and volume group. Make sure that you have
installed the po l i cyco reuti l s-pytho n package (yum i nstal l po l i cyco reuti l s-pytho n)
before running the command.
# semanage fcontext -a -t virt_image_t -f -b /dev/sda2
# restorecon /dev/sda2
K VM an d SELin u x
The following table shows the SELinux Booleans which affect KVM when launched by libvirt.
K VM SELin u x B o o lean s
SELin u x B o o lean

D escrip t io n

virt_use_comm

Allow virt to use serial/parallel communication

ports.
Allow virt to read fuse files.
Allow virt to manage NFS files.
Allow virt to manage CIFS files.
Allow sanlock to manage virt lib files.
Allow virt to manage device configuration (PCI).
Allow virtual machine to interact with the xserver.
Allow virt to use USB devices.

virt_use_fusefs
virt_use_nfs
virt_use_samba
virt_use_sanlock
virt_use_sysfs
virt_use_xserver
virt_use_usb

2.4 . Virt ualiz at ion Firewall Informat ion

Various ports are used for communication between guest virtual machines and cooresponding
management utilities.

15

Virt ualiz at ion Administ rat ion G uide

Note
Any network service on a guest virtual machine must have the applicable ports open on the
guest virtual machine to allow external access. If a network service on a guest virtual machine
is firewalled it will be inaccessible. Always verify the guest virtual machine's network
configuration first.
ICMP requests must be accepted. ICMP packets are used for network testing. You cannot ping
guest virtual machines if the ICMP packets are blocked.
Port 22 should be open for SSH access and the initial installation.
Ports 80 or 443 (depending on the security settings on the RHEV Manager) are used by the vdsmreg service to communicate information about the host physical machine.
Ports 5634 to 6166 are used for guest virtual machine console access with the SPICE protocol.
Ports 49152 to 49216 are used for migrations with KVM. Migration may use any port in this range
depending on the number of concurrent migrations occurring.
Enabling IP forwarding (net. i pv4 . i p_fo rward = 1) is also required for shared bridges and
the default bridge. Note that installing libvirt enables this variable so it will be enabled when the
virtualization packages are installed unless it was manually disabled.

Note
Note that enabling IP forwarding is n o t required for physical bridge devices. When a guest
virtual machine is connected through a physical bridge, traffic only operates at a level that
does not require IP configuration such as IP forwarding.

16

Chapt er 3. sVirt

Chapter 3. sVirt
sVirt is a technology included in Red Hat Enterprise Linux 6 that integrates SELinux and
virtualization. sVirt applies Mandatory Access Control (MAC) to improve security when using guest
virtual machines. This integrated technology improves security and hardens the system against bugs
in the hypervisor. It is particulary helpful in preventing attacks on the host physical machine or on
another guest virtual machine.
This chapter describes how sVirt integrates with virtualization technologies in Red Hat Enterprise
Linux 6.
N o n - virt u aliz ed En viro n men t s
In a non-virtualized environment, host physical machines are separated from each other physically
and each host physical machine has a self-contained environment, consisting of services such as a
web server, or a D NS server. These services communicate directly to their own user space, host
physical machine's kernel and physical hardware, offering their services directly to the network. The
following image represents a non-virtualized environment:

User Space - memory area where all user mode applications and some drivers execute.

Web App (web application server) - delivers web content that can be accessed through the a
browser.
Host Kernel - is strictly reserved for running the host physical machine's privileged kernel,
kernel extensions, and most device drivers.

D NS Server - stores D NS records allowing users to access web pages using logical names
instead of IP addresses.
Virt u aliz ed En viro n men t s
In a virtualized environment, several virtual operating systems can run on a single kernel residing on
a host physical machine. The following image represents a virtualized environment:

17

Virt ualiz at ion Administ rat ion G uide

3.1. Securit y and Virt ualiz at ion

When services are not virtualized, machines are physically separated. Any exploit is usually
contained to the affected machine, with the obvious exception of network attacks. When services are
grouped together in a virtualized environment, extra vulnerabilities emerge in the system. If there is a
security flaw in the hypervisor that can be exploited by a guest virtual machine, this guest virtual
machine may be able to not only attack the host physical machine, but also other guest virtual
machines running on that host physical machine. These attacks can extend beyond the guest virtual
machine and could expose other guest virtual machines to an attack as well.
sVirt is an effort to isolate guest virtual machines and limit their ability to launch further attacks if
exploited. This is demonstrated in the following image, where an attack can not break out of the guest
virtual machine and invade other guest virtual machines:

SELinux introduces a pluggable security framework for virtualized instances in its implementation of
Mandatory Access Control (MAC). The sVirt framework allows guest virtual machines and their
resources to be uniquely labeled. Once labeled, rules can be applied which can reject access
between different guest virtual machines.

18

Chapt er 3. sVirt

3.2. sVirt Labeling

Like other services under the protection of SELinux, sVirt uses process-based mechanisms and
restrictions to provide an extra layer of security over guest virtual machines. Under typical use, you
should not even notice that sVirt is working in the background. This section describes the labeling
features of sVirt.
As shown in the following output, when using sVirt, each virtualized guest virtual machine process is
labeled and runs with a dynamically generated level. Each process is isolated from other VMs with
different levels:
# ps -eZ | grep qemu
system_u:system_r:svirt_t:s0:c87,c520 27950 ?

00:00:17 qemu-kvm

The actual disk images are automatically labeled to match the processes, as shown in the following
output:
# ls -lZ /var/lib/libvirt/images/*
system_u:object_r:svirt_image_t:s0:c87,c520

image1

The following table outlines the different context labels that can be assigned when using sVirt:
T ab le 3.1. sVirt co n t ext lab els
SELin u x C o n t ext

T yp e / D escrip t io n

system_u:system_r:svirt_t:MCS1

Guest virtual machine processes. MCS1 is a

random MCS field. Approximately 500,000
labels are supported.
Guest virtual machine images. Only svirt_t
processes with the same MCS fields can
read/write these images.
Guest virtual machine shared read/write content.
All svirt_t processes can write to the
svirt_image_t:s0 files.

system_u:object_r:svirt_image_t:MCS1

system_u:object_r:svirt_image_t:s0

It is also possible to perform static labeling when using sVirt. Static labels allow the administrator to
select a specific label, including the MCS/MLS field, for a guest virtual machine. Administrators who
run statically-labeled virtualized guest virtual machines are responsible for setting the correct label
on the image files. The guest virtual machine will always be started with that label, and the sVirt
system will never modify the label of a statically-labeled virtual machine's content. This allows the
sVirt component to run in an MLS environment. You can also run multiple guest virtual machines with
different sensitivity levels on a system, depending on your requirements.

19

Virt ualiz at ion Administ rat ion G uide

Chapter 4. KVM Live Migration

This chapter covers migrating guest virtual machines running on one host physical machine to
another. In both instances, the host physical machines are running the KVM hypervisor.
Migration describes the process of moving a guest virtual machine from one host physical machine
to another. This is possible because guest virtual machines are running in a virtualized environment
instead of directly on the hardware. Migration is useful for:
Load balancing - guest virtual machines can be moved to host physical machines with lower
usage when their host physical machine becomes overloaded, or another host physical machine
is under-utilized.
Hardware independence - when we need to upgrade, add, or remove hardware devices on the
host physical machine, we can safely relocate guest virtual machines to other host physical
machines. This means that guest virtual machines do not experience any downtime for hardware
improvements.
Energy saving - guest virtual machines can be redistributed to other host physical machines and
can thus be powered off to save energy and cut costs in low usage periods.
Geographic migration - guest virtual machines can be moved to another location for lower latency
or in serious circumstances.
Migration works by sending the state of the guest virtual machine's memory and any virtualized
devices to a destination host physical machine. It is recommended to use shared, networked storage
to store the guest virtual machine's images to be migrated. It is also recommended to use libvirtmanaged storage pools for shared storage when migrating virtual machines.
Migrations can be performed live or not.
In a live migration, the guest virtual machine continues to run on the source host physical machine
while its memory pages are transferred, in order, to the destination host physical machine. D uring
migration, KVM monitors the source for any changes in pages it has already transferred, and begins
to transfer these changes when all of the initial pages have been transferred. KVM also estimates
transfer speed during migration, so when the remaining amount of data to transfer will take a certain
configurable period of time (10ms by default), KVM suspends the original guest virtual machine,
transfers the remaining data, and resumes the same guest virtual machine on the destination host
physical machine.
A migration that is not performed live, suspends the guest virtual machine, then moves an image of
the guest virtual machine's memory to the destination host physical machine. The guest virtual
machine is then resumed on the destination host physical machine and the memory the guest virtual
machine used on the source host physical machine is freed. The time it takes to complete such a
migration depends on network bandwidth and latency. If the network is experiencing heavy use or
low bandwidth, the migration will take much longer.
If the original guest virtual machine modifies pages faster than KVM can transfer them to the
destination host physical machine, offline migration must be used, as live migration would never
complete.

4 .1. Live Migrat ion Requirement s

Migrating guest virtual machines requires the following:

20

Chapt er 4 . KVM Live Migrat ion

Mig rat io n req u iremen t s

A guest virtual machine installed on shared storage using one of the following protocols:
Fibre Channel-based LUNs
iSCSI
FCoE
NFS
GFS2
SCSI RD MA protocols (SCSI RCP): the block export protocol used in Infiniband and 10GbE
iWARP adapters
The migration platforms and versions should be checked against table Table 4.1, Live Migration
Compatibility . It should also be noted that Red Hat Enterprise Linux 6 supports live migration of
guest virtual machines using raw and qcow2 images on shared storage.
Both systems must have the appropriate TCP/IP ports open. In cases where a firewall is used refer
to Section 2.4, Virtualization Firewall Information for detailed port information.
A separate system exporting the shared storage medium. Storage should not reside on either of
the two host physical machines being used for migration.
Shared storage must mount at the same location on source and destination systems. The
mounted directory names must be identical. Although it is possible to keep the images using
different paths, it is not recommended. Note that, if you are intending to use virt-manager to
perform the migration, the path names must be identical. If however you intend to use virsh to
perform the migration, different network configurations and mount directories can be used with the
help of --xml option or pre-hooks when doing migrations. Even without shared storage,
migration can still succeed with the option --co py-sto rag e-al l (deprecated). For more
information on preho o ks, refer to libvirt.org, and for more information on the XML option, refer to
Chapter 20, Manipulating the Domain XML.
When migration is attempted on an existing guest virtual machine in a public bridge+tap network,
the source and destination host physical machines must be located in the same network.
Otherwise, the guest virtual machine network will not operate after migration.
In Red Hat Enterprise Linux 5 and 6, the default cache mode of KVM guest virtual machines is set
to none, which prevents inconsistent disk states. Setting the cache option to none (using vi rsh
attach-d i sk cache no ne, for example), causes all of the guest virtual machine's files to be
opened using the O_DIRECT flag (when calling the o pen syscall), thus bypassing the host
physical machine's cache, and only providing caching on the guest virtual machine. Setting the
cache mode to none prevents any potential inconsistency problems, and when used makes it
possible to live-migrate virtual machines. For information on setting cache to none, refer to
Section 13.3, Adding Storage D evices to Guests .
Make sure that the l i bvi rtd service is enabled (# chkco nfi g l i bvi rtd o n) and running (#
servi ce l i bvi rtd start). It is also important to note that the ability to migrate effectively is
dependent on the parameter settings in the /etc/l i bvi rt/l i bvi rtd . co nf configuration file.
Pro ced u re 4 .1. C o n f ig u rin g lib virt d .co n f
1. Opening the l i bvi rtd . co nf requires running the command as root:
# vim /etc/libvirt/libvirtd.conf

21

Virt ualiz at ion Administ rat ion G uide

2. Change the parameters as needed and save the file.

3. Restart the l i bvi rtd service:
# service libvirtd restart

4 .2. Live Migrat ion and Red Hat Ent erprise Linux Version Compat ibilit y
Live Migration is supported as shown in table Table 4.1, Live Migration Compatibility :
T ab le 4 .1. Live Mig rat io n C o mp at ib ilit y
Mig rat io n
Met h o d

R elease T yp e

Examp le

Live Mig rat io n

Su p p o rt

Forward
Forward

Major release
Minor release

5.x 6.y
5.x 5.y (y>x,
x>=4)

Not supported
Fully supported

Forward

Minor release

6.x 6.y (y>x,

x>=0)

Fully supported

Backward
Backward

Major release
Minor release

6.x 5.y
5.x 5.y
(x>y,y>=4)

Not supported
Supported

Backward

Minor release

6.x 6.y (x>y,

y>=0)

Supported

N o t es

Any issues
should be
reported
Any issues
should be
reported
Refer to
Troubleshooting
problems with
migration for
known issues
Refer to
Troubleshooting
problems with
migration for
known issues

T ro u b lesh o o t in g p ro b lems wit h mig rat io n

Issu es wit h SPIC E It has been found that SPICE has an incompatible change when
migrating from Red Hat Enterprise Linux 6.0 6.1. In such cases, the client may disconnect and
then reconnect, causing a temporary loss of audio and video. This is only temporary and all
services will resume.
Issu es wit h U SB Red Hat Enterprise Linux 6.2 added USB functionality which included
migration support, but not without certain caveats which reset USB devices and caused any
application running over the device to abort. This problem was fixed in Red Hat Enterprise Linux
6.4, and should not occur in future versions. To prevent this from happening in a version prior to
6.4, abstain from migrating while USB devices are in use.
Issu es wit h t h e mig rat io n p ro t o co l If backward migration ends with " unknown section
error" , repeating the migration process can repair the issue as it may be a transient error. If not,
please report the problem.

C o n f ig u rin g N et wo rk St o rag e

22

Chapt er 4 . KVM Live Migrat ion

Configure shared storage and install a guest virtual machine on the shared storage.
Alternatively, use the NFS example in Section 4.3, Shared Storage Example: NFS for a Simple
Migration

4 .3. Shared St orage Example: NFS for a Simple Migrat ion

Important
This example uses NFS to share guest virtual machine images with other KVM host physical
machines. Although not practical for large installations, it is presented to demonstrate
migration techniques only. D o not use this example for migrating or running more than a few
guest virtual machines. In addition, it is required that the sync parameter is enabled. This is
required for proper export of the NFS storage. In addition, it is strongly recommended that the
the NFS is mounted on source host physical machine, and the guest virtual machine's image
needs to be created on the NFS mounted directory located on source host physical machine. It
should also be noted that NFS filelocking must n o t be used as it is not supported in KVM.
iSCSI storage is a better choice for large deployments. Refer to Section 12.5, iSCSI-based
Storage Pools for configuration details.

Also note, that the instructions provided in this section are not meant to replace the detailed
instructions found in Red Hat Linux Storage Administration Guide. Refer to this guide for information on
configuring NFS, opening IP tables, and configuring the firewall.
1. C reat e a d irect o ry f o r t h e d isk imag es
This shared directory will contain the disk images for the guest virtual machines. To do this
create a directory in a location different from /var/l i b/l i bvi rt/i mag es. For example:
# mkdir /var/lib/libvirt-img/images
2. Ad d t h e n ew d irect o ry p at h t o t h e N FS co n f ig u rat io n f ile
The NFS configuration file is a text file located in /etc/expo rts. Open the file and edit it
adding the path to the new file you created in step 1.
# echo "/var/lib/libvirt-img/images" >> /etc/exports/[NFS-ConfigFILENAME.txt]
3. St art N FS
a. Make sure that the ports for NFS in i ptabl es (2049, for example) are opened and
add NFS to the /etc/ho sts. al l o w file.
b. Start the NFS service:
# service nfs start
4. Mo u n t t h e sh ared st o rag e o n b o t h t h e so u rce an d t h e d est in at io n

23

Virt ualiz at ion Administ rat ion G uide

Mount the /var/l i b/l i bvi rt/i mag es directory on both the source and destination
system, running the following command twice. Once on the source system and again on the
destination system.
# mount source_host:/var/lib/libvirt-img/images
/var/lib/libvirt/images

Warning
Make sure that the direcrtories you create in this procedure is compliant with the
requirements as outlined in Section 4.1, Live Migration Requirements . In addition, the
directory may need to be labeled with the correct SELinux label. For more information
consult the NFS chapter in the Red Hat Enterprise Linux Storage Administration Guide.

4 .4 . Live KVM Migrat ion wit h virsh

A guest virtual machine can be migrated to another host physical machine with the vi rsh command.
The mi g rate command accepts parameters in the following format:
# virsh migrate --live GuestName DestinationURL
Note that the --l i ve option may be eliminated when live migration is not desired. Additional options
are listed in Section 4.4.2, Additional Options for the virsh migrate Command .
The GuestName parameter represents the name of the guest virtual machine which you want to
migrate.
The DestinationURL parameter is the connection URL of the destination host physical machine.
The destination system must run the same version of Red Hat Enterprise Linux, be using the same
hypervisor and have l i bvi rt running.

Note
The DestinationURL parameter for normal migration and peer2peer migration has different
semantics:
normal migration: the DestinationURL is the URL of the target host physical machine as
seen from the source guest virtual machine.
peer2peer migration: DestinationURL is the URL of the target host physical machine as
seen from the source host physical machine.

Once the command is entered, you will be prompted for the root password of the destination system.

24

Chapt er 4 . KVM Live Migrat ion

Important
An entry for the destination host physical machine, in the /etc/ho sts file on the source
server is required for migration to succeed. Enter the IP address and hostname for the
destination host physical machine in this file as shown in the following example, substituting
your destination host physical machine's IP address and hostname:
10.0.0.20 host2.example.com

Examp le: Live Mig rat io n wit h virsh

This example migrates from ho st1. exampl e. co m to ho st2. exampl e. co m. Change the host
physical machine names for your environment. This example migrates a virtual machine named
g uest1-rhel 6 -6 4 .
This example assumes you have fully configured shared storage and meet all the prerequisites
(listed here: Migration requirements).
1.
Verif y t h e g u est virt u al mach in e is ru n n in g
From the source system, ho st1. exampl e. co m, verify g uest1-rhel 6 -6 4 is running:
[root@ host1 ~]# virsh list
Id Name
State
---------------------------------10 guest1-rhel6-64
running
2.
Mig rat e t h e g u est virt u al mach in e
Execute the following command to live migrate the guest virtual machine to the destination,
ho st2. exampl e. co m. Append /system to the end of the destination URL to tell libvirt that
you need full access.
# virsh migrate --live guest1-rhel6-64
qemu+ssh://host2.example.com/system
Once the command is entered you will be prompted for the root password of the destination
system.
3.
Wait
The migration may take some time depending on load and the size of the guest virtual
machine. vi rsh only reports errors. The guest virtual machine continues to run on the
source host physical machine until fully migrated.

25

Virt ualiz at ion Administ rat ion G uide

Note
D uring the migration, the completion percentage indicator number is likely to decrease
multiple times before the process finishes. This is caused by a recalculation of the
overall progress, as source memory pages that are changed after the migration starts
need to be be copied again. Therefore, this behavior is expected and does not indicate
any problems with the migration.
4.
Verif y t h e g u est virt u al mach in e h as arrived at t h e d est in at io n h o st
From the destination system, ho st2. exampl e. co m, verify g uest1-rhel 6 -6 4 is running:
[root@ host2 ~]# virsh list
Id Name
State
---------------------------------10 guest1-rhel6-64
running
The live migration is now complete.

Note
libvirt supports a variety of networking methods including TLS/SSL, UNIX sockets, SSH, and
unencrypted TCP. Refer to Chapter 5, Remote Management of Guests for more information on
using other methods.

Note
Non-running guest virtual machines cannot be migrated with the vi rsh mi g rate command.
To migrate a non-running guest virtual machine, the following script should be used:
virsh dumpxml Guest1 > Guest1.xml
virsh -c qemu+ssh://<target-system-FQDN>
virsh undefine Guest1

define Guest1.xml

4 .4 .1. Addit ional T ips for Migrat ion wit h virsh

It is possible to perform multiple, concurrent live migrations where each migration runs in a separate
command shell. However, this should be done with caution and should involve careful calculations
as each migration instance uses one MAX_CLIENT from each side (source and target). As the default
setting is 20, there is enough to run 10 instances without changing the settings. Should you need to
change the settings, refer to the procedure Procedure 4.1, Configuring libvirtd.conf .
1. Open the libvirtd.conf file as described in Procedure 4.1, Configuring libvirtd.conf .
2. Look for the Processing controls section.

26

Chapt er 4 . KVM Live Migrat ion

#################################################################
#
# Processing controls
#
# The maximum number of concurrent client connections to allow
# over all sockets combined.
#max_clients = 20

# The minimum limit sets the number of workers to start up

# initially. If the number of active clients exceeds this,
# then more threads are spawned, upto max_workers limit.
# Typically you'd want max_workers to equal maximum number
# of clients allowed
#min_workers = 5
#max_workers = 20

# The number of priority workers. If all workers from above

# pool will stuck, some calls marked as high priority
# (notably domainDestroy) can be executed in this pool.
#prio_workers = 5
# Total global limit on concurrent RPC calls. Should be
# at least as large as max_workers. Beyond this, RPC requests
# will be read into memory and queued. This directly impact
# memory usage, currently each request requires 256 KB of
# memory. So by default upto 5 MB of memory is used
#
# XXX this isn't actually enforced yet, only the per-client
# limit is used so far
#max_requests = 20
# Limit on concurrent requests from a single client
# connection. To avoid one client monopolizing the server
# this should be a small fraction of the global max_requests
# and max_workers parameter
#max_client_requests = 5
#################################################################
3. Change the max_clients and max_workers parameters settings. It is recommended that
the number be the same in both parameters. The max_clients will use 2 clients per
migration (one per side) and max_workers will use 1 worker on the source and 0 workers on
the destination during the perform phase and 1 worker on the destination during the finish
phase.

27

Virt ualiz at ion Administ rat ion G uide

Important
The max_clients and max_workers parameters settings are effected by all guest
virtual machine connections to the libvirtd service. This means that any user that is
using the same guest virtual machine and is performing a migration at the same time
will also beholden to the limits set in the the max_clients and max_workers
parameters settings. This is why the maximum value needs to be considered carefully
before performing a concurrent live migration.
4. Save the file and restart the service.

Note
There may be cases where a migration connection drops because there are too many
ssh sessions that have been started, but not yet authenticated. By default, sshd allows
only 10 sessions to be in a " pre-authenticated state" at any time. This setting is
controlled by the MaxStartups parameter in the sshd configuration file (located here:
/etc/ssh/sshd _co nfi g ), which may require some adjustment. Adjusting this
parameter should be done with caution as the limitation is put in place to prevent D oS
attacks (and over-use of resources in general). Setting this value too high will negate
its purpose. To change this parameter, edit the file /etc/ssh/sshd _co nfi g , remove
the # from the beginning of the MaxStartups line, and change the 10 (default value)
to a higher number. Remember to save the file and restart the sshd service. For more
information, refer to the sshd _co nfi g man page.

4 .4 .2. Addit ional Opt ions for t he virsh migrat e Command

In addition to --l i ve, virsh migrate accepts the following options:
--d i rect - used for direct migration
--p2p - used for peer-2-peer migration
--tunnel l ed - used for tunneled migration
--persi stent - leaves the domain in a persistent state on the destination host physical machine
--und efi neso urce - removes the guest virtual machine on the source host physical machine
--suspend - leaves the domain in a paused state on the destination host physical machine
--chang e-pro tecti o n - enforces that no incompatible configuration changes will be made to
the domain while the migration is underway; this option is implicitly enabled when supported by
the hypervisor, but can be explicitly used to reject the migration if the hypervisor lacks change
protection support.
--unsafe - forces the migration to occur, ignoring all safety procedures.
--verbo se - displays the progress of migration as it is occurring
--abo rt-o n-erro r - cancels the migration if a soft error (such as an I/O error) happens during
the migration process.
--mi g rateuri - the migration URI which is usually omitted.

28

Chapt er 4 . KVM Live Migrat ion

--d o mai n [string]- domain name, id or uuid

--d esturi [string]- connection URI of the destination host physical machine as seen from the
client(normal migration) or source(p2p migration)
--mi g rateuri - migration URI, usually can be omitted
--ti meo ut [seconds]- forces a guest virtual machine to suspend when the live migration counter
exceeds N seconds. It can only be used with a live migration. Once the timeout is initiated, the
migration continues on the suspended guest virtual machine.
--d name [string] - changes the name of the guest virtual machine to a new name during migration
(if supported)
--xml - the filename indicated can be used to supply an alternative XML file for use on the
destination to supply a larger set of changes to any host-specific portions of the domain XML,
such as accounting for naming differences between source and destination in accessing
underlying storage. This option is usually omitted.
Refer to the virsh man page for more information.

4 .5. Migrat ing wit h virt -manager

This section covers migrating a KVM guest virtual machine with vi rt-manag er from one host
physical machine to another.
1. O p en virt - man ag er
Open vi rt-manag er. Choose Ap p licat io n s Syst em T o o ls Virt u al Mach in e
Man ag er from the main menu bar to launch vi rt-manag er.

Fig u re 4 .1. Virt - Man ag er main men u

2. C o n n ect t o t h e t arg et h o st p h ysical mach in e

29

Virt ualiz at ion Administ rat ion G uide

Connect to the target host physical machine by clicking on the File menu, then click Ad d
C o n n ect io n .

Fig u re 4 .2. O p en Ad d C o n n ect io n win d o w

3. Ad d co n n ect io n
The Ad d C o nnecti o n window appears.

Fig u re 4 .3. Ad d in g a co n n ect io n t o t h e t arg et h o st p h ysical mach in e

30

Chapt er 4 . KVM Live Migrat ion

Enter the following details:

Hypervi so r: Select Q EMU /K VM.
Metho d : Select the connection method.
Username: Enter the username for the remote host physical machine.
Ho stname: Enter the hostname for the remote host physical machine.
Click the C o nnect button. An SSH connection is used in this example, so the specified user's
password must be entered in the next step.

Fig u re 4 .4 . En t er p asswo rd
4. Mig rat e g u est virt u al mach in es
Open the list of guests inside the source host physical machine (click the small triangle on
the left of the host name) and right click on the guest that is to be migrated (g u est 1- rh el6 6 4 in this example) and click Mig rat e.

31

Virt ualiz at ion Administ rat ion G uide

Fig u re 4 .5. C h o o sin g t h e g u est t o b e mig rat ed

In the N ew H o st field, use the drop-down list to select the host physical machine you wish to
migrate the guest virtual machine to and click Mig rat e.

32

Chapt er 4 . KVM Live Migrat ion

Fig u re 4 .6 . C h o o sin g t h e d est in at io n h o st p h ysical mach in e an d st art in g t h e

mig rat io n p ro cess
A progress window will appear.

Fig u re 4 .7. Pro g ress win d o w

33

Virt ualiz at ion Administ rat ion G uide

virt - man ag er now displays the newly migrated guest virtual machine running in the
destination host. The guest virtual machine that was running in the source host physical
machine is now listed inthe Shutoff state.

Fig u re 4 .8. Mig rat ed g u est virt u al mach in e ru n n in g in t h e d est in at io n h o st

p h ysical mach in e
5. O p t io n al - View t h e st o rag e d et ails f o r t h e h o st p h ysical mach in e
In the Ed it menu, click C o n n ect io n D et ails, the Connection D etails window appears.
Click the Sto rag e tab. The iSCSI target details for the destination host physical machine is
shown. Note that the migrated guest virtual machine is listed as using the storage

34

Chapt er 4 . KVM Live Migrat ion

Fig u re 4 .9 . St o rag e d et ails

This host was defined by the following XML configuration:

<pool type='iscsi'>
<name>iscsirhel6guest</name>
<source>
<host name='virtlab22.example.com.'/>
<device path='iqn.2001-05.com.iscsivendor:0-8a0906fbab74a06-a700000017a4cc89-rhevh'/>
</source>
<target>
<path>/dev/disk/by-path</path>
</target>
</pool>
...
Fig u re 4 .10. XML co n f ig u rat io n f o r t h e d est in at io n h o st p h ysical mach in e

35

Virt ualiz at ion Administ rat ion G uide

Chapter 5. Remote Management of Guests

This section explains how to remotely manage your guests using ssh or TLS and SSL. More
information on SSH can be found in the Red Hat Enterprise Linux Deployment Guide.

5.1. Remot e Management wit h SSH

The ssh package provides an encrypted network protocol which can securely send management
functions to remote virtualization servers. The method described uses the l i bvi rt management
connection securely tunneled over an SSH connection to manage the remote machines. All the
authentication is done using SSH public key cryptography and passwords or passphrases gathered
by your local SSH agent. In addition the VN C console for each guest is tunneled over SSH .
Be aware of the issues with using SSH for remotely managing your virtual machines, including:
you require root log in access to the remote machine for managing virtual machines,
the initial connection setup process may be slow,
there is no standard or trivial way to revoke a user's key on all hosts or guests, and
ssh does not scale well with larger numbers of remote machines.

Note
Red Hat Enterprise Virtualization enables remote management of large numbers of virtual
machines. Refer to the Red Hat Enterprise Virtualization documentation for further details.
The following packages are required for ssh access:
openssh
openssh-askpass
openssh-clients
openssh-server

C o n f ig u rin g Passwo rd - less o r Passwo rd - man ag ed SSH Access f o r vi rt-manag er

The following instructions assume you are starting from scratch and do not already have SSH keys
set up. If you have SSH keys set up and copied to the other systems you can skip this procedure.

36

Chapt er 5. Remot e Management of G uest s

Important
SSH keys are user dependent and may only be used by their owners. A key's owner is the one
who generated it. Keys may not be shared.
vi rt-manag er must be run by the user who owns the keys to connect to the remote host. That
means, if the remote systems are managed by a non-root user vi rt-manag er must be run in
unprivileged mode. If the remote systems are managed by the local root user then the SSH
keys must be owned and created by root.
You cannot manage the local host as an unprivileged user with vi rt-manag er.

1. O p t io n al: C h an g in g u ser
Change user, if required. This example uses the local root user for remotely managing the
other hosts and the local host.
$ su 2. G en erat in g t h e SSH key p air
Generate a public key pair on the machine vi rt-manag er is used. This example uses the
default key location, in the ~ /. ssh/ directory.
# ssh-keyg en -t rsa
3. C o p yin g t h e keys t o t h e remo t e h o st s
Remote login without a password, or with a passphrase, requires an SSH key to be
distributed to the systems being managed. Use the ssh-co py-i d command to copy the key
to root user at the system address provided (in the example, [email protected]).
# ssh-co py-i d -i ~ /. ssh/i d _rsa. pub ro o t@ ho st2. exampl e. co m
root@ host2.example.com's password:
Now try logging into the machine, with the ssh ro o t@ ho st2. exampl e. co m command and
check in the . ssh/autho ri zed _keys file to make sure unexpected keys have not been
added.
Repeat for other systems, as required.
4. O p t io n al: Ad d t h e p assp h rase t o t h e ssh - ag en t
The instructions below describe how to add a passphrase to an existing ssh-agent. It will fail
to run if the ssh-agent is not running. To avoid errors or conflicts make sure that your SSH
parameters are set correctly. Refer to the Red Hat Enterprise Linux Deployment Guide for more
information.
Add the passphrase for the SSH key to the ssh-ag ent, if required. On the local host, use the
following command to add the passphrase (if there was one) to enable password-less login.
# ssh-ad d ~ /. ssh/i d _rsa

37

Virt ualiz at ion Administ rat ion G uide

The SSH key is added to the remote system.

T h e l i bvi rt D aemo n ( l i bvi rtd )

The l i bvi rt daemon provides an interface for managing virtual machines. You must have the
l i bvi rtd daemon installed and running on every remote host that needs managing.
$ ssh ro o t@ somehost
# chkco nfi g l i bvi rtd o n
# servi ce l i bvi rtd start
After l i bvi rtd and SSH are configured you should be able to remotely access and manage your
virtual machines. You should also be able to access your guests with VNC at this point.

Accessin g R emo t e H o st s wit h virt - man ag er

Remote hosts can be managed with the virt-manager GUI tool. SSH keys must belong to the user
executing virt-manager for password-less login to work.
1. Start virt-manager.
2. Open the File->Ad d C o n n ect io n menu.

Fig u re 5.1. Ad d co n n ect io n men u

3. Use the drop down menu to select hypervisor type, and click the C o n n ect t o remo t e h o st
check box to open the Connection Met h o d (in this case Remote tunnel over SSH), and enter
the desired U ser n ame and H o st n ame, then click C o n n ect .

38

Chapt er 5. Remot e Management of G uest s

5.2. Remot e Management Over T LS and SSL

You can manage virtual machines using TLS and SSL. TLS and SSL provides greater scalability but
is more complicated than ssh (refer to Section 5.1, Remote Management with SSH ). TLS and SSL is
the same technology used by web browsers for secure connections. The l i bvi rt management
connection opens a TCP port for incoming connections, which is securely encrypted and
authenticated based on x509 certificates. The procedures that follow provide instructions on creating
and deploying authentication certificates for TLS and SSL management.
Pro ced u re 5.1. C reat in g a cert if icat e au t h o rit y ( C A) key f o r T LS man ag emen t
1. Before you begin, confirm that the certto o l utility is installed. If not:
# yum i nstal l g nutl s-uti l s
2. Generate a private key, using the following command:
# certto o l --g enerate-pri vkey > cakey. pem
3. Once the key generates, the next step is to create a signature file so the key can be selfsigned. To do this, create a file with signature details and name it ca. i nfo . This file should
contain the following:
# vi m ca. i nfo
cn = Name of your organization
ca
cert_signing_key
4. Generate the self-signed key with the following command:
# certto o l --g enerate-sel f-si g ned --l o ad -pri vkey cakey. pem -templ ate ca. i nfo --o utfi l e cacert. pem
Once the file generates, the ca.info file may be deleted using the rm command. The file that
results from the generation process is named cacert. pem. This file is the public key
(certificate). The loaded file cakey. pem is the private key. This file should not be kept in a
shared space. Keep this key private.
5. Install the cacert. pem Certificate Authority Certificate file on all clients and servers in the
/etc/pki /C A/cacert. pem directory to let them know that the certificate issued by your CA
can be trusted. To view the contents of this file, run:
# certto o l -i --i nfi l e cacert. pem
This is all that is required to set up your CA. Keep the CA's private key safe as you will need it
in order to issue certificates for your clients and servers.
Pro ced u re 5.2. Issu in g a server cert if icat e

39

Virt ualiz at ion Administ rat ion G uide

This procedure demonstrates how to issue a certificate with the X.509 CommonName (CN)field set to
the hostname of the server. The CN must match the hostname which clients will be using to connect to
the server. In this example, clients will be connecting to the server using the URI:
q emu: //myco mmo nname/system, so the CN field should be identical, ie mycommoname.
1. Create a private key for the server.
# certto o l --g enerate-pri vkey > serverkey. pem
2. Generate a signature for the CA's private key by first creating a template file called
server. i nfo . Make sure that the CN is set to be the same as the server's hostname:
organization = Name of your organization
cn = mycommonname
tls_www_server
encryption_key
signing_key
3. Create the certificate with the following command:
# certto o l --g enerate-certi fi cate --l o ad -pri vkey serverkey. pem -l o ad -ca-certi fi cate cacert. pem --l o ad -ca-pri vkey cakey. pem \ -templ ate server. i nfo --o utfi l e servercert. pem
4. This results in two files being generated:
serverkey.pem - The server's private key
servercert.pem - The server's public key
Make sure to keep the location of the private key secret. To view the contents of the file,
perform the following command:
# certto o l -i --i ni fi l e servercert. pem
When opening this file the C N= parameter should be the same as the CN that you set earlier.
For example, myco mmo nname.
5. Install the two files in the following locations:
serverkey. pem - the server's private key. Place this file in the following location:
/etc/pki /l i bvi rt/pri vate/serverkey. pem
servercert. pem - the server's certificate. Install it in the following location on the server:
/etc/pki /l i bvi rt/servercert. pem
Pro ced u re 5.3. Issu in g a clien t cert if icat e
1. For every client (ie. any program linked with libvirt, such as virt-manager), you need to issue
a certificate with the X.509 D istinguished Name (D N) set to a suitable name. This needs to be
decided on a corporate level.
For example purposes the following information will be used:
C=USA,ST=North Carolina,L=Raleigh,O=Red Hat,CN=name_of_client

40

Chapt er 5. Remot e Management of G uest s

This process is quite similar to Procedure 5.2, Issuing a server certificate , with the following
exceptions noted.
2. Make a private key with the following command:
# certto o l --g enerate-pri vkey > cl i entkey. pem
3. Generate a signature for the CA's private key by first creating a template file called
cl i ent. i nfo . The file should contain the following (fields should be customized to reflect
your region/location):
country = USA
state = North Carolina
locality = Raleigh
organization = Red Hat
cn = client1
tls_www_client
encryption_key
signing_key
4. Sign the certificate with the following command:
# certto o l --g enerate-certi fi cate --l o ad -pri vkey cl i entkey. pem -l o ad -ca-certi fi cate cacert. pem \ --l o ad -ca-pri vkey cakey. pem -templ ate cl i ent. i nfo --o utfi l e cl i entcert. pem
5. Install the certificates on the client machine:
# cp cl i entkey. pem /etc/pki /l i bvi rt/pri vate/cl i entkey. pem
# cp cl i entcert. pem /etc/pki /l i bvi rt/cl i entcert. pem

5.3. T ransport Modes

For remote management, l i bvi rt supports the following transport modes:

T ran sp o rt Layer Secu rit y ( T LS)

Transport Layer Security TLS 1.0 (SSL 3.1) authenticated and encrypted TCP/IP socket, usually
listening on a public port number. To use this you will need to generate client and server certificates.
The standard port is 16514.

U N IX So cket s
UNIX domain sockets are only accessible on the local machine. Sockets are not encrypted, and use
UNIX permissions or SELinux for authentication. The standard socket names are
/var/run/l i bvi rt/l i bvi rt-so ck and /var/run/l i bvi rt/l i bvi rt-so ck-ro (for read-only
connections).

41

Virt ualiz at ion Administ rat ion G uide

SSH
Transported over a Secure Shell protocol (SSH) connection. Requires Netcat (the nc package)
installed. The libvirt daemon (l i bvi rtd ) must be running on the remote machine. Port 22 must be
open for SSH access. You should use some sort of SSH key management (for example, the sshag ent utility) or you will be prompted for a password.

ext
The ext parameter is used for any external program which can make a connection to the remote
machine by means outside the scope of libvirt. This parameter is unsupported.

TCP
Unencrypted TCP/IP socket. Not recommended for production use, this is normally disabled, but an
administrator can enable it for testing or use over a trusted network. The default port is 16509.
The default transport, if no other is specified, is TLS.

R emo t e U R Is
A Uniform Resource Identifier (URI) is used by vi rsh and libvirt to connect to a remote host. URIs can
also be used with the --co nnect parameter for the vi rsh command to execute single commands or
migrations on remote hosts. Remote URIs are formed by taking ordinary local URIs and adding a
hostname and/or transport name. As a special case, using a URI scheme of 'remote', will tell the
remote libvirtd server to probe for the optimal hypervisor driver. This is equivalent to passing a NULL
URI for a local connection
libvirt URIs take the general form (content in square brackets, " []" , represents optional functions):
driver[+transport]://[username@ ][hostname][:port]/path[?extraparameters]
Note that if the hypervisor(driver) is QEMU, the path is mandatory. If it is XEN, it is optional.
The following are examples of valid remote URIs:
qemu://hostname/
xen://hostname/
xen+ssh://hostname/
The transport method or the hostname must be provided to target an external location. For more
information refer to http://libvirt.org/guide/html/Application_D evelopment_Guide-ArchitectureRemote_URIs.html.

Examp les o f remo t e man ag emen t p aramet ers

42

Chapt er 5. Remot e Management of G uest s

Connect to a remote KVM host named ho st2, using SSH transport and the SSH username
vi rtuser.The connect command for each is co nnect [<name>] [--read o nl y], where
<name> is a valid URI as explained here. For more information about the vi rsh co nnect
command refer to Section 14.1.5, connect
q emu+ ssh: //vi rtuser@ ho t2/
Connect to a remote KVM hypervisor on the host named ho st2 using TLS.
q emu: //ho st2/

T est in g examp les

Connect to the local KVM hypervisor with a non-standard UNIX socket. The full path to the UNIX
socket is supplied explicitly in this case.
q emu+ uni x: ///system?so cket= /o pt/l i bvi rt/run/l i bvi rt/l i bvi rt-so ck
Connect to the libvirt daemon with an unencrypted TCP/IP connection to the server with the IP
address 10.1.1.10 on port 5000. This uses the test driver with default settings.
test+ tcp: //10 . 1. 1. 10 : 50 0 0 /d efaul t

Ext ra U R I Paramet ers

Extra parameters can be appended to remote URIs. The table below Table 5.1, Extra URI
parameters covers the recognized parameters. All other parameters are ignored. Note that parameter
values must be URI-escaped (that is, a question mark (?) is appended before the parameter and
special characters are converted into the URI format).
T ab le 5.1. Ext ra U R I p aramet ers
N ame

T ran sp o rt mo d e

D escrip t io n

Examp le u sag e

name

all modes

The name passed to

the remote
virConnectOpen
function. The name is
normally formed by
removing transport,
hostname, port
number, username and
extra parameters from
the remote URI, but in
certain very complex
cases it may be better
to supply the name
explicitly.

name=qemu:///system

43

Virt ualiz at ion Administ rat ion G uide

N ame

T ran sp o rt mo d e

D escrip t io n

Examp le u sag e

command

ssh and ext

command=/opt/openss
h/bin/ssh

socket

unix and ssh

netcat

ssh

The external command.

For ext transport this is
required. For ssh the
default is ssh. The
PATH is searched for
the command.
The path to the UNIX
domain socket, which
overrides the default.
For ssh transport, this
is passed to the remote
netcat command (see
netcat).
The netcat command
can be used to connect
to remote systems. The
default netcat
parameter uses the nc
command. For SSH
transport, libvirt
constructs an SSH
command using the
form below:

socket=/opt/libvirt/run/li
bvirt/libvirt-sock

netcat=/opt/netcat/bin/n
c

command -p port [-l

username] hostname
netcat -U socket
The port, username
and hostname
parameters can be
specified as part of the
remote URI. The
command, netcat and
socket come from
other extra parameters.
no_verify

44

tls

If set to a non-zero
value, this disables
client checks of the
server's certificate.
Note that to disable
server checks of the
client's certificate or IP
address you must
change the libvirtd
configuration.

no_verify=1

Chapt er 5. Remot e Management of G uest s

N ame

T ran sp o rt mo d e

D escrip t io n

Examp le u sag e

no_tty

ssh

If set to a non-zero
value, this stops ssh
from asking for a
password if it cannot
log in to the remote
machine automatically
. Use this when you do
not have access to a
terminal .

no_tty=1

45

Virt ualiz at ion Administ rat ion G uide

Chapter 6. Overcommitting with KVM

6.1. Int roduct ion
The KVM hypervisor supports overcommitting CPUs and overcommitting memory. Overcommitting is
allocating more virtualized CPUs or memory than there are physical resources on the system. With
CPU overcommit, under-utilized virtualized servers or desktops can run on fewer servers which saves
a number of system resources, with the net effect of less power, cooling, and investment in server
hardware.
As most processes do not access 100% of their allocated memory all the time, KVM can use this
behavior to its advantage and allocate more memory for guest virtual machines than the host
physical machine actually has available, in a process called overcommitting of resources.

Important
Overcommitting is not an ideal solution for all memory issues as the recommended method to
deal with memory shortage is to allocate less memory per guest so that the sum of all guests
memory (+4G for the host O/S) is lower than the host physical machine's physical memory. If
the guest virtual machines need more memory, then increase the guest virtual machines' swap
space allocation. If however, should you decide to overcommit, do so with caution.
Guest virtual machines running on a KVM hypervisor do not have dedicated blocks of physical RAM
assigned to them. Instead, each guest virtual machine functions as a Linux process where the host
physical machine's Linux kernel allocates memory only when requested. In addition the host
physical machine's memory manager can move the guest virtual machine's memory between its own
physical memory and swap space. This is why overcommitting requires allotting sufficient swap
space on the host physical machine to accommodate all guest virtual machines as well as enough
memory for the host physical machine's processes. As a basic rule, the host physical machine's
operating system requires a maximum of 4GB of memory along with a minimum of 4GB of swap
space. Refer to Example 6.1, Memory overcommit example for more information.
Red Hat Knowledgebase has an article on safely and efficiently determining the size of the swap
partition.

Note
The example below is provided as a guide for configuring swap only. The settings listed may
not be appropriate for your environment.

Examp le 6 .1. Memo ry o verco mmit examp le

This example demonstrates how to calculate swap space for overcommitting. Although it may
appear to be simple in nature, the ramifications of overcommitting should not be ignored. Refer to
Important before proceeding.
ExampleServer1 has 32GB of physical RAM. The system is being configured to run 50 guest
virtual machines, each requiring 1GB of virtualized memory. As mentioned above, the host
physical machine's system itself needs a maximum of 4GB (apart from the guest virtual machines)
as well as an additional 4GB as a swap space minimum.

46

Chapt er 6 . O vercommit t ing wit h KVM

The swap space is calculated as follows:

Calculate the amount of memory needed for the sum of all the guest virtual machines - In this
example: (50 guest virtual machines * 1GB of memory per guest virtual machine) = 50GB
Add the guest virtual machine's memory amount to the amount needed for the host physical
machine's OS and for the host physical machine's minimum swap space - In this example:
50GB guest virtual machine memory + 4GB host physical machine's OS + 4GB minimal swap =
58GB
Subtract this amount from the amount of physical RAM there is on the system - In this example
58GB - 32GB = 26GB
The answer is the amount of swap space that needs to be allocated. In this example 26GB

Note
Overcommitting does not work with all guest virtual machines, but has been found to work in a
desktop virtualization setup with minimal intensive usage or running several identical guest
virtual machines with KSM. It should be noted that configuring swap and memory overcommit
is not a simple plug-in and configure formula, as each environment and setup is different.
Proceed with caution before changing these settings and make sure you completely
understand your environment and setup before making any changes.
For more information on KSM and overcommitting, refer to Chapter 7, KSM.

6.2. Overcommit t ing Virt ualiz ed CPUs

The KVM hypervisor supports overcommitting virtualized CPUs. Virtualized CPUs can be
overcommitted as far as load limits of guest virtual machines allow. Use caution when overcommitting
VCPUs as loads near 100% may cause dropped requests or unusable response times.
Virtualized CPUs (VCPUs) are overcommitted best when a single host physical machine has multiple
guest virtual machines, where the guests do not share the same VCPU. The Linux scheduler is very
efficient with this type of load. KVM should safely support guest virtual machines with loads under
100% at a ratio of five VCPUs (on 5 virtual machines) to one physical CPU on one single host
physical machine. KVM will switch between all of the machines making sure that the load is balanced.
You cannot overcommit symmetric multiprocessing guest virtual machines on more than the physical
number of processing cores. For example a guest virtual machine with four VCPUs should not be run
on a host physical machine with a dual core processor. Overcommitting symmetric multiprocessing
guest virtual machines in over the physical number of processing cores will cause significant
performance degradation.
Assigning guest virtual machines VCPUs up to the number of physical cores is appropriate and
works as expected. For example, running guest virtual machines with four VCPUs on a quad core
host. Guest virtual machines with less than 100% loads should function effectively in this setup.

47

Virt ualiz at ion Administ rat ion G uide

Important
D o not overcommit memory or CPUs in a production environment without extensive testing.
Applications which use 100% of memory or processing resources may become unstable in
overcommitted environments. Test before deploying.
For more information on how to get the best performance out of your virtual machine, refer to the Red
Hat Enterprise Linux 6 Virtualization Tuning and Optimization Guide.

48

Chapt er 7 . KSM

Chapter 7. KSM
The concept of shared memory is common in modern operating systems. For example, when a
program is first started it shares all of its memory with the parent program. When either the child or
parent program tries to modify this memory, the kernel allocates a new memory region, copies the
original contents and allows the program to modify this new region. This is known as copy on write.
KSM is a new Linux feature which uses this concept in reverse. KSM enables the kernel to examine
two or more already running programs and compare their memory. If any memory regions or pages
are identical, KSM reduces multiple identical memory pages to a single page. This page is then
marked copy on write. If the contents of the page is modified by a guest virtual machine, a new page
is created for that guest virtual machine.
This is useful for virtualization with KVM. When a guest virtual machine is started, it inherits only the
memory from the parent q emu-kvm process. Once the guest virtual machine is running, the contents
of the guest virtual machine operating system image can be shared when guests are running the
same operating system or applications.

Note
The page deduplication technology (used also by the KSM implementation) may introduce
side channels that could potentially be used to leak information across multiple guests. In
case this is a concern, KSM can be disabled on a per-guest basis.
KSM provides enhanced memory speed and utilization. With KSM, common process data is stored in
cache or in main memory. This reduces cache misses for the KVM guests which can improve
performance for some applications and operating systems. Secondly, sharing memory reduces the
overall memory usage of guests which allows for higher densities and greater utilization of
resources.

Note
Starting in Red Hat Enterprise Linux 6.5, KSM is NUMA aware. This allows it to take NUMA
locality into account while coalescing pages, thus preventing performance drops related to
pages being moved to a remote node. Red Hat recommends avoiding cross-node memory
merging when KSM is in use. If KSM is in use, change the
/sys/kernel /mm/ksm/merg e_acro ss_no d es tunable to 0 to avoid merging pages across
NUMA nodes. Kernel memory accounting statistics can eventually contradict each other after
large amounts of cross-node merging. As such, numad can become confused after the KSM
daemon merges large amounts of memory. If your system has a large amount of free memory,
you may achieve higher performance by turning off and disabling the KSM daemon. Refer to
the Red Hat Enterprise Linux Performance Tuning Guide for more information on NUMA.
Red Hat Enterprise Linux uses two separate methods for controlling KSM:
The ksm service starts and stops the KSM kernel thread.
The ksmtuned service controls and tunes the ksm, dynamically managing same-page merging.
The ksmtuned service starts ksm and stops the ksm service if memory sharing is not necessary.
The ksmtuned service must be told with the retune parameter to run when new guests are
created or destroyed.

49

Virt ualiz at ion Administ rat ion G uide

Both of these services are controlled with the standard service management tools.
T h e K SM Service
The ksm service is included in the qemu-kvm package. KSM is off by default on Red Hat Enterprise
Linux 6. When using Red Hat Enterprise Linux 6 as a KVM host physical machine, however, it is likely
turned on by the ksm/ksmtuned services.
When the ksm service is not started, KSM shares only 2000 pages. This default is low and provides
limited memory saving benefits.
When the ksm service is started, KSM will share up to half of the host physical machine system's main
memory. Start the ksm service to enable KSM to share more memory.
# service ksm start
Starting ksm:

OK

The ksm service can be added to the default startup sequence. Make the ksm service persistent with
the chkconfig command.
# chkconfig ksm on
T h e K SM T u n in g Service
The ksmtuned service does not have any options. The ksmtuned service loops and adjusts ksm.
The ksmtuned service is notified by libvirt when a guest virtual machine is created or destroyed.
# service ksmtuned start
Starting ksmtuned:

OK

The ksmtuned service can be tuned with the retune parameter. The retune parameter instructs
ksmtuned to run tuning functions manually.
Before changing the parameters in the file, there are a few terms that need to be clarified:
thres - Activation threshold, in kbytes. A KSM cycle is triggered when the thres value added to
the sum of all qemu-kvm processes RSZ exceeds total system memory. This parameter is the
equivalent in kbytes of the percentage defined in KSM_THRES_COEF.
The /etc/ksmtuned . co nf file is the configuration file for the ksmtuned service. The file output
below is the default ksmtuned . co nf file.
# Configuration file for ksmtuned.
# How long ksmtuned should sleep between tuning adjustments
# KSM_MONITOR_INTERVAL=60
# Millisecond sleep between ksm scans for 16Gb server.
# Smaller servers sleep more, bigger sleep less.
# KSM_SLEEP_MSEC=10
# KSM_NPAGES_BOOST is added to the npages value, when free memory is less
than thres.
# KSM_NPAGES_BOOST=300
# KSM_NPAGES_DECAY Value given is subtracted to the npages value, when

50

Chapt er 7 . KSM

free memory is greater than thres.

# KSM_NPAGES_DECAY=-50
# KSM_NPAGES_MIN is the lower limit for the npages value.
# KSM_NPAGES_MIN=64
# KSM_NAGES_MAX is the upper limit for the npages value.
# KSM_NPAGES_MAX=1250
# KSM_TRES_COEF - is the RAM percentage to be calculated in parameter
thres.
# KSM_THRES_COEF=20
# KSM_THRES_CONST - If this is a low memory system, and the thres value
is less than KSM_THRES_CONST, then reset thres value to KSM_THRES_CONST
value.
# KSM_THRES_CONST=2048
# uncomment the following to enable ksmtuned debug information
# LOGFILE=/var/log/ksmtuned
# DEBUG=1
K SM Variab les an d Mo n it o rin g
KSM stores monitoring data in the /sys/kernel /mm/ksm/ directory. Files in this directory are
updated by the kernel and are an accurate record of KSM usage and statistics.
The variables in the list below are also configurable variables in the /etc/ksmtuned . co nf file as
noted below.
T h e /sys/kernel /mm/ksm/ f iles
f u ll_scan s
Full scans run.
p ag es_sh ared
Total pages shared.
p ag es_sh arin g
Pages presently shared.
p ag es_t o _scan
Pages not scanned.
p ag es_u n sh ared
Pages no longer shared.
p ag es_vo lat ile
Number of volatile pages.
ru n
Whether the KSM process is running.

51

Virt ualiz at ion Administ rat ion G uide

sleep _millisecs
Sleep milliseconds.
KSM tuning activity is stored in the /var/l o g /ksmtuned log file if the DEBUG=1 line is added to the
/etc/ksmtuned . co nf file. The log file location can be changed with the LOGFILE parameter.
Changing the log file location is not advised and may require special configuration of SELinux
settings.
D eact ivat in g K SM
KSM has a performance overhead which may be too large for certain environments or host physical
machine systems.
KSM can be deactivated by stopping the ksmtuned and the ksm service. Stopping the services
deactivates KSM but does not persist after restarting.
# service ksmtuned stop
Stopping ksmtuned:
# service ksm stop
Stopping ksm:

OK

OK

Persistently deactivate KSM with the chkco nfi g command. To turn off the services, run the following
commands:
# chkconfig ksm off
# chkconfig ksmtuned off

Important
Ensure the swap size is sufficient for the committed RAM even with KSM. KSM reduces the RAM
usage of identical or similar guests. Overcommitting guests with KSM without sufficient swap
space may be possible but is not recommended because guest virtual machine memory use
can result in pages becoming unshared.

52

Chapt er 8 . Advanced G uest Virt ual Machine Administ rat ion

Chapter 8. Advanced Guest Virtual Machine Administration

This chapter covers advanced administration tools for fine tuning and controlling system resources
as they are made available to guest virtual machines.

8.1. Cont rol Groups (cgroups)

Red Hat Enterprise Linux 6 provides a new kernel feature: control groups, which are often referred to as
cgroups. Cgroups allow you to allocate resources such as CPU time, system memory, network
bandwidth, or a combination of these resources among user-defined groups of tasks (processes)
running on a system. You can monitor the cgroups you configure, deny cgroups access to certain
resources, and even reconfigure your cgroups dynamically on a running system.
The cgroup functionality is fully supported by lib virt . By default, lib virt puts each guest into a
separate control group for various controllers (such as memory, cpu, blkio, device).
When a guest is started, it is already in a cgroup. The only configuration that may be required is the
setting of policies on the cgroups. Refer to the Red Hat Enterprise Linux Resource Management Guide for
more information on cgroups.

8.2. Huge Page Support

This section provides information about huge page support.
In t ro d u ct io n
x86 CPUs usually address memory in 4kB pages, but they are capable of using larger pages known
as huge pages. KVM guests can be deployed with huge page memory support in order to improve
performance by increasing CPU cache hits against the Transaction Lookaside Buffer (TLB). Huge
pages can significantly increase performance, particularly for large memory and memory-intensive
workloads. Red Hat Enterprise Linux 6 is able to more effectively manage large amounts of memory
by increasing the page size through the use of huge pages.
By using huge pages for a KVM guest, less memory is used for page tables and TLB misses are
reduced, thereby significantly increasing performance, especially for memory-intensive situations.
T ran sp aren t H u g e Pag es
Transparent huge pages (THP) is a kernel feature that reduces TLB entries needed for an application.
By also allowing all free memory to be used as cache, performance is increased.
To use transparent huge pages, no special configuration in the q emu. co nf file is required. Huge
pages are used by default if /sys/kernel /mm/red hat_transparent_hug epag e/enabl ed is set
to always.
Transparent huge pages do not prevent the use of hugetlbfs. However, when hugetlbfs is not used,
KVM will use transparent huge pages instead of the regular 4kB page size.

Note
See the Red Hat Enterprise Linux 7 Virtualization Tuning and Optimization Guide for
instructions on tuning memory performance with huge pages.

53

Virt ualiz at ion Administ rat ion G uide

8.3. Running Red Hat Ent erprise Linux as a Guest Virt ual Machine on a
Hyper-V Hypervisor
It is possible to run a Red Hat Enterprise Linux guest virtual machine on a Microsoft Windows host
physical machine running the Microsoft Windows Hyper-V hypervisor. In particular, the following
enhancements have been made to allow for easier deployment and management of Red Hat
Enterprise Linux guest virtual machines:
Upgraded VMBUS protocols - VMBUS protocols have been upgraded to Windows 8 level. As part
of this work, now VMBUS interrupts can be processed on all available virtual CPUs in the guest.
Furthermore, the signaling protocol between the Red Hat Enterprise Linux guest virtual machine
and the Windows host physical machine has been optimized.
Synthetic frame buffer driver - Provides enhanced graphics performance and superior resolution
for Red Hat Enterprise Linux desktop users.
Live Virtual Machine Backup support - Provisions uninterrupted backup support for live Red Hat
Enterprise Linux guest virtual machines.
D ynamic expansion of fixed size Linux VHD s - Allows expansion of live mounted fixed size Red
Hat Enterprise Linux VHD s.
For more information, refer to the following article: Enabling Linux Support on Windows Server 2012
R2 Hyper-V.

Note
The Hyper-V hypervisor supports shrinking a GPT-partitioned disk on a Red Hat Enterprise
Linux guest if there is free space after the last partition, by allowing the user to drop the
unused last part of the disk. However, this operation will silently delete the secondary GPT
header on the disk, which may produce error messages when the guest examines the partition
table (for example, when printing the partition table with parted ). This is a known limit of
Hyper-V. As a workaround, it is possible to manually restore the secondary GPT header after
shrinking the GPT disk by using the expert menu in g d i sk and the e command. Furthermore,
using the " expand" option in the Hyper-V manager also places the GPT secondary header in
a location other than at the end of disk, but this can be moved with parted . See the g d i sk
and parted man pages for more information on these commands.

8.4 . Guest Virt ual Machine Memory Allocat ion

The following procedure shows how to allocate memory for a guest virtual machine. This allocation
and assignment works only at boot time and any changes to any of the memory values will not take
effect until the next reboot. The maximum memory that can be allocated per guest is 4 TiB, providing
that this memory allocation isn't more than what the host physical machine resources can provide.
Valid memory units include:
b or bytes for bytes
KB for kilobytes (10 3 or blocks of 1,000 bytes)
k or KiB for kibibytes (2 10 or blocks of 1024 bytes)

54

Chapt er 8 . Advanced G uest Virt ual Machine Administ rat ion

MB for megabytes (10 6 or blocks of 1,000,000 bytes)

M or MiB for mebibytes (2 20 or blocks of 1,048,576 bytes)
GB for gigabytes (10 9 or blocks of 1,000,000,000 bytes)
G or GiB for gibibytes (2 30 or blocks of 1,073,741,824 bytes)
TB for terabytes (10 12 or blocks of 1,000,000,000,000 bytes)
T or TiB for tebibytes (2 40 or blocks of 1,099,511,627,776 bytes)
Note that all values will be rounded up to the nearest kibibyte by libvirt, and may be further rounded
to the granularity supported by the hypervisor. Some hypervisors also enforce a minimum, such as
4000KiB (or 4000 x 2 10 or 4,096,000 bytes). The units for this value are determined by the optional
attribute memory unit, which defaults to the kibibytes (KiB) as a unit of measure where the value
given is multiplied by 2 10 or blocks of 1024 bytes.
In the cases where the guest virtual machine crashes the optional attribute dumpCore can be used to
control whether the guest virtual machine's memory should be included in the generated coredump
(dumpCore='on') or not included (dumpCore='off'). Note that the default setting is on so if the
parameter is not set to off, the guest virtual machine memory will be included in the coredump file.
The currentMemory attribute determines the actual memory allocation for a guest virtual machine.
This value can be less than the maximum allocation, to allow for ballooning up the guest virtual
machines memory on the fly. If this is omitted, it defaults to the same value as the memory element.
The unit attribute behaves the same as for memory.
In all cases for this section, the domain XML needs to be altered as follows:
<domain>
<memory unit='KiB' dumpCore='off'>524288</memory>
<!-- changes the memory unit to KiB and does not allow the guest
virtual machine's memory to be included in the generated coredump file ->
<currentMemory unit='KiB'>524288</currentMemory>
<!-- makes the current memory unit 524288 KiB -->
...
</domain>

8.5. Aut omat ically St art ing Guest Virt ual Machines
This section covers how to make guest virtual machines start automatically during the host physical
machine system's boot phase.
This example uses vi rsh to set a guest virtual machine, TestServer, to automatically start when
the host physical machine boots.
# virsh autostart TestServer
Domain TestServer marked as autostarted
The guest virtual machine now automatically starts with the host physical machine.

55

Virt ualiz at ion Administ rat ion G uide

To stop a guest virtual machine automatically booting use the --disable parameter
# virsh autostart --disable TestServer
Domain TestServer unmarked as autostarted
The guest virtual machine no longer automatically starts with the host physical machine.

8.6. Disable SMART Disk Monit oring for Guest Virt ual Machines
SMART disk monitoring can be safely disabled as virtual disks and the physical storage devices are
managed by the host physical machine.
# service smartd stop
# chkconfig --del smartd

8.7. Configuring a VNC Server

To configure a VNC server, use the R emo t e D eskt o p application in Syst em > Pref eren ces.
Alternatively, you can run the vi no -preferences command.
Use the following step set up a dedicated VNC server session:
If needed, Create and then Edit the ~ /. vnc/xstartup file to start a GNOME session whenever
vn cserver is started. The first time you run the vn cserver script it will ask you for a password you
want to use for your VNC session. For more information on vnc server files refer to the Red Hat
Enterprise Linux Installation Guide.

8.8. Generat ing a New Unique MAC Address

In some cases you will need to generate a new and unique MAC address for a guest virtual machine.
There is no command line tool available to generate a new MAC address at the time of writing. The
script provided below can generate a new MAC address for your guest virtual machines. Save the
script on your guest virtual machine as macg en. py. Now from that directory you can run the script
using . /macg en. py and it will generate a new MAC address. A sample output would look like the
following:
$ ./macgen.py
00:16:3e:20:b0:11
#!/usr/bin/python
# macgen.py script to generate a MAC address for guest virtual machines
#
import random
#
def randomMAC():
mac = [ 0x00, 0x16, 0x3e,
random.randint(0x00, 0x7f),
random.randint(0x00, 0xff),
random.randint(0x00, 0xff) ]
return ':'.join(map(lambda x: "%02x" % x, mac))
#
print randomMAC()

56

Chapt er 8 . Advanced G uest Virt ual Machine Administ rat ion

8.8.1. Anot her Met hod t o Generat e a New MAC for Your Guest Virt ual Machine
You can also use the built-in modules of pytho n-vi rti nst to generate a new MAC address and
UUID for use in a guest virtual machine configuration file:
# echo 'import virtinst.util ; print\
virtinst.util.uuidToString(virtinst.util.randomUUID())' | python
# echo 'import virtinst.util ; print virtinst.util.randomMAC()' |
python
The script above can also be implemented as a script file as seen below.
#!/usr/bin/env python
# -*- mode: python; -*print ""
print "New UUID:"
import virtinst.util ; print
virtinst.util.uuidToString(virtinst.util.randomUUID())
print "New MAC:"
import virtinst.util ; print virtinst.util.randomMAC()
print ""

8.9. Improving Guest Virt ual Machine Response T ime

Guest virtual machines can sometimes be slow to respond with certain workloads and usage
patterns. Examples of situations which may cause slow or unresponsive guest virtual machines:
Severely overcommitted memory.
Overcommitted memory with high processor usage
Other (not q emu-kvm processes) busy or stalled processes on the host physical machine.
KVM guest virtual machines function as Linux processes. Linux processes are not permanently kept
in main memory (physical RAM) and will be placed into swap space (virtual memory) especially if
they are not being used. If a guest virtual machine is inactive for long periods of time, the host
physical machine kernel may move the guest virtual machine into swap. As swap is slower than
physical memory it may appear that the guest is not responding. This changes once the guest is
loaded into the main memory. Note that the process of loading a guest virtual machine from swap to
main memory may take several seconds per gigabyte of RAM assigned to the guest virtual machine,
depending on the type of storage used for swap and the performance of the components.
KVM guest virtual machines processes may be moved to swap regardless of whether memory is
overcommitted or overall memory usage.
Using unsafe overcommit levels or overcommitting with swap turned off guest virtual machine
processes or other critical processes is not recommended. Always ensure the host physical machine
has sufficient swap space when overcommitting memory.
For more information on overcommitting with KVM, refer to Section 6.1, Introduction .

57

Virt ualiz at ion Administ rat ion G uide

Warning
Virtual memory allows a Linux system to use more memory than there is physical RAM on the
system. Underused processes are swapped out which allows active processes to use memory,
improving memory utilization. D isabling swap reduces memory utilization as all processes are
stored in physical RAM.
If swap is turned off, do not overcommit guest virtual machines. Overcommitting guest virtual
machines without any swap can cause guest virtual machines or the host physical machine
system to crash.

8.10. Virt ual Machine T imer Management wit h libvirt

Accurate time keeping on guest virtual machines is a key challenge for virtualization platforms.
D ifferent hypervisors attempt to handle the problem of time keeping in a variety of ways. libvirt
provides hypervisor independent configuration settings for time management, using the <clock> and
<timer> elements in the domain XML. The domain XML can be edited using the vi rsh ed i t
command. See Section 14.6, Editing a Guest Virtual Machine's configuration file for details.
The <cl o ck> element is used to determine how the guest virtual machine clock is synchronized with
the host physical machine clock. The clock element has the following attributes:
o ffset determines how the guest virtual machine clock is offset from the host physical machine
clock. The offset attribute has the following possible values:
T ab le 8.1. O f f set at t rib u t e valu es
Valu e

D escrip t io n

utc

The guest virtual machine clock will be

synchronized to UTC when booted.
The guest virtual machine clock will be
synchronized to the host physical machine's
configured timezone when booted, if any.
The guest virtual machine clock will be
synchronized to a given timezone, specified by
the timezone attribute.
The guest virtual machine clock will be
synchronized to an arbitrary offset from UTC.
The delta relative to UTC is specified in
seconds, using the adjustment attribute. The
guest virtual machine is free to adjust the Real
Time Clock (RTC) over time and expect that it
will be honored following the next reboot. This
is in contrast to utc mode, where any RTC
adjustments are lost at each reboot.

localtime

timezone

variable

58

Chapt er 8 . Advanced G uest Virt ual Machine Administ rat ion

Note
The value u t c is set as the clock offset in a virtual machine by default. However, if the guest
virtual machine clock is run with the lo calt ime value, the clock offset needs to be changed
to a different value in order to have the guest virtual machine clock synchronized with the
host physical machine clock.
The ti mezo ne attribute determines which timezone is used for the guest virtual machine clock.
The ad justment attribute provides the delta for guest virtual machine clock synchronization. In
seconds, relative to UTC.

Examp le 8.1. Always syn ch ro n iz e t o U T C

<clock offset="utc" />

Examp le 8.2. Always syn ch ro n iz e t o t h e h o st p h ysical mach in e t imez o n e

<clock offset="localtime" />

Examp le 8.3. Syn ch ro n iz e t o an arb it rary t imez o n e

<clock offset="timezone" timezone="Europe/Paris" />

Examp le 8.4 . Syn ch ro n iz e t o U T C + arb it rary o f f set

<clock offset="variable" adjustment="123456" />

8.10.1. t imer Child Element for clock

A clock element can have zero or more timer elements as children. The timer element specifies a time
source used for guest virtual machine clock synchronization. The timer element has the following
attributes. Only the name is required, all other attributes are optional.
The name attribute dictates the type of the time source to use, and can be one of the following:
T ab le 8.2. n ame at t rib u t e valu es
Valu e

D escrip t io n

pit

Programmable Interval Timer - a timer with

periodic interrupts.
Real Time Clock - a continuously running timer
with periodic interrupts.
Time Stamp Counter - counts the number of ticks
since reset, no interrupts.

rtc
tsc

59

Virt ualiz at ion Administ rat ion G uide

Valu e

D escrip t io n

kvmclock

KVM clock - recommended clock source for KVM

guest virtual machines. KVM pvclock, or kvmclock lets guest virtual machines read the host
physical machines wall clock time.

8.10.2. t rack
The track attribute specifies what is tracked by the timer. Only valid for a name value of rtc.
T ab le 8.3. t rack at t rib u t e valu es
Valu e

D escrip t io n

boot

Corresponds to old host physical machine option,

this is an unsupported tracking option.
RTC always tracks guest virtual machine time.
RTC always tracks host time.

guest
wall

8.10.3. t ickpolicy
The ti ckpo l i cy attribute assigns the policy used to pass ticks on to the guest virtual machine. The
following values are accepted:
T ab le 8.4 . t ickp o licy at t rib u t e valu es
Valu e

D escrip t io n

delay

Continue to deliver at normal rate (i.e. ticks are

delayed).
D eliver at a higher rate to catch up.
Ticks merged into one single tick.
All missed ticks are discarded.

catchup
merge
discard

8.10.4 . frequency, mode, and present

The freq uency attribute is used to set a fixed frequency, and is measured in Hz. This attribute is
only relevant when the name element has a value of tsc. All other timers operate at a fixed frequency
(pi t, rtc).
mo d e determines how the time source is exposed to the guest virtual machine. This attribute is only
relevant for a name value of tsc. All other timers are always emulated. Command is as follows:
<ti mer name= ' tsc' freq uency= ' NNN' mo d e= ' auto | nati ve| emul ate| smpsafe' />.
Mode definitions are given in the table.
T ab le 8.5. mo d e at t rib u t e valu es
Valu e

D escrip t io n

auto

Native if TSC is unstable, otherwise allow native

TSC access.
Always allow native TSC access.
Always emulate TSC.
Always emulate TSC and interlock SMP

native
emulate
smpsafe

60

Chapt er 8 . Advanced G uest Virt ual Machine Administ rat ion

present is used to override the default set of timers visible to the guest virtual machine..
T ab le 8.6 . p resen t at t rib u t e valu es
Valu e

D escrip t io n

yes

Force this timer to the visible to the guest virtual

machine.
Force this timer to not be visible to the guest
virtual machine.

no

8.10.5. Examples Using Clock Synchroniz at ion

Examp le 8.5. C lo ck syn ch ro n iz in g t o lo cal t ime wit h R T C an d PIT t imers
In this example. the clock is synchronized to local time with RTC and PIT timers
<clock offset="localtime">
<timer name="rtc" tickpolicy="catchup" track="guest virtual machine"
/>
<timer name="pit" tickpolicy="delay" />
</clock>

Note
The PIT clocksource can be used with a 32-bit guest running under a 64-bit host (which
cannot use PIT), with the following conditions:
Guest virtual machine may have only one CPU
APIC timer must be disabled (use the " noapictimer" command line option)
NoHZ mode must be disabled in the guest (use the " nohz=off" command line option)
High resolution timer mode must be disabled in the guest (use the " highres=off" command
line option)
The PIT clocksource is not compatible with either high resolution timer mode, or NoHz
mode.

8.11. Using PMU t o Monit or Guest Virt ual Machine Performance

In Red Hat Enterprise Linux 6.4, vPMU (virtual PMU )was introduced as technical-preview. vPMU is
based on Intel's PMU (Performance Monitoring Units) and may only be used on Intel machines. PMU
allows the tracking of statistics which indicate how a guest virtual machine is functioning.
Using performance monitoring, allows developers to use the CPU's PMU counter while using the
performance tool for profiling. The virtual performance monitoring unit feature allows virtual machine
users to identify sources of possible performance problems in their guest virtual machines, thereby
improving the ability to profile a KVM guest virtual machine.
To enable the feature, the -cpu ho st flag must be set.

61

Virt ualiz at ion Administ rat ion G uide

This feature is only supported with guest virtual machines running Red Hat Enterprise Linux 6 and is
disabled by default. This feature only works using the Linux perf tool. Make sure the perf package is
installed using the command:
# yum i nstal l perf.
See the man page on perf for more information on the perf commands.

8.12. Guest Virt ual Machine Power Management

It is possible to forcibly enable or disable BIOS advertisements to the guest virtual machine's
operating system by changing the following parameters in the D omain XML for Libvirt:
...
<pm>
<suspend-to-disk enabled='no'/>
<suspend-to-mem enabled='yes'/>
</pm>
...
The element pm enables ('yes') or disables ('no') BIOS support for S3 (suspend-to-disk) and S4
(suspend-to-mem) ACPI sleep states. If nothing is specified, then the hypervisor will be left with its
default value.

62

Chapt er 9 . G uest virt ual machine device configurat ion

Chapter 9. Guest virtual machine device configuration

Red Hat Enterprise Linux 6 supports three classes of devices for guest virtual machines:
Emulated devices are purely virtual devices that mimic real hardware, allowing unmodified guest
operating systems to work with them using their standard in-box drivers. Red Hat Enterprise Linux
6 supports up to 216 virtio devices.
Virtio devices are purely virtual devices designed to work optimally in a virtual machine. Virtio
devices are similar to emulated devices, however, non-Linux virtual machines do not include the
drivers they require by default. Virtualization management software like the Virtual Machine
Manager (virt - man ag er) and the Red Hat Enterprise Virtualization Hypervisor install these
drivers automatically for supported non-Linux guest operating systems. Red Hat Enterprise Linux
6 supports up to 700 scsi disks.
Assigned devices are physical devices that are exposed to the virtual machine. This method is also
known as 'passthrough'. D evice assignment allows virtual machines exclusive access to PCI
devices for a range of tasks, and allows PCI devices to appear and behave as if they were
physically attached to the guest operating system. Red Hat Enterprise Linux 6 supports up to 32
assigned devices per virtual machine.
D evice assignment is supported on PCIe devices, including select graphics devices. Nvidia K-series
Quadro, GRID , and Tesla graphics card GPU functions are now supported with device assignment
in Red Hat Enterprise Linux 6. Parallel PCI devices may be supported as assigned devices, but they
have severe limitations due to security and system configuration conflicts.

Note
The number of devices that can be attached to a virtual machine depends on several factors.
One factor is the number of files open by the QEMU process (configured in
/etc/securi ty/l i mi ts. co nf, which can be overridden by /etc/l i bvi rt/q emu. co nf).
Other limitation factors include the the number of slots available on the virtual bus, as well as
the system-wide limit on open files set by sysctl.
For more information on specific devices and for limitations refer to Section 20.16, D evices .
Red Hat Enterprise Linux 6 supports PCI hotplug of devices exposed as single function slots to the
virtual machine. Single function host devices and individual functions of multi-function host devices
may be configured to enable this. Configurations exposing devices as multi-function PCI slots to the
virtual machine are recommended only for non-hotplug applications.

63

Virt ualiz at ion Administ rat ion G uide

Note
Platform support for interrupt remapping is required to fully isolate a guest with assigned
devices from the host. Without such support, the host may be vulnerable to interrupt injection
attacks from a malicious guest. In an environment where guests are trusted, the admin may
opt-in to still allow PCI device assignment using the al l o w_unsafe_i nterrupts option to
the vfio_iommu_type1 module. This may either be done persistently by adding a .conf file
(e.g. l o cal . co nf) to /etc/mo d pro be. d containing the following:
options vfio_iommu_type1 allow_unsafe_interrupts=1
or dynamically using the sysfs entry to do the same:
# echo 1 >
/sys/module/vfio_iommu_type1/parameters/allow_unsafe_interrupts

9.1. PCI Devices

PCI device assignment is only available on hardware platforms supporting either Intel VT-d or AMD
IOMMU. These Intel VT-d or AMD IOMMU specifications must be enabled in BIOS for PCI device
assignment to function.
Pro ced u re 9 .1. Prep arin g an In t el syst em f o r PC I d evice assig n men t
1. En ab le t h e In t el VT - d sp ecif icat io n s
The Intel VT-d specifications provide hardware support for directly assigning a physical
device to a virtual machine. These specifications are required to use PCI device assignment
with Red Hat Enterprise Linux.
The Intel VT-d specifications must be enabled in the BIOS. Some system manufacturers
disable these specifications by default. The terms used to refer to these specifications can
differ between manufacturers; consult your system manufacturer's documentation for the
appropriate terms.
2. Act ivat e In t el VT - d in t h e kern el
Activate Intel VT-d in the kernel by adding the intel_iommu=on parameter to the end of the
GRUB_CMD LINX_LINUX line, within the quotes, in the /etc/sysco nfi g /g rub file.
The example below is a modified g rub file with Intel VT-d activated.
GRUB_CMDLINE_LINUX="rd.lvm.lv=vg_VolGroup00/LogVol01
vconsole.font=latarcyrheb-sun16 rd.lvm.lv=vg_VolGroup_1/root
vconsole.keymap=us $([ -x /usr/sbin/rhcrashkernel-param ] & &
/usr/sbin/
rhcrashkernel-param || :) rhgb quiet i ntel _i o mmu= o n"
3. R eg en erat e co n f ig f ile
Regenerate /etc/grub2.cfg by running:

64

Chapt er 9 . G uest virt ual machine device configurat ion

grub2-mkconfig -o /etc/grub2.cfg
Note that if you are using a UEFI-based host, the target file should be /etc/g rub2efi . cfg .
4. R ead y t o u se
Reboot the system to enable the changes. Your system is now capable of PCI device
assignment.
Pro ced u re 9 .2. Prep arin g an AMD syst em f o r PC I d evice assig n men t
1. En ab le t h e AMD IO MMU sp ecif icat io n s
The AMD IOMMU specifications are required to use PCI device assignment in Red Hat
Enterprise Linux. These specifications must be enabled in the BIOS. Some system
manufacturers disable these specifications by default.
2. En ab le IO MMU kern el su p p o rt
Append amd_iommu=on to the end of the GRUB_CMD LINX_LINUX line, within the quotes, in
/etc/sysco nfi g /g rub so that AMD IOMMU specifications are enabled at boot.
3. R eg en erat e co n f ig f ile
Regenerate /etc/grub2.cfg by running:
grub2-mkconfig -o /etc/grub2.cfg
Note that if you are using a UEFI-based host, the target file should be /etc/g rub2efi . cfg .
4. R ead y t o u se
Reboot the system to enable the changes. Your system is now capable of PCI device
assignment.

9.1.1. Assigning a PCI Device wit h virsh

These steps cover assigning a PCI device to a virtual machine on a KVM hypervisor.
This example uses a PCIe network controller with the PCI identifier code, pci _0 0 0 0 _0 1_0 0 _0 , and
a fully virtualized guest machine named guest1-rhel6-64.
Pro ced u re 9 .3. Assig n in g a PC I d evice t o a g u est virt u al mach in e wit h virsh
1. Id en t if y t h e d evice
First, identify the PCI device designated for device assignment to the virtual machine. Use the
l spci command to list the available PCI devices. You can refine the output of l spci with
g rep.
This example uses the Ethernet controller highlighted in the following output:
# lspci | grep Ethernet

65

Virt ualiz at ion Administ rat ion G uide

0 0 : 19 . 0 Ethernet co ntro l l er: Intel C o rpo rati o n 8256 7LM-2 G i g abi t

Netwo rk C o nnecti o n
01:00.0 Ethernet controller: Intel Corporation 82576 Gigabit
Network Connection (rev 01)
01:00.1 Ethernet controller: Intel Corporation 82576 Gigabit
Network Connection (rev 01)
This Ethernet controller is shown with the short identifier 0 0 : 19 . 0 . We need to find out the
full identifier used by vi rsh in order to assign this PCI device to a virtual machine.
To do so, use the vi rsh no d ed ev-l i st command to list all devices of a particular type
(pci ) that are attached to the host machine. Then look at the output for the string that maps
to the short identifier of the device you wish to use.
This example highlights the string that maps to the Ethernet controller with the short identifier
0 0 : 19 . 0 . In this example, the : and . characters are replaced with underscores in the full
identifier.
# virsh nodedev-list --cap pci
pci_0000_00_00_0
pci_0000_00_01_0
pci_0000_00_03_0
pci_0000_00_07_0
pci_0000_00_10_0
pci_0000_00_10_1
pci_0000_00_14_0
pci_0000_00_14_1
pci_0000_00_14_2
pci_0000_00_14_3
pci_0000_0 0 _19 _0
pci_0000_00_1a_0
pci_0000_00_1a_1
pci_0000_00_1a_2
pci_0000_00_1a_7
pci_0000_00_1b_0
pci_0000_00_1c_0
pci_0000_00_1c_1
pci_0000_00_1c_4
pci_0000_00_1d_0
pci_0000_00_1d_1
pci_0000_00_1d_2
pci_0000_00_1d_7
pci_0000_00_1e_0
pci_0000_00_1f_0
pci_0000_00_1f_2
pci_0000_00_1f_3
pci_0000_01_00_0
pci_0000_01_00_1
pci_0000_02_00_0
pci_0000_02_00_1
pci_0000_06_00_0
pci_0000_07_02_0
pci_0000_07_03_0
Record the PCI device number that maps to the device you want to use; this is required in
other steps.

66

Chapt er 9 . G uest virt ual machine device configurat ion

2. R eview d evice in f o rmat io n

Information on the domain, bus, and function are available from output of the vi rsh
no d ed ev-d umpxml command:
virsh nodedev-dumpxml pci_0000_00_19_0
<device>
<name>pci_0000_00_19_0</name>
<parent>computer</parent>
<driver>
<name>e1000e</name>
</driver>
<capability type='pci'>
<domain>0</domain>
<bus>0</bus>
<slot>25</slot>
<function>0</function>
<product id='0x1502'>82579LM Gigabit Network
Connection</product>
<vendor id='0x8086'>Intel Corporation</vendor>
<iommuGroup number='7'>
<address domain='0x0000' bus='0x00' slot='0x19'
function='0x0'/>
</iommuGroup>
</capability>
</device>

67

Virt ualiz at ion Administ rat ion G uide

Note
An IOMMU group is determined based on the visibility and isolation of devices from the
perspective of the IOMMU. Each IOMMU group may contain one or more devices. When
multiple devices are present, all endpoints within the IOMMU group must be claimed for
any device within the group to be assigned to a guest. This can be accomplished
either by also assigning the extra endpoints to the guest or by detaching them from the
host driver using vi rsh no d ed ev-d etach. D evices contained within a single group
may not be split between multiple guests or split between host and guest. Nonendpoint devices such as PCIe root ports, switch ports, and bridges should not be
detached from the host drivers and will not interfere with assignment of endpoints.
D evices within an IOMMU group can be determined using the iommuGroup section of
the vi rsh no d ed ev-d umpxml output. Each member of the group is provided via a
separate " address" field. This information may also be found in sysfs using the
following:
$ ls /sys/bus/pci/devices/0000:01:00.0/iommu_group/devices/
An example of the output from this would be:
0000:01:00.0

0000:01:00.1

To assign only 0000.01.00.0 to the guest, the unused endpoint should be detached
from the host before starting the guest:
$ virsh nodedev-detach pci_0000_01_00_1

3. D et ermin e req u ired co n f ig u rat io n d et ails

Refer to the output from the vi rsh no d ed ev-d umpxml pci _0 0 0 0 _0 0 _19 _0 command
for the values required for the configuration file.
The example device has the following values: bus = 0, slot = 25 and function = 0. The decimal
configuration uses those three values:
bus='0'
slot='25'
function='0'
4. Ad d co n f ig u rat io n d et ails
Run vi rsh ed i t, specifying the virtual machine name, and add a device entry in the
<so urce> section to assign the PCI device to the guest virtual machine.
# virsh edit guest1-rhel6-64
<hostdev mode='subsystem' type='pci' managed='yes'>
<source>
<address domain='0' bus='0' slot='25' function='0'/>
</source>
</hostdev>

68

Chapt er 9 . G uest virt ual machine device configurat ion

Alternately, run vi rsh attach-d evi ce, specifying the virtual machine name and the
guest's XML file:
virsh attach-device guest1-rhel6-64 fi l e. xml
5. St art t h e virt u al mach in e
# virsh start guest1-rhel6-64
The PCI device should now be successfully assigned to the virtual machine, and accessible to the
guest operating system.

9.1.2. Assigning a PCI Device wit h virt -manager

PCI devices can be added to guest virtual machines using the graphical vi rt-manag er tool. The
following procedure adds a Gigabit Ethernet controller to a guest virtual machine.
Pro ced u re 9 .4 . Assig n in g a PC I d evice t o a g u est virt u al mach in e u sin g virt - man ag er
1. O p en t h e h ard ware set t in g s
Open the guest virtual machine and click the Ad d Hard ware button to add a new device to
the virtual machine.

Fig u re 9 .1. T h e virt u al mach in e h ard ware in f o rmat io n win d o w

69

Virt ualiz at ion Administ rat ion G uide

2. Select a PC I d evice
Select PC I H o st D evice from the Hard ware list on the left.
Select an unused PCI device. If you select a PCI device that is in use by another guest an
error may result. In this example, a spare 82576 network device is used. Click Fi ni sh to
complete setup.

Fig u re 9 .2. T h e Ad d n ew virt u al h ard ware wiz ard

3. Ad d t h e n ew d evice
The setup is complete and the guest virtual machine now has direct access to the PCI device.

70

Chapt er 9 . G uest virt ual machine device configurat ion

Fig u re 9 .3. T h e virt u al mach in e h ard ware in f o rmat io n win d o w

Note
If device assignment fails, there may be other endpoints in the same IOMMU group that are still
attached to the host. There is no way to retrieve group information using virt-manager, but
virsh commands can be used to analyze the bounds of the IOMMU group and if necessary
sequester devices.
Refer to the Note in Section 9.1.1, Assigning a PCI D evice with virsh for more information on
IOMMU groups and how to detach endpoint devices using virsh.

9.1.3. PCI Device Assignment wit h virt -inst all

To use virt - in st all to assign a PCI device, use the --host-device parameter.
Pro ced u re 9 .5. Assig n in g a PC I d evice t o a virt u al mach in e wit h virt - in st all
1. Id en t if y t h e d evice

71

Virt ualiz at ion Administ rat ion G uide

Identify the PCI device designated for device assignment to the guest virtual machine.
# lspci
00:19.0
Network
01:00.0
Network
01:00.1
Network

| grep Ethernet
Ethernet controller: Intel Corporation 82567LM-2 Gigabit
Connection
Ethernet controller: Intel Corporation 82576 Gigabit
Connection (rev 01)
Ethernet controller: Intel Corporation 82576 Gigabit
Connection (rev 01)

The vi rsh no d ed ev-l i st command lists all devices attached to the system, and identifies
each PCI device with a string. To limit output to only PCI devices, run the following command:
# virsh nodedev-list --cap pci
pci_0000_00_00_0
pci_0000_00_01_0
pci_0000_00_03_0
pci_0000_00_07_0
pci_0000_00_10_0
pci_0000_00_10_1
pci_0000_00_14_0
pci_0000_00_14_1
pci_0000_00_14_2
pci_0000_00_14_3
pci_0000_00_19_0
pci_0000_00_1a_0
pci_0000_00_1a_1
pci_0000_00_1a_2
pci_0000_00_1a_7
pci_0000_00_1b_0
pci_0000_00_1c_0
pci_0000_00_1c_1
pci_0000_00_1c_4
pci_0000_00_1d_0
pci_0000_00_1d_1
pci_0000_00_1d_2
pci_0000_00_1d_7
pci_0000_00_1e_0
pci_0000_00_1f_0
pci_0000_00_1f_2
pci_0000_00_1f_3
pci_0000_01_00_0
pci_0000_01_00_1
pci_0000_02_00_0
pci_0000_02_00_1
pci_0000_06_00_0
pci_0000_07_02_0
pci_0000_07_03_0
Record the PCI device number; the number is needed in other steps.
Information on the domain, bus and function are available from output of the vi rsh
no d ed ev-d umpxml command:
# virsh nodedev-dumpxml pci_0000_01_00_0

72

Chapt er 9 . G uest virt ual machine device configurat ion

<device>
<name>pci_0000_01_00_0</name>
<parent>pci_0000_00_01_0</parent>
<driver>
<name>igb</name>
</driver>
<capability type='pci'>
<domain>0</domain>
<bus>1</bus>
<slot>0</slot>
<function>0</function>
<product id='0x10c9'>82576 Gigabit Network Connection</product>
<vendor id='0x8086'>Intel Corporation</vendor>
<iommuGroup number='7'>
<address domain='0x0000' bus='0x00' slot='0x19'
function='0x0'/>
</iommuGroup>
</capability>
</device>

Note
If there are multiple endpoints in the IOMMU group and not all of them are assigned to
the guest, you will need to manually detach the other endpoint(s) from the host by
running the following command before you start the guest:
$ virsh nodedev-detach pci_0000_00_19_1
Refer to the Note in Section 9.1.1, Assigning a PCI D evice with virsh for more
information on IOMMU groups.

2. Ad d t h e d evice
Use the PCI identifier output from the vi rsh no d ed ev command as the value for the -host-device parameter.
virt-install \
--name=guest1-rhel6-64 \
--disk path=/var/lib/libvirt/images/guest1-rhel6-64.img,size=8 \
--nonsparse --graphics spice \
--vcpus=2 --ram=2048 \
--location=http://example1.com/installation_tree/RHEL6.0-Serverx86_64/os \
--nonetworks \
--os-type=linux \
--os-variant=rhel6
--host-device=pci_0000_01_00_0
3. C o mp let e t h e in st allat io n
Complete the guest installation. The PCI device should be attached to the guest.

73

Virt ualiz at ion Administ rat ion G uide

9.1.4 . Det aching an Assigned PCI Device

When a host PCI device has been assigned to a guest machine, the host can no longer use the
device. Read this section to learn how to detach the device from the guest with vi rsh or virt man ag er so it is available for host use.
Pro ced u re 9 .6 . D et ach in g a PC I d evice f ro m a g u est wit h virsh
1. D et ach t h e d evice
Use the following command to detach the PCI device from the guest by removing it in the
guest's XML file:
# virsh detach-device name_of_guest file.xml
2. R e- at t ach t h e d evice t o t h e h o st ( o p t io n al)
If the device is in managed mode, skip this step. The device will be returned to the host
automatically.
If the device is not using managed mode, use the following command to re-attach the PCI
device to the host machine:
# virsh nodedev-reattach device
For example, to re-attach the pci _0 0 0 0 _0 1_0 0 _0 device to the host:
virsh nodedev-reattach pci_0000_01_00_0
The device is now available for host use.
Pro ced u re 9 .7. D et ach in g a PC I D evice f ro m a g u est wit h virt - man ag er
1. O p en t h e virt u al h ard ware d et ails screen
In virt - man ag er, double-click on the virtual machine that contains the device. Select the
Sho w vi rtual hard ware d etai l s button to display a list of virtual hardware.

Fig u re 9 .4 . T h e virt u al h ard ware d et ails b u t t o n

2. Select an d remo ve t h e d evice
Select the PCI device to be detached from the list of virtual devices in the left panel.

74

Chapt er 9 . G uest virt ual machine device configurat ion

Fig u re 9 .5. Select in g t h e PC I d evice t o b e d et ach ed

Click the R emo ve button to confirm. The device is now available for host use.

9.1.5. Creat ing PCI Bridges

Peripheral Component Interconnects (PCI) bridges are used to attach to devices such as network
cards, modems and sound cards. Just like their physical counterparts, virtual devices can also be
attached to a PCI Bridge. In the past, only 31 PCI devices could be added to any guest virtual
machine. Now, when a 31st PCI device is added, a PCI bridge is automatically placed in the 31st slot
moving the additional PCI device to the PCI bridge. Each PCI bridge has 31 slots for 31 additional
devices, all of which can be bridges. In this manner, over 900 devices can be available for guest
virtual machines.

Note
This action cannot be performed when the guest virtual machine is running. You must add the
PCI device on a guest virtual machine that is shutdown.

9.1.6. PCI Passt hrough

A PCI network device (specified by the <so urce> element) is directly assigned to the guest using
generic device passthrough, after first optionally setting the device's MAC address to the configured
value, and associating the device with an 802.1Qbh capable switch using an optionally specified
<vi rtual po rt> element (see the examples of virtualport given above for type='direct' network

75

Virt ualiz at ion Administ rat ion G uide

devices). D ue to limitations in standard single-port PCI ethernet card driver design - only SR-IOV
(Single Root I/O Virtualization) virtual function (VF) devices can be assigned in this manner; to
assign a standard single-port PCI or PCIe Ethernet card to a guest, use the traditional <ho std ev>
device definition.
To use VFIO device assignment rather than traditional/legacy KVM device assignment (VFIO is a new
method of device assignment that is compatible with UEFI Secure Boot), a <type= ' ho std ev' >
interface can have an optional driver sub-element with a name attribute set to " vfio" . To use legacy
KVM device assignment you can set name to " kvm" (or simply omit the <d ri ver> element, since
<d ri ver= ' kvm' > is currently the default).

Note
Intelligent passthrough of network devices is very similar to the functionality of a standard
<ho std ev> device, the difference being that this method allows specifying a MAC address
and <vi rtual po rt> for the passed-through device. If these capabilities are not required, if
you have a standard single-port PCI, PCIe, or USB network card that does not support SR-IOV
(and hence would anyway lose the configured MAC address during reset after being assigned
to the guest domain), or if you are using a version of libvirt older than 0.9.11, you should use
standard <ho std ev> to assign the device to the guest instead of <i nterface
type= ' ho std ev' />.

<devices>
<interface type='hostdev'>
<driver name='vfio'/>
<source>
<address type='pci' domain='0x0000' bus='0x00' slot='0x07'
function='0x0'/>
</source>
<mac address='52:54:00:6d:90:02'>
<virtualport type='802.1Qbh'>
<parameters profileid='finance'/>
</virtualport>
</interface>
</devices>
Fig u re 9 .6 . XML examp le f o r PC I d evice assig n men t

9.1.7. Configuring PCI Assignment (Passt hrough) wit h SR-IOV Devices

This section is for SR-IOV devices only. SR-IOV network cards provide multiple Virtual Functions (VFs)
that can each be individually assigned to a guest virtual machines using PCI device assignment.
Once assigned, each will behave as a full physical network device. This permits many guest virtual
machines to gain the performance advantage of direct PCI device assignment, while only using a
single slot on the host physical machine.
These VFs can be assigned to guest virtual machines in the traditional manner using the element
<ho std ev>, but as SR-IOV VF network devices do not have permanent unique MAC addresses, it
causes issues where the guest virtual machine's network settings would have to be re-configured
each time the host physical machine is rebooted. To remedy this, you would need to set the MAC

76

Chapt er 9 . G uest virt ual machine device configurat ion

address prior to assigning the VF to the host physical machine and you would need to set this each
and every time the guest virtual machine boots. In order to assign this MAC address as well as other
options, refert to the procedure described in Procedure 9.8, Configuring MAC addresses, vLAN, and
virtual ports for assigning PCI devices on SR-IOV .
Pro ced u re 9 .8. C o n f ig u rin g MAC ad d resses, vLAN , an d virt u al p o rt s f o r assig n in g PC I
d evices o n SR - IO V
It is important to note that the <ho std ev> element cannot be used for function-specific items like
MAC address assignment, vLAN tag ID assignment, or virtual port assignment because the <mac>,
<vl an>, and <vi rtual po rt> elements are not valid children for <ho std ev>. As they are valid for
<i nterface>, support for a new interface type was added (<i nterface type= ' ho std ev' >).
This new interface device type behaves as a hybrid of an <i nterface> and <ho std ev>. Thus,
before assigning the PCI device to the guest virtual machine, libvirt initializes the network-specific
hardware/switch that is indicated (such as setting the MAC address, setting a vLAN tag, and/or
associating with an 802.1Qbh switch) in the guest virtual machine's XML configuration file. For
information on setting the vLAN tag, refer to Section 18.14, Setting vLAN Tags .
1. Sh u t d o wn t h e g u est virt u al mach in e
Using vi rsh shutd o wn command (refer to Section 14.9.1, Shutting D own a Guest Virtual
Machine ), shutdown the guest virtual machine named guestVM.
# vi rsh shutd o wn guestVM
2. G at h er in f o rmat io n
In order to use <i nterface type= ' ho std ev' >, you must have an SR-IOV-capable
network card, host physical machine hardware that supports either the Intel VT-d or AMD
IOMMU extensions, and you must know the PCI address of the VF that you wish to assign.
3. O p en t h e XML f ile f o r ed it in g
Run the # vi rsh save-i mag e-ed i t command to open the XML file for editing (refer to
Section 14.8.10, Edit D omain XML Configuration Files for more information). As you would
want to restore the guest virtual machine to its former running state, the --runni ng would be
used in this case. The name of the configuration file in this example is guestVM.xml, as the
name of the guest virtual machine is guestVM.
# vi rsh save-i mag e-ed i t guestVM.xml --runni ng
The guestVM.xml opens in your default editor.
4. Ed it t h e XML f ile
Update the configuration file (guestVM.xml) to have a <d evi ces> entry similar to the
following:

<devices>
...
<interface type='hostdev' managed='yes'>
<source>
<address type='pci' domain='0x0' bus='0x00' slot='0x07'

77

Virt ualiz at ion Administ rat ion G uide

function='0x0'/> <!--these values can be decimal as well-->

</source>
<mac address='52:54:00:6d:90:02'/>
<!--sets the mac address-->
<virtualport type='802.1Qbh'>
<!--sets the virtual port for the 802.1Qbh switch-->
<parameters profileid='finance'/>
</virtualport>
<vlan>
<!--sets the vlan tag-->
<tag id='42'/>
</vlan>
</interface>
...
</devices>

Fig u re 9 .7. Samp le d o main XML f o r h o st d ev in t erf ace t yp e

Note that if you do not provide a MAC address, one will be automatically generated, just as
with any other type of interface device. Also, the <vi rtual po rt> element is only used if you
are connecting to an 802.11Qgh hardware switch (802.11Qbg (a.k.a. " VEPA" ) switches are
currently not supported.
5. R e- st art t h e g u est virt u al mach in e
Run the vi rsh start command to restart the guest virtual machine you shutdown in the first
step (example uses guestVM as the guest virtual machine's domain name). Refer to
Section 14.8.1, Starting a D efined D omain for more information.
# vi rsh start guestVM
When the guest virtual machine starts, it sees the network device provided to it by the physical
host machine's adapter, with the configured MAC address. This MAC address will remain
unchanged across guest virtual machine and host physical machine reboots.

9.1.8. Set t ing PCI Device Assignment from a Pool of SR-IOV Virt ual Funct ions
Hard coding the PCI addresses of a particular Virtual Functions (VFs) into a guest's configuration has
two serious limitations:
The specified VF must be available any time the guest virtual machine is started, implying that the
administrator must permanently assign each VF to a single guest virtual machine (or modify the
configuration file for every guest virtual machine to specify a currently unused VF's PCI address
each time every guest virtual machine is started).
If the guest virtual machine is moved to another host physical machine, that host physical
machine must have exactly the same hardware in the same location on the PCI bus (or, again, the
guest virtual machine configuration must be modified prior to start).
It is possible to avoid both of these problems by creating a libvirt network with a device pool
containing all the VFs of an SR-IOV device. Once that is done you would configure the guest virtual
machine to reference this network. Each time the guest is started, a single VF will be allocated from
the pool and assigned to the guest virtual machine. When the guest virtual machine is stopped, the
VF will be returned to the pool for use by another guest virtual machine.

78

Chapt er 9 . G uest virt ual machine device configurat ion

Pro ced u re 9 .9 . C reat in g a d evice p o o l

1. Sh u t d o wn t h e g u est virt u al mach in e
Using vi rsh shutd o wn command (refer to Section 14.9, Shutting D own, Rebooting, and
Forcing Shutdown of a Guest Virtual Machine ), shutdown the guest virtual machine named
guestVM.
# vi rsh shutd o wn guestVM
2. C reat e a co n f ig u rat io n f ile
Using your editor of chocice create an XML file (named passthrough.xml, for example) in the
/tmp directory. Make sure to replace pf d ev= ' eth3' with the netdev name of your own SRIOV device's PF
The following is an example network definition that will make available a pool of all VFs for
the SR-IOV adapter with its physical function (PF) at " eth3' on the host physical machine:

<network>
<name>passthrough</name>
<!--This is the name of the file you created-->
<forward mode='hostdev' managed='yes'>
<pf dev='myNetDevName'/>
<!--Use the netdev name of your SR-IOV devices PF here-->
</forward>
</network>

Fig u re 9 .8. Samp le n et wo rk d ef in it io n d o main XML

3. Lo ad t h e n ew XML f ile
Run the following command, replacing /tmp/passthrough.xml, with the name and location of
your XML file you created in the previous step:
# vi rsh net-d efi ne /tmp/passthrough.xml
4. R est art in g t h e g u est
Run the following replacing passthrough.xml, with the name of your XML file you created in the
previous step:
# vi rsh net-auto start passthrough # vi rsh net-start passthrough
5. R e- st art t h e g u est virt u al mach in e
Run the vi rsh start command to restart the guest virtual machine you shutdown in the first
step (example uses guestVM as the guest virtual machine's domain name). Refer to
Section 14.8.1, Starting a D efined D omain for more information.

79

Virt ualiz at ion Administ rat ion G uide

# vi rsh start guestVM

6. In it iat in g p asst h ro u g h f o r d evices
Although only a single device is shown, libvirt will automatically derive the list of all VFs
associated with that PF the first time a guest virtual machine is started with an interface
definition in its domain XML like the following:

<interface type='network'>
<source network='passthrough'>
</interface>

Fig u re 9 .9 . Samp le d o main XML f o r in t erf ace n et wo rk d ef in it io n

7. Verif icat io n
You can verify this by running vi rsh net-d umpxml passthrough command after starting
the first guest that uses the network; you will get output similar to the following:

<network connections='1'>
<name>passthrough</name>
<uuid>a6b49429-d353-d7ad-3185-4451cc786437</uuid>
<forward mode='hostdev' managed='yes'>
<pf dev='eth3'/>
<address type='pci' domain='0x0000' bus='0x02' slot='0x10'
function='0x1'/>
<address type='pci' domain='0x0000' bus='0x02' slot='0x10'
function='0x3'/>
<address type='pci' domain='0x0000' bus='0x02' slot='0x10'
function='0x5'/>
<address type='pci' domain='0x0000' bus='0x02' slot='0x10'
function='0x7'/>
<address type='pci' domain='0x0000' bus='0x02' slot='0x11'
function='0x1'/>
<address type='pci' domain='0x0000' bus='0x02' slot='0x11'
function='0x3'/>
<address type='pci' domain='0x0000' bus='0x02' slot='0x11'
function='0x5'/>
</forward>
</network>

Fig u re 9 .10. XML d u mp f ile passthrough co n t en t s

80

Chapt er 9 . G uest virt ual machine device configurat ion

9.2. USB Devices

This section gives the commands required for handling USB devices.

9.2.1. Assigning USB Devices t o Guest Virt ual Machines

Most devices such as web cameras, card readers, keyboards, mice etc., are connected to a computer
using a USB port and cable. There are two ways to pass such devices to a guest virtual machine:
Using USB passthrough - this requires the device to be physically connected to the host physical
machine that is hosting the guest virtual machine. SPICE is not needed in this case. USB devices
on the host can be passed to the guest using the command line or virt - man ag er. Refer to
Section 15.3.1, Attaching USB D evices to a Guest Virtual Machine for virt man ag er directions.

Note
virt - man ag er should not be used for hot plugging or hot unplugging devices. If you want
to hot plug/or hot unplug a USB device, refer to Procedure 14.1, Hotplugging USB devices
for use by the guest virtual machine .
Using USB re-direction - USB re-direction is best used in cases where there is a host physical
machine that is running in a data center. The user connects to his/her guest virtual machine from
a local machine or thin client. On this local machine there is a SPICE client. The user can attach
any USB device to the thin client and the SPICE client will redirect the device to the host physical
machine on the data center so it can be used by the guest virtual machine that is running on the
thin client. For instructions on USB re-direction using the virt-manager, refer to Section 15.3.1,
Attaching USB D evices to a Guest Virtual Machine It should be noted that USB redirection is not
possible using the TCP protocol (Refer to BZ #1085318).

9.2.2. Set t ing a Limit on USB Device Redirect ion

To filter out certain devices from redirection, pass the filter property to -device usb-redir. The
filter property takes a string consisting of filter rules, the format for a rule is:
<cl ass>: <vend o r>: <pro d uct>: <versi o n>: <al l o w>
Use the value -1 to designate it to accept any value for a particular field. You may use multiple rules
on the same command line using | as a separator.

Important
If a device matches none of the rule filters, redirecting it will not be allowed!

Examp le 9 .1. An examp le o f limit in g red irect io n wit h a win d o ws g u est virt u al mach in e
1. Prepare a Windows 7 guest virtual machine.
2. Add the following code excerpt to the guest virtual machine's' domain xml file:

81

Virt ualiz at ion Administ rat ion G uide

<redirdev bus='usb' type='spicevmc'>

<alias name='redir0'/>
<address type='usb' bus='0' port='3'/>
</redirdev>
<redirfilter>
<usbdev class='0x08' vendor='0x1234' product='0xBEEF'
version='2.0' allow='yes'/>
<usbdev class='-1' vendor='-1' product='-1' version='-1'
allow='no'/>
</redirfilter>
3. Start the guest virtual machine and confirm the setting changes by running the following:
#ps -ef | g rep $g uest_name
-d evi ce usb-red i r,chard ev= charred i r0 ,i d = red i r0 ,/
fi l ter= 0 x0 8: 0 x1234 : 0 xBEEF: 0 x0 20 0 : 1| -1: -1: -1: 1: 0 ,bus= usb. 0 ,po rt= 3
4. Plug a USB device into a host physical machine, and use virt - man ag er to connect to the
guest virtual machine.
5. Click R ed irect U SB Service in the menu, which will produce the following message:
" Some USB devices are blocked by host policy" . Click O K to confirm and continue.
The filter takes effect.
6. To make sure that the filter captures properly check the USB device vendor and product,
then make the following changes in the guest virtual machine's domain XML to allow for
USB redirection.
<redirfilter>
<usbdev class='0x08' vendor='0x0951' product='0x1625'
version='2.0' allow='yes'/>
<usbdev allow='no'/>
</redirfilter>
7. Restart the guest virtual machine, then use virt - viewer to connect to the guest virtual
machine. The USB device will now redirect traffic to the guest virtual machine.

9.3. Configuring Device Cont rollers

D epending on the guest virtual machine architecture, some device buses can appear more than
once, with a group of virtual devices tied to a virtual controller. Normally, libvirt can automatically
infer such controllers without requiring explicit XML markup, but in some cases it is better to explicitly
set a virtual controller element.

...
<devices>
<controller type='ide' index='0'/>

82

Chapt er 9 . G uest virt ual machine device configurat ion

<controller type='virtio-serial' index='0' ports='16' vectors='4'/>

<controller type='virtio-serial' index='1'>
<address type='pci' domain='0x0000' bus='0x00' slot='0x0a'
function='0x0'/>
</controller>
...
</devices>
...
Fig u re 9 .11. D o main XML examp le f o r virt u al co n t ro llers
Each controller has a mandatory attribute <co ntro l l er type>, which must be one of:
ide
fdc
scsi
sata
usb
ccid
virtio-serial
pci
The <co ntro l l er> element has a mandatory attribute <co ntro l l er i nd ex> which is the decimal
integer describing in which order the bus controller is encountered (for use in controller attributes of
<ad d ress> elements). When <co ntro l l er type = ' vi rti o -seri al ' > there are two additional
optional attributes (named po rts and vecto rs), which control how many devices can be connected
through the controller. Note that Red Hat Enterprise Linux 6 does not support the use of more than 32
vectors per device. Using more vectors will cause failures in migrating the guest virtual machine.
When <co ntro l l er type = ' scsi ' >, there is an optional attribute mo d el model, which can have
the following values:
auto
buslogic
ibmvscsi
lsilogic
lsisas1068
lsisas1078
virtio-scsi
vmpvscsi
When <co ntro l l er type = ' usb' >, there is an optional attribute mo d el model, which can have
the following values:
piix3-uhci

83

Virt ualiz at ion Administ rat ion G uide

piix4-uhci
ehci
ich9-ehci1
ich9-uhci1
ich9-uhci2
ich9-uhci3
vt82c686b-uhci
pci-ohci
nec-xhci

Note
If the USB bus needs to be explicitly disabled for the guest virtual machine,
<mo d el = ' no ne' > may be used. .
For controllers that are themselves devices on a PCI or USB bus, an optional sub-element
<ad d ress> can specify the exact relationship of the controller to its master bus, with semantics as
shown in Section 9.4, Setting Addresses for D evices .
An optional sub-element <d ri ver> can specify the driver specific options. Currently it only supports
attribute queues, which specifies the number of queues for the controller. For best performance, it's
recommended to specify a value matching the number of vCPUs.
USB companion controllers have an optional sub-element <master> to specify the exact
relationship of the companion to its master controller. A companion controller is on the same bus as
its master, so the companion i nd ex value should be equal.
An example XML which can be used is as follows:

...
<devices>
<controller type='usb' index='0'
<address type='pci' domain='0'
</controller>
<controller type='usb' index='0'
<master startport='0'/>
<address type='pci' domain='0'
multifunction='on'/>
</controller>
...
</devices>
...

84

model='ich9-ehci1'>
bus='0' slot='4' function='7'/>
model='ich9-uhci1'>
bus='0' slot='4' function='0'

Chapt er 9 . G uest virt ual machine device configurat ion

Fig u re 9 .12. D o main XML examp le f o r U SB co n t ro llers

PCI controllers have an optional mo d el attribute with the following possible values:
pci-root
pcie-root
pci-bridge
dmi-to-pci-bridge
The root controllers (pci -ro o t and pci e-ro o t) have an optional pci ho l e6 4 element specifying
how big (in kilobytes, or in the unit specified by pci ho l e6 4 's uni t attribute) the 64-bit PCI hole
should be. Some guest virtual machines (such as Windows Server 2003) may cause a crash, unless
uni t is disabled (set to 0 uni t= ' 0 ' ).
For machine types which provide an implicit PCI bus, the pci-root controller with i nd ex= ' 0 ' is
auto-added and required to use PCI devices. pci-root has no address. PCI bridges are auto-added if
there are too many devices to fit on the one bus provided by mo d el = ' pci -ro o t' , or a PCI bus
number greater than zero was specified. PCI bridges can also be specified manually, but their
addresses should only refer to PCI buses provided by already specified PCI controllers. Leaving
gaps in the PCI controller indexes might lead to an invalid configuration. The following XML example
can be added to the <d evi ces> section:

...
<devices>
<controller type='pci' index='0' model='pci-root'/>
<controller type='pci' index='1' model='pci-bridge'>
<address type='pci' domain='0' bus='0' slot='5' function='0'
multifunction='off'/>
</controller>
</devices>
...
Fig u re 9 .13. D o main XML examp le f o r PC I b rid g e
For machine types which provide an implicit PCI Express (PCIe) bus (for example, the machine types
based on the Q35 chipset), the pcie-root controller with i nd ex= ' 0 ' is auto-added to the domain's
configuration. pcie-root has also no address, but provides 31 slots (numbered 1-31) and can only be
used to attach PCIe devices. In order to connect standard PCI devices on a system which has a pcieroot controller, a pci controller with mo d el = ' d mi -to -pci -bri d g e' is automatically added. A
dmi-to-pci-bridge controller plugs into a PCIe slot (as provided by pcie-root), and itself provides 31
standard PCI slots (which are not hot-pluggable). In order to have hot-pluggable PCI slots in the
guest system, a pci-bridge controller will also be automatically created and connected to one of the
slots of the auto-created dmi-to-pci-bridge controller; all guest devices with PCI addresses that are
auto-determined by libvirt will be placed on this pci-bridge device.

...
<devices>

85

Virt ualiz at ion Administ rat ion G uide

<controller type='pci' index='0'

<controller type='pci' index='1'
<address type='pci' domain='0'
</controller>
<controller type='pci' index='2'
<address type='pci' domain='0'
</controller>
</devices>
...

model='pcie-root'/>
model='dmi-to-pci-bridge'>
bus='0' slot='0xe' function='0'/>
model='pci-bridge'>
bus='1' slot='1' function='0'/>

Fig u re 9 .14 . D o main XML examp le f o r PC Ie ( PC I exp ress)

9.4 . Set t ing Addresses for Devices

Many devices have an optional <ad d ress> sub-element which is used to describe where the device
is placed on the virtual bus presented to the guest virtual machine. If an address (or any optional
attribute within an address) is omitted on input, libvirt will generate an appropriate address; but an
explicit address is required if more control over layout is required. See Figure 9.6, XML example for
PCI device assignment for domain XML device examples including an <ad d ress> element.
Every address has a mandatory attribute type that describes which bus the device is on. The choice
of which address to use for a given device is constrained in part by the device and the architecture of
the guest virtual machine. For example, a <d i sk> device uses type= ' d ri ve' , while a <co nso l e>
device would use type= ' pci ' on i686 or x86_64 guest virtual machine architectures. Each
address type has further optional attributes that control where on the bus the device will be placed as
described in the table:
T ab le 9 .1. Su p p o rt ed d evice ad d ress t yp es
Ad d ress t yp e

D escrip t io n

type='pci'

PCI addresses have the following additional

attributes:
domain (a 2-byte hex integer, not currently
used by qemu)
bus (a hex value between 0 and 0xff,
inclusive)
slot (a hex value between 0x0 and 0x1f,
inclusive)
function (a value between 0 and 7, inclusive)
multifunction controls turning on the
multifunction bit for a particular slot/function
in the PCI control register By default it is set
to 'off', but should be set to 'on' for function 0
of a slot that will have multiple functions
used.

86

Chapt er 9 . G uest virt ual machine device configurat ion

Ad d ress t yp e

D escrip t io n

type='drive'

D rive addresses have the following additional

attributes:
controller (a 2-digit controller number)
bus (a 2-digit bus number
target (a 2-digit bus number)
unit (a 2-digit unit number on the bus)

type='virtio-serial'

Each virtio-serial address has the following

additional attributes:
controller (a 2-digit controller number)
bus (a 2-digit bus number)
slot (a 2-digit slot within the bus)

type='ccid'

A CCID address, for smart-cards, has the

following additional attributes:
bus (a 2-digit bus number)
slot attribute (a 2-digit slot within the bus)

type='usb'

USB addresses have the following additional

attributes:
bus (a hex value between 0 and 0xfff,
inclusive)
port (a dotted notation of up to four octets,
such as 1.2 or 2.1.3.1)

type='isa'

ISA addresses have the following additional

attributes:
iobase
irq

9.5. Managing St orage Cont rollers in a Guest Virt ual Machine

Starting from Red Hat Enterprise Linux 6.4, it is supported to add SCSI and virtio-SCSI devices to
guest virtual machines that are running Red Hat Enterprise Linux 6.4 or later. Unlike virtio disks, SCSI
devices require the presence of a controller in the guest virtual machine. Virtio-SCSI provides the
ability to connect directly to SCSI LUNs and significantly improves scalability compared to virtio-blk.
The advantage of virtio-SCSI is that it is capable of handling hundreds of devices compared to virtioblk which can only handle 28 devices and exhausts PCI slots. Virtio-SCSI is now capable of
inheriting the feature set of the target device with the ability to:
attach a virtual hard drive or CD through the virtio-scsi controller,
pass-through a physical SCSI device from the host to the guest via the QEMU scsi-block device,
and allow the usage of hundreds of devices per guest; an improvement from the 28-device limit of
virtio-blk.
This section details the necessary steps to create a virtual SCSI controller (also known as " Host Bus
Adapter" , or HBA) and to add SCSI storage to the guest virtual machine.

87

Virt ualiz at ion Administ rat ion G uide

Pro ced u re 9 .10. C reat in g a virt u al SC SI co n t ro ller

1. D isplay the configuration of the guest virtual machine (G uest1) and look for a pre-existing
SCSI controller:
# virsh dumpxml Guest1 | grep controller.*scsi
If a device controller is present, the command will output one or more lines similar to the
following:
<controller type='scsi' model='virtio-scsi' index='0'/>
2. If the previous step did not show a device controller, create the description for one in a new
file and add it to the virtual machine, using the following steps:
a. Create the device controller by writing a <co ntro l l er> element in a new file and
save this file with an XML extension. vi rti o -scsi -co ntro l l er. xml , for example.
<controller type='scsi' model='virtio-scsi'/>
b. Associate the device controller you just created in vi rti o -scsi -co ntro l l er. xml
with your guest virtual machine (Guest1, for example):
# virsh attach-device --config Guest1 ~/virtio-scsicontroller.xml
In this example the --co nfi g option behaves the same as it does for disks. Refer to
Procedure 13.2, Adding physical block devices to guests for more information.
3. Add a new SCSI disk or CD -ROM. The new disk can be added using the methods in sections
Section 13.3.1, Adding File-based Storage to a Guest and Section 13.3.2, Adding Hard
D rives and Other Block D evices to a Guest . In order to create a SCSI disk, specify a target
device name that starts with sd.
# virsh attach-disk Guest1 /var/lib/libvirt/images/FileName.img sdb
--cache none
D epending on the version of the driver in the guest virtual machine, the new disk may not be
detected immediately by a running guest virtual machine. Follow the steps in the Red Hat
Enterprise Linux Storage Administration Guide.

9.6. Random Number Generat or (RNG) Device

vi rti o -rng is a virtual RNG (random number generator) device that feeds RNG data to the guest
virtual machine's operating system, thereby providing fresh entropy for guest virtual machines on
request.
Using an RNG is particularly useful when a device such as a keyboard, mouse and other inputs are
not enough to generate sufficient entropy on the guest virtual machine. The virtio-rng device is
available for both Red Hat Enterprise Linux and Windows guest virtual machines. Refer to the Note
for instructions on installing the Windows requirements. Unless noted, the following descriptions are
for both Red Hat Enterprise Linux and Windows guest virtual machines.
When virtio-rng is enabled on a Linux guest virtual machine, a chardev is created in the guest virtual

88

Chapt er 9 . G uest virt ual machine device configurat ion

machine at the location /d ev/hwrng . This chardev can then be opened and read to fetch entropy
from the host physical machine. In order for guest virtual machines' applications to benefit from
using randomness from the virtio-rng device transparently, the input from /d ev/hwrng must be
relayed to the kernel entropy pool in the guest virtual machine. This can be accomplished if the
information in this location is coupled with the rgnd daemon (contained within the rng-tools).
This coupling results in the entropy to be routed to the guest virtual machine's /d ev/rand o m file.
The process is done manually in Red Hat Enterprise Linux 6 guest virtual machines.
Red Hat Enterprise Linux 6 guest virtual machines are coupled by running the following command:
# rngd -b -r /dev/hwrng -o /dev/random
For more assistance, run the man rng d command for an explanation of the command options
shown here. For further examples, refer to Procedure 9.11, Implementing virtio-rng with the command
line tools for configuring the virtio-rng device.

Note
Windows guest virtual machines require the driver vi o rng to be installed. Once installed, the
virtual RNG device will work using the CNG (crypto next generation) API provided by Microsoft.
Once the driver is installed, the vi rtrng device appears in the list of RNG providers.

Pro ced u re 9 .11. Imp lemen t in g virt io - rn g wit h t h e co mman d lin e t o o ls

1. Shut down the guest virtual machine.
2. In a terminal window, using the vi rsh ed i t domain-name command, open the XML file for
the desired guest virtual machine.
3. Edit the <d evi ces> element to include the following:

...
<devices>
<rng model='virtio'>
<rate period="2000" bytes="1234"/>
<backend model='random'>/dev/random</backend>
<source mode='bind' service='1234'>
<source mode='connect' host='192.0.2.1' service='1234'>
</backend>
</rng>
</devices>
...

89

Virt ualiz at ion Administ rat ion G uide

Chapter 10. QEMU-img and QEMU Guest Agent

This chapter contain useful hints and tips for using the qemu-img package with guest virtual
machines. If you are looking for information on QEMU trace events and arguments, refer to the
READ ME file located here: /usr/share/d o c/q emu-*/R EAD ME. systemtap.

10.1. Using qemu-img

The q emu-i mg command line tool is used for formatting, modifying and verifying various file
systems used by KVM. q emu-i mg options and usages are listed below.
C h eck
Perform a consistency check on the disk image filename.
#

qemu-img check -f qcow2 --output=qcow2 -r all filename-img.qcow2

Note
Only the q co w2 and vd i formats support consistency checks.
Using the -r tries to repair any inconsistencies that are found during the check, but when used with
-r leaks cluster leaks are repaired and when used with -r all all kinds of errors are fixed. Note
that this has a risk of choosing the wrong fix or hiding corruption issues that may have already
occurred.
C o mmit
Commits any changes recorded in the specified file (filename) to the file's base image with the q emui mg co mmi t command. Optionally, specify the file's format type (format).
# qemu-img commit [-f format] [-t cache] filename

C o n vert
The convert option is used to convert one recognized image format to another image format.
Command format:
# qemu-img convert [-c] [-p] [-f format] [-t cache] [-O output_format] [-o
options] [-S sparse_size] filename output_filename
The -p parameter shows the progress of the command (optional and not for every command) and -S
option allows for the creation of a sparse file, which is included within the disk image. Sparse files in
all purposes function like a standard file, except that the physical blocks that only contain zeros (i.e.,
nothing). When the Operating System sees this file, it treats it as it exists and takes up actual disk

90

Chapt er 1 0 . Q EMU- img and Q EMU G uest Agent

space, even though in reality it doesn't take any. This is particularly helpful when creating a disk for
a guest virtual machine as this gives the appearance that the disk has taken much more disk space
than it has. For example, if you set -S to 50Gb on a disk image that is 10Gb, then your 10Gb of disk
space will appear to be 60Gb in size even though only 10Gb is actually being used.
Convert the disk image filename to disk image output_filename using format output_format.
The disk image can be optionally compressed with the -c option, or encrypted with the -o option by
setting -o encrypti o n. Note that the options available with the -o parameter differ with the selected
format.
Only the q co w2 format supports encryption or compression. q co w2 encryption uses the AES format
with secure 128-bit keys. q co w2 compression is read-only, so if a compressed sector is converted
from q co w2 format, it is written to the new format as uncompressed data.
Image conversion is also useful to get a smaller image when using a format which can grow, such as
q co w or co w. The empty sectors are detected and suppressed from the destination image.

C reat e
Create the new disk image filename of size size and format format.
# qemu-img create [-f format] [-o options] filename [size][preallocation]
If a base image is specified with -o backi ng _fi l e= filename, the image will only record
differences between itself and the base image. The backing file will not be modified unless you use
the co mmi t command. No size needs to be specified in this case.
Preallocation is an option that may only be used with creating qcow2 images. Accepted values
include -o preal l o cati o n= off|meta|full|falloc. Images with preallocated metadata are
larger than images without. However in cases where the image size increases, performance will
improve as the image grows.
It should be noted that using ful l allocation can take a long time with large images. In cases where
you want full allocation and time is of the essence, using fal l o c will save you time.

In f o
The i nfo parameter displays information about a disk image filename. The format for the i nfo
option is as follows:
# qemu-img info [-f format] filename
This command is often used to discover the size reserved on disk which can be different from the
displayed size. If snapshots are stored in the disk image, they are displayed also. This command will
show for example, how much space is being taken by a q co w2 image on a block device. This is
done by running the q emu-i mg . You can check that the image in use is the one that matches the
output of the q emu-i mg i nfo command with the q emu-i mg check command. Refer to
Section 10.1, Using qemu-img .
# qemu-img info /dev/vg-90.100-sluo/lv-90-100-sluo
image: /dev/vg-90.100-sluo/lv-90-100-sluo
file format: qcow2
virtual size: 20G (21474836480 bytes)

91

Virt ualiz at ion Administ rat ion G uide

disk size: 0
cluster_size: 65536

Map
The # q emu-i mg map [-f format] [--o utput= output_format] fi l ename command
dumps the metadata of the image filename and its backing file chain. Specifically, this command
dumps the allocation state of every sector of a specified file, together with the topmost file that
allocates it in the backing file chain. For example, if you have a chain such as c.qcow2 b.qcow2
a.qcow2, a.qcow is the original file, b.qcow2 is the changes made to a.qcow2 and c.qcow2 is the
delta file from b.qcow2. When this chain is created the image files stores the normal image data, plus
information about what is in which file and where it is located within the file. This information is
referred to as the image's metadata. The -f format option is the format of the specified image file.
Formats such as raw, qcow2, vhdx and vmdk may be used. There are two output options possible:
human and jso n.
human is the default setting. It is designed to be more readable to the human eye, and as such,
this format should not be parsed. For clarity and simplicity, the default human format only dumps
known-nonzero areas of the file. Known-zero parts of the file are omitted altogether, and likewise
for parts that are not allocated throughout the chain. When the command is executed, qemu-img
output will identify a file from where the data can be read, and the offset in the file. The output is
displayed as a table with four columns; the first three of which are hexadecimal numbers.
# qemu-img map -f qcow2 --output=human /tmp/test.qcow2
Offset
Length
Mapped to
File
0
0x20000
0x50000
/tmp/test.qcow2
0x100000
0x80000
0x70000
/tmp/test.qcow2
0x200000
0x1f0000
0xf0000
/tmp/test.qcow2
0x3c00000
0x20000
0x2e0000
/tmp/test.qcow2
0x3fd0000
0x10000
0x300000
/tmp/test.qcow2
jso n, or JSON (JavaScript Object Notation), is readable by humans, but as it is a programming
language, it is also designed to be parsed. For example, if you want to parse the output of " qemuimg map" in a parser then you should use the option --o utput= jso n.
# qemu-img map -f qcow2 --output=json /tmp/test.qcow2
[{ "start": 0, "length": 131072, "depth": 0, "zero": false, "data":
true, "offset": 327680},
{ "start": 131072, "length": 917504, "depth": 0, "zero": true, "data":
false},
For more information on the JSON format, refer to the qemu-img(1) man page.
R eb ase
Changes the backing file of an image.
# qemu-img rebase [-f format] [-t cache] [-p] [-u] -b backing_file [-F
backing_format] filename
The backing file is changed to backing_file and (if the format of filename supports the feature), the
backing file format is changed to backing_format.

92

Chapt er 1 0 . Q EMU- img and Q EMU G uest Agent

Note
Only the q co w2 format supports changing the backing file (rebase).
There are two different modes in which rebase can operate: Saf e and U n saf e.
Saf e mo d e is used by default and performs a real rebase operation. The new backing file may differ
from the old one and the q emu-i mg rebase command will take care of keeping the guest virtual
machine-visible content of filename unchanged. In order to achieve this, any clusters that differ
between backing_file and old backing file of filename are merged into filename before making any
changes to the backing file.
Note that safe mode is an expensive operation, comparable to converting an image. The old backing
file is required for it to complete successfully.
U n saf e mo d e is used if the -u option is passed to q emu-i mg rebase. In this mode, only the
backing file name and format of filename is changed, without any checks taking place on the file
contents. Make sure the new backing file is specified correctly or the guest-visible content of the
image will be corrupted.
This mode is useful for renaming or moving the backing file. It can be used without an accessible old
backing file. For instance, it can be used to fix an image whose backing file has already been moved
or renamed.
R esiz e
Change the disk image filename as if it had been created with size size. Only images in raw format can
be resized regardless of version. Red Hat Enterprise Linux 6.1 and later adds the ability to grow (but
not shrink) images in q co w2 format.
Use the following to set the size of the disk image filename to size bytes:
# qemu-img resize filename size
You can also resize relative to the current size of the disk image. To give a size relative to the current
size, prefix the number of bytes with + to grow, or - to reduce the size of the disk image by that
number of bytes. Adding a unit suffix allows you to set the image size in kilobytes (K), megabytes (M),
gigabytes (G) or terabytes (T).
# qemu-img resize filename [+|-]size[K|M|G|T]

Warning
Before using this command to shrink a disk image, you must use file system and partitioning
tools inside the VM itself to reduce allocated file systems and partition sizes accordingly.
Failure to do so will result in data loss.
After using this command to grow a disk image, you must use file system and partitioning tools
inside the VM to actually begin using the new space on the device.

Sn ap sh o t

93

Virt ualiz at ion Administ rat ion G uide

List, apply, create, or delete an existing snapshot (snapshot) of an image (filename).

# qemu-img snapshot [ -l | -a snapshot | -c snapshot | -d snapshot ]
filename
-l lists all snapshots associated with the specified disk image. The apply option, -a, reverts the disk
image (filename) to the state of a previously saved snapshot. -c creates a snapshot (snapshot) of an
image (filename). -d deletes the specified snapshot.

Su p p o rt ed Fo rmat s
q emu - img is designed to convert files to one of the following formats:
raw
Raw disk image format (default). This can be the fastest file-based format. If your file system
supports holes (for example in ext2 or ext3 on Linux or NTFS on Windows), then only the
written sectors will reserve space. Use q emu-i mg i nfo to obtain the real size used by the
image or l s -l s on Unix/Linux. Although Raw images give optimal performance, only very
basic features are available with a Raw image (no snapshots etc.).
q co w2
QEMU image format, the most versatile format with the best feature set. Use it to have
optional AES encryption, zlib-based compression, support of multiple VM snapshots, and
smaller images, which are useful on file systems that do not support holes (non-NTFS file
systems on Windows). Note that this expansive feature set comes at the cost of
performance.
Although only the formats above can be used to run on a guest virtual machine or host physical
machine machine, q emu - img also recognizes and supports the following formats in order to convert
from them into either raw or q co w2 format. The format of an image is usually detected automatically.
In addition to converting these formats into raw or q co w2 , they can be converted back from raw or
q co w2 to the original format.
bo chs
Bochs disk image format.
cl o o p
Linux Compressed Loop image, useful only to reuse directly compressed CD -ROM images
present for example in the Knoppix CD -ROMs.
co w
User Mode Linux Copy On Write image format. The co w format is included only for
compatibility with previous versions. It does not work with Windows.
d mg
Mac disk image format.
nbd
Network block device.
paral l el s

94

Chapt er 1 0 . Q EMU- img and Q EMU G uest Agent

Parallels virtualization disk image format.

q co w
Old QEMU image format. Only included for compatibility with older versions.
vd i
Oracle VM VirtualBox hard disk image format.
vmd k
VMware compatible image format (read-write support for versions 1 and 2, and read-only
support for version 3).
vpc
Windows Virtual PC disk image format. Also referred to as vhd , or Microsoft virtual hard
disk image format.
vvfat
Virtual VFAT disk image format.

10.2. QEMU Guest Agent

The QEMU guest agent runs inside the guest and allows the host machine to issue commands to the
guest operating system using libvirt. The guest operating system then responds to those commands
asynchronously. This chapter covers the libvirt commands and options available to the guest agent.

Important
Note that it is only safe to rely on the guest agent when run by trusted guests. An untrusted
guest may maliciously ignore or abuse the guest agent protocol, and although built-in
safeguards exist to prevent a denial of service attack on the host, the host requires guest cooperation for operations to run as expected.
Note that QEMU guest agent can be used to enable and disable virtual CPUs (vCPUs) while the guest
is running, thus adjusting the number of vCPUs without using the hot plug and hot unplug features.
Refer to Section 14.13.6, Configuring Virtual CPU Count for more information.

10.2.1. Inst all and Enable t he Guest Agent

Install qemu-guest-agent on the guest virtual machine with the yum i nstal l q emu-g uest-ag ent
command and make it run automatically at every boot as a service (qemu-guest-agent.service).

10.2.2. Set t ing up Communicat ion bet ween Guest Agent and Host
The host machine communicates with the guest agent through a VirtIO serial connection between the
host and guest machines. A VirtIO serial channel is connected to the host via a character device
driver (typically a Unix socket), and the guest listens on this serial channel. The following procedure
shows how to set up the host and guest machines for guest agent use.

95

Virt ualiz at ion Administ rat ion G uide

Note
For instructions on how to set up the QEMU guest agent on Windows guests, refer to the
instructions found here.

Pro ced u re 10.1. Set t in g u p co mmu n icat io n b et ween g u est ag en t an d h o st

1. O p en t h e g u est XML
Open the guest XML with the QEMU guest agent configuration. You will need the guest name
to open the file. Use the command # vi rsh l i st on the host machine to list the guests that
it can recognize. In this example, the guest's name is rhel6:
# vi rsh ed i t rhel6
2. Ed it t h e g u est XML f ile
Add the following elements to the XML file and save the changes.

<channel type='unix'>
<source mode='bind' path='/var/lib/libvirt/qemu/rhel6.agent'/>
<target type='virtio' name='org.qemu.guest_agent.0'/>
</channel>

Fig u re 10.1. Ed it in g t h e g u est XML t o co n f ig u re t h e Q EMU g u est ag en t

3. St art t h e Q EMU g u est ag en t in t h e g u est
D ownload and install the guest agent in the guest virtual machine using yum i nstal l
q emu-g uest-ag ent if you have not done so already. Once installed, start the service as
follows:
# servi ce start q emu-g uest-ag ent
You can now communicate with the guest by sending valid libvirt commands over the established
character device driver.

10.2.3. Using t he QEMU Guest Agent

The QEMU guest agent protocol (QEMU GA) package, qemu-guest-agent, is fully supported in Red Hat
Enterprise Linux 6.5 and newer. However, there are the following limitations with regards to isaserial/virtio-serial transport:
The qemu-guest-agent cannot detect whether or not a client has connected to the channel.
There is no way for a client to detect whether or not qemu-guest-agent has disconnected or
reconnected to the back-end.

96

Chapt er 1 0 . Q EMU- img and Q EMU G uest Agent

If the virtio-serial device resets and qemu-guest-agent has not connected to the channel (generally
caused by a reboot or hotplug), data from the client will be dropped.
If qemu-guest-agent has connected to the channel following a virtio-serial device reset, data from
the client will be queued (and eventually throttled if available buffers are exhausted), regardless of
whether or not qemu-guest-agent is still running or connected.

10.2.4 . Using t he QEMU Guest Agent wit h libvirt

Installing the QEMU guest agent allows various other libvirt commands to become more powerful.
The guest agent enhances the following vi rsh commands:
vi rsh shutd o wn --mo d e= ag ent - This shutdown method is more reliable than vi rsh
shutd o wn --mo d e= acpi , as vi rsh shutd o wn used with the QEMU guest agent is
guaranteed to shut down a cooperative guest in a clean state. If the agent is not present, libvirt
has to instead rely on injecting an ACPI shutdown event, but some guests ignore that event and
thus will not shut down.
Can be used with the same syntax for vi rsh rebo o t.
vi rsh snapsho t-create --q ui esce - Allows the guest to flush its I/O into a stable state
before the snapshot is created, which allows use of the snapshot without having to perform a fsck
or losing partial database transactions. The guest agent allows a high level of disk contents
stability by providing guest co-operation.
vi rsh setvcpus --g uest - Instructs the guest to take CPUs offline.
vi rsh d o mpmsuspend - Suspends a running guest gracefully using the guest operating
system's power management functions.

10.2.5. Creat ing a Guest Virt ual Machine Disk Backup

libvirt can communicate with qemu-ga to assure that snapshots of guest virtual machine file systems
are consistent internally and ready for use on an as needed basis. Improvements in Red Hat
Enterprise Linux 6 have been made to make sure that both file and application level synchronization
(flushing) is done. Guest system administrators can write and install application-specific freeze/thaw
hook scripts. Before freezing the filesystems, the qemu-ga invokes the main hook script (included in
the qemu-ga package). The freezing process temporarily deactivates all guest virtual machine
applications.
Just before filesystems are frozen, the following actions occur:
File system applications / databases flush working buffers to the virtual disk and stop accepting
client connections
Applications bring their data files into a consistent state
Main hook script returns
qemu-ga freezes the filesystems and management stack takes a snapshot
Snapshot is confirmed
Filesystem function resumes
Thawing happens in reverse order.
Use the snapsho t-create-as command to create a snapshot of the guest disk. See
Section 14.15.2.2, Creating a snapshot for the current domain for more details on this command.

97

Virt ualiz at ion Administ rat ion G uide

Note
An application-specific hook script might need various SELinux permissions in order to run
correctly, as is done when the script needs to connect to a socket in order to talk to a
database. In general, local SELinux policies should be developed and installed for such
purposes. Accessing file system nodes should work out of the box, after issuing the
resto reco n -FvvR command listed in Table 10.1, QEMU guest agent package contents in
the table row labeled /etc/q emu-g a/fsfreeze-ho o k. d /.
The qemu-guest-agent binary RPM includes the following files:
T ab le 10.1. Q EMU g u est ag en t p ackag e co n t en t s
File n ame

D escrip t io n

/etc/rc. d /i ni t. d /q emu-g a

Service control script (start/stop) for the QEMU

guest agent.
Configuration file for the QEMU guest agent, as
it is read by the /etc/rc. d /i ni t. d /q emug a control script. The settings are documented
in the file with shell script comments.
QEMU guest agent binary file.
Root directory for hook scripts.
Main hook script. No modifications are needed
here.
D irectory for individual, application-specific
hook scripts. The guest system administrator
should copy hook scripts manually into this
directory, ensure proper file mode bits for them,
and then run resto reco n -FvvR on this
directory.
D irectory with sample scripts (for example
purposes only). The scripts contained here are
not executed.

/etc/sysco nfi g /q emu-g a

/usr/bi n/q emu-g a

/usr/l i bexec/q emu-g a/
/usr/l i bexec/q emu-g a/fsfreeze-ho o k
/usr/l i bexec/q emu-g a/fsfreezeho o k. d /

/usr/share/q emu-kvm/q emu-g a/

The main hook script, /usr/l i bexec/q emu-g a/fsfreeze-ho o k logs its own messages, as well
as the application-specific script's standard output and error messages, in the following log file:
/var/l o g /q emu-g a/fsfreeze-ho o k. l o g . For more information, refer to the qemu-guest-agent
wiki page at wiki.qemu.org or libvirt.org.

10.3. Running t he QEMU Guest Agent on a Windows Guest

A Red Hat Enterprise Linux host machine can issue commands to Windows guests by running the
QEMU guest agent in the guest. This is supported in hosts running Red Hat Enterprise Linux 6.5 and
newer, and in the following Windows guest operating systems:
Windows XP Service Pack 3 (VSS is not supported)
Windows Server 2003 R2 - x86 and AMD 64 (VSS is not supported)
Windows Server 2008
Windows Server 2008 R2

98

Chapt er 1 0 . Q EMU- img and Q EMU G uest Agent

Windows 7 - x86 and AMD 64

Windows Server 2012
Windows Server 2012 R2
Windows 8 - x86 and AMD 64
Windows 8.1 - x86 and AMD 64

Note
Windows guest virtual machines require the QEMU guest agent package for Windows, qemuguest-agent-win. This agent is required for VSS (Volume Shadow Copy Service) support for
Windows guest virtual machines running on Red Hat Enterprise Linux. More information can
be found here.

Pro ced u re 10.2. C o n f ig u rin g t h e Q EMU g u est ag en t o n a Win d o ws g u est

Follow these steps for Windows guests running on a Red Hat Enterprise Linux host machine.
1. Prep are t h e R ed H at En t erp rise Lin u x h o st mach in e
Make sure the following package is installed on the Red Hat Enterprise Linux host physical
machine:
virtio-win, located in /usr/share/vi rti o -wi n/
To copy the drivers in the Windows guest, make an *. i so file for the q xl driver using the
following command:
# mki so fs -o /var/l i b/l i bvi rt/i mag es/vi rti o wi n. i so
/usr/share/vi rti o -wi n/d ri vers
2. Prep are t h e Win d o ws g u est
Install the virtio-serial driver in guest by mounting the *. i so to the Windows guest in order
to update the driver. Start the guest, then attach the driver .iso file to the guest as shown
(using a disk named hdb):
# vi rsh attach-d i sk g uest /var/l i b/l i bvi rt/i mag es/vi rti o wi n. i so
hdb
To install the drivers using the Windows C o n t ro l Pan el, navigate to the following menus:
To install the virtio-win driver - Select H ard ware an d So u n d > D evice man ag er >
virt io - serial d river.
3. U p d at e t h e Win d o ws g u est XML co n f ig u rat io n f ile
The guest XML file for the Windows guest is located on the Red Hat Enterprise Linux host
machine. To gain access to this file, you need the Windows guest name. Use the # vi rsh
l i st command on the host machine to list the guests that it can recognize. In this example,
the guest's name is win7x86.

99

Virt ualiz at ion Administ rat ion G uide

Add the following elements to the XML file using the # vi rsh ed i t win7x86 command and
save the changes. Note that the source socket name must be unique in the host, named
win7x86.agent in this example:

...
<channel type='unix'>
<source mode='bind'
path='/var/lib/libvirt/qemu/win7x86.agent'/>
<target type='virtio' name='org.qemu.guest_agent.0'/>
<address type='virtio-serial' controller='0' bus='0'
port='1'/>
</channel>
<channel type='spicevmc'>
<target type='virtio' name='com.redhat.spice.0'/>
<address type='virtio-serial' controller='0' bus='0'
port='2'/>
</channel>
...

Fig u re 10.2. Ed it in g t h e Win d o ws g u est XML t o co n f ig u re t h e Q EMU g u est ag en t

4. R eb o o t t h e Win d o ws g u est
Reboot the Windows guest to apply the changes:
# vi rsh rebo o t win7x86
5. Prep are t h e Q EMU g u est ag en t in t h e Win d o ws g u est
To prepare the guest agent in a Windows guest:
a. In st all t h e lat est virtio-win p ackag e
Run the following command on the Red Hat Enterprise Linux host physical machine
terminal window to locate the file to install. Note that the file shown below may not be
exactly the same as the one your system finds, but it should be latest official version.
# rpm -q a| g rep vi rti o -wi n
virtio-win-1.6.8-5.el6.noarch
# rpm -i v vi rti o -wi n-1. 6 . 8-5. el 6 . no arch
b. C o n f irm t h e in st allat io n co mp let ed
After the virtio-win package finishes installing, check the /usr/share/vi rti o wi n/g uest-ag ent/ folder and you will find an file named qemu-ga-x64.msi or the
qemu-ga-x86.msi as shown:
# ls -l /usr/share/virtio-win/guest-agent/

100

Chapt er 1 0 . Q EMU- img and Q EMU G uest Agent

total 1544
-rw-r--r--. 1 root root 856064 Oct 23 04:58 qemu-ga-x64.msi
-rw-r--r--. 1 root root 724992 Oct 23 04:58 qemu-ga-x86.msi

c. In st all t h e .msi f ile

From the Windows guest (win7x86, for example) install the qemu-ga-x64.msi or the
qemu-ga-x86.msi by double clicking on the file. Once installed, it will be shown as a
qemu-ga service in the Windows guest within the System Manager. This same
manager can be used to monitor the status of the service.

10.3.1. Using libvirt Commands wit h t he QEMU Guest Agent on Windows

Guest s
The QEMU guest agent can use the following vi rsh commands with Windows guests:
vi rsh shutd o wn --mo d e= ag ent - This shutdown method is more reliable than vi rsh
shutd o wn --mo d e= acpi , as vi rsh shutd o wn used with the QEMU guest agent is
guaranteed to shut down a cooperative guest in a clean state. If the agent is not present, libvirt
has to instead rely on injecting an ACPI shutdown event, but some guests ignore that event and
thus will not shut down.
Can be used with the same syntax for vi rsh rebo o t.
vi rsh snapsho t-create --q ui esce - Allows the guest to flush its I/O into a stable state
before the snapshot is created, which allows use of the snapshot without having to perform a fsck
or losing partial database transactions. The guest agent allows a high level of disk contents
stability by providing guest co-operation.
vi rsh d o mpmsuspend - Suspends a running guest gracefully using the guest operating
system's power management functions.

10.4 . Set t ing a Limit on Device Redirect ion

To filter out certain devices from redirection, pass the filter property to -device usb-redir. The
filter property takes a string consisting of filter rules. The format for a rule is:
<cl ass>: <vend o r>: <pro d uct>: <versi o n>: <al l o w>
Use the value -1 to designate it to accept any value for a particular field. You may use multiple rules
on the same command line using | as a separator. Note that if a device matches none of the filter
rules, the redirection will not be allowed.

Examp le 10.1. Limit in g red irect io n wit h a Win d o ws g u est virt u al mach in e
1. Prepare a Windows 7 guest virtual machine.
2. Add the following code excerpt to the guest virtual machine's XML file:
<redirdev bus='usb' type='spicevmc'>
<alias name='redir0'/>

101

Virt ualiz at ion Administ rat ion G uide

<address type='usb' bus='0' port='3'/>

</redirdev>
<redirfilter>
<usbdev class='0x08' vendor='0x1234' product='0xBEEF'
version='2.0' allow='yes'/>
<usbdev class='-1' vendor='-1' product='-1' version='-1'
allow='no'/>
</redirfilter>
3. Start the guest virtual machine and confirm the setting changes by running the following:
# ps -ef | g rep $g uest_name
-d evi ce usb-red i r,chard ev= charred i r0 ,i d = red i r0 ,/
fi l ter= 0 x0 8: 0 x1234 : 0 xBEEF: 0 x0 20 0 : 1| -1: -1: -1: 1: 0 ,bus= usb. 0 ,po rt= 3
4. Plug a USB device into a host physical machine, and use virt - viewer to connect to the
guest virtual machine.
5. Click U SB d evice select io n in the menu, which will produce the following message:
" Some USB devices are blocked by host policy" . Click O K to confirm and continue.
The filter takes effect.
6. To make sure that the filter captures properly check the USB device vendor and product,
then make the following changes in the host physical machine's domain XML to allow for
USB redirection.
<redirfilter>
<usbdev class='0x08' vendor='0x0951' product='0x1625'
version='2.0' allow='yes'/>
<usbdev allow='no'/>
</redirfilter>
7. Restart the guest virtual machine, then use virt - viewer to connect to the guest virtual
machine. The USB device will now redirect traffic to the guest virtual machine.

10.5. Dynamically Changing a Host Physical Machine or a Net work

Bridge t hat is At t ached t o a Virt ual NIC
This section demonstrates how to move the vNIC of a guest virtual machine from one bridge to
another while the guest virtual machine is running without compromising the guest virtual machine
1. Prepare guest virtual machine with a configuration similar to the following:
<interface type='bridge'>
<mac address='52:54:00:4a:c9:5e'/>
<source bridge='virbr0'/>
<model type='virtio'/>
</interface>
2. Prepare an XML file for interface update:

102

Chapt er 1 0 . Q EMU- img and Q EMU G uest Agent

# cat br1. xml

<interface type='bridge'>
<mac address='52:54:00:4a:c9:5e'/>
<source bridge='virbr1'/>
<model type='virtio'/>
</interface>
3. Start the guest virtual machine, confirm the guest virtual machine's network functionality, and
check that the guest virtual machine's vnetX is connected to the bridge you indicated.
# brctl sho w
bridge name
virbr0
virbr0-nic

bridge id
8000.5254007da9f2

STP enabled
yes

vnet0
virbr1
virbr1-nic

8000.525400682996

yes

interfaces

4. Update the guest virtual machine's network with the new interface parameters with the
following command:
# vi rsh upd ate-d evi ce test1 br1. xml
Device updated successfully

5. On the guest virtual machine, run servi ce netwo rk restart. The guest virtual machine
gets a new IP address for virbr1. Check the guest virtual machine's vnet0 is connected to the
new bridge(virbr1)
# brctl sho w
bridge name
virbr0
virbr1
vnet0

bridge id
8000.5254007da9f2
8000.525400682996

STP enabled
yes
yes

interfaces
virbr0-nic
virbr1-nic

103

Virt ualiz at ion Administ rat ion G uide

Chapter 11. Storage Concepts

This chapter introduces the concepts used for describing and managing storage devices. Terms
such as Storage pools and Volumes are explained in the sections that follow.

11.1. St orage Pools

A storage pool is a file, directory, or storage device managed by libvirt for the purpose of providing
storage to guest virtual machines. The storage pool can be local or it can be shared over a network.
A storage pool is a quantity of storage set aside by an administrator, often a dedicated storage
administrator, for use by guest virtual machines. Storage pools are divided into storage volumes
either by the storage administrator or the system administrator, and the volumes are assigned to
guest virtual machines as block devices. In short storage volumes are to partitions what storage
pools are to disks. Although the storage pool is a virtual container it is limited by two factors:
maximum size allowed to it by qemu-kvm and the size of the disk on the host physical machine.
Storage pools may not exceed the size of the disk on the host physical machine. The maximum sizes
are as follows:
virtio-blk = 2^63 bytes or 8 Exabytes(using raw files or disk)
Ext4 = ~ 16 TB (using 4 KB block size)
XFS = ~8 Exabytes
qcow2 and host file systems keep their own metadata and scalability should be evaluated/tuned
when trying very large image sizes. Using raw disks means fewer layers that could affect
scalability or max size.
libvirt uses a directory-based storage pool, the /var/l i b/l i bvi rt/i mag es/ directory, as the
default storage pool. The default storage pool can be changed to another storage pool.
Lo cal st o rag e p o o ls - Local storage pools are directly attached to the host physical machine
server. Local storage pools include: local directories, directly attached disks, physical partitions,
and LVM volume groups. These storage volumes store guest virtual machine images or are
attached to guest virtual machines as additional storage. As local storage pools are directly
attached to the host physical machine server, they are useful for development, testing and small
deployments that do not require migration or large numbers of guest virtual machines. Local
storage pools are not suitable for many production environments as local storage pools do not
support live migration.
N et wo rked ( sh ared ) st o rag e p o o ls - Networked storage pools include storage devices
shared over a network using standard protocols. Networked storage is required when migrating
virtual machines between host physical machines with virt-manager, but is optional when
migrating with virsh. Networked storage pools are managed by libvirt. Supported protocols for
networked storage pools include:
Fibre Channel-based LUNs
iSCSI
NFS
GFS2
SCSI RD MA protocols (SCSI RCP), the block export protocol used in InfiniBand and 10GbE
iWARP adapters.

104

Chapt er 1 1 . St orage Concept s

Note
Multi-path storage pools should not be created or used as they are not fully supported.

11.2. Volumes
Storage pools are divided into storage volumes. Storage volumes are an abstraction of physical
partitions, LVM logical volumes, file-based disk images and other storage types handled by libvirt.
Storage volumes are presented to guest virtual machines as local storage devices regardless of the
underlying hardware.
R ef eren cin g Vo lu mes
To reference a specific volume, three approaches are possible:
T h e n ame o f t h e vo lu me an d t h e st o rag e p o o l
A volume may be referred to by name, along with an identifier for the storage pool it belongs
in. On the virsh command line, this takes the form --pool storage_pool volume_name.
For example, a volume named firstimage in the guest_images pool.
# virsh vol-info --pool guest_images firstimage
Name:
firstimage
Type:
block
Capacity:
20.00 GB
Allocation:
20.00 GB
virsh #
T h e f u ll p at h t o t h e st o rag e o n t h e h o st p h ysical mach in e syst em
A volume may also be referred to by its full path on the file system. When using this
approach, a pool identifier does not need to be included.
For example, a volume named secondimage.img, visible to the host physical machine system
as /images/secondimage.img. The image can be referred to as /images/secondimage.img.
# virsh vol-info /images/secondimage.img
Name:
secondimage.img
Type:
file
Capacity:
20.00 GB
Allocation:
136.00 kB
T h e u n iq u e vo lu me key
When a volume is first created in the virtualization system, a unique identifier is generated
and assigned to it. The unique identifier is termed the volume key. The format of this volume
key varies upon the storage used.
When used with block based storage such as LVM, the volume key may follow this format:
c3pKz4-qPVc-Xf7M-7WNM-WJc8-qSiz-mtvpGn

105

Virt ualiz at ion Administ rat ion G uide

When used with file based storage, the volume key may instead be a copy of the full path to
the volume storage.
/images/secondimage.img
For example, a volume with the volume key of Wlvnf7-a4a3-Tlje-lJDa-9eak-PZBv-LoZuUr:
# virsh vol-info Wlvnf7-a4a3-Tlje-lJDa-9eak-PZBv-LoZuUr
Name:
firstimage
Type:
block
Capacity:
20.00 GB
Allocation:
20.00 GB
vi rsh provides commands for converting between a volume name, volume path, or volume key:
vo l- n ame
Returns the volume name when provided with a volume path or volume key.
# virsh vol-name /dev/guest_images/firstimage
firstimage
# virsh vol-name Wlvnf7-a4a3-Tlje-lJDa-9eak-PZBv-LoZuUr
vo l- p at h
Returns the volume path when provided with a volume key, or a storage pool identifier and
volume name.
# virsh vol-path Wlvnf7-a4a3-Tlje-lJDa-9eak-PZBv-LoZuUr
/dev/guest_images/firstimage
# virsh vol-path --pool guest_images firstimage
/dev/guest_images/firstimage
T h e vo l- key co mman d
Returns the volume key when provided with a volume path, or a storage pool identifier and
volume name.
# virsh vol-key /dev/guest_images/firstimage
Wlvnf7-a4a3-Tlje-lJDa-9eak-PZBv-LoZuUr
# virsh vol-key --pool guest_images firstimage
Wlvnf7-a4a3-Tlje-lJDa-9eak-PZBv-LoZuUr

106

Chapt er 1 2 . St orage Pools

Chapter 12. Storage Pools

This chapter includes instructions on creating storage pools of assorted types. A storage pool is a
quantity of storage set aside by an administrator, often a dedicated storage administrator, for use by
virtual machines. Storage pools are often divided into storage volumes either by the storage
administrator or the system administrator, and the volumes are assigned to guest virtual machines as
block devices.

Examp le 12.1. N FS st o rag e p o o l

Suppose a storage administrator responsible for an NFS server creates a share to store guest
virtual machines' data. The system administrator defines a pool on the host physical machine with
the details of the share (nfs.example.com:/path/to /share should be mounted on /vm_d ata).
When the pool is started, libvirt mounts the share on the specified directory, just as if the system
administrator logged in and executed mo unt nfs. exampl e. co m: /path/to /share /vmd ata.
If the pool is configured to autostart, libvirt ensures that the NFS share is mounted on the directory
specified when libvirt is started.
Once the pool starts, the files that the NFS share, are reported as volumes, and the storage
volumes' paths are then queried using the libvirt APIs. The volumes' paths can then be copied into
the section of a guest virtual machine's XML definition file describing the source storage for the
guest virtual machine's block devices. With NFS, applications using the libvirt APIs can create and
delete volumes in the pool (files within the NFS share) up to the limit of the size of the pool (the
maximum storage capacity of the share). Not all pool types support creating and deleting volumes.
Stopping the pool negates the start operation, in this case, unmounts the NFS share. The data on
the share is not modified by the destroy operation, despite the name. See man virsh for more
details.

Note
Storage pools and volumes are not required for the proper operation of guest virtual
machines. Pools and volumes provide a way for libvirt to ensure that a particular piece of
storage will be available for a guest virtual machine, but some administrators will prefer to
manage their own storage and guest virtual machines will operate properly without any pools
or volumes defined. On systems that do not use pools, system administrators must ensure the
availability of the guest virtual machines' storage using whatever tools they prefer, for
example, adding the NFS share to the host physical machine's fstab so that the share is
mounted at boot time.

12.1. Disk-based St orage Pools

This section covers creating disk based storage devices for guest virtual machines.

107

Virt ualiz at ion Administ rat ion G uide

Warning
Guests should not be given write access to whole disks or block devices (for example,
/d ev/sd b). Use partitions (for example, /d ev/sd b1) or LVM volumes.
If you pass an entire block device to the guest, the guest will likely partition it or create its own
LVM groups on it. This can cause the host physical machine to detect these partitions or LVM
groups and cause errors.

12.1.1. Creat ing a Disk-based St orage Pool Using virsh

This procedure creates a new storage pool using a disk device with the vi rsh command.

Warning
D edicating a disk to a storage pool will reformat and erase all data presently stored on the
disk device. It is strongly recommended to back up the storage device before commencing with
the following procedure.

1. C reat e a G PT d isk lab el o n t h e d isk

The disk must be relabeled with a GUID Partition Table (GPT) disk label. GPT disk labels allow
for creating a large numbers of partitions, up to 128 partitions, on each device. GPT partition
tables can store partition data for far more partitions than the MS-D OS partition table.
# parted /dev/sdb
GNU Parted 2.1
Using /dev/sdb
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) mklabel
New disk label type? gpt
(parted) quit
Information: You may need to update /etc/fstab.
#
2. C reat e t h e st o rag e p o o l co n f ig u rat io n f ile
Create a temporary XML text file containing the storage pool information required for the new
device.
The file must be in the format shown below, and contain the following fields:
<n ame>g u est _imag es_d isk</n ame>
The name parameter determines the name of the storage pool. This example uses
the name guest_images_disk in the example below.
<d evice p at h = ' /dev/sdb' />
The device parameter with the path attribute specifies the device path of the
storage device. This example uses the device /dev/sdb.

108

Chapt er 1 2 . St orage Pools

<t arg et > <p at h >/dev</p at h ></t arg et >

The file system target parameter with the path sub-parameter determines the
location on the host physical machine file system to attach volumes created with
this storage pool.
For example, sdb1, sdb2, sdb3. Using /dev/, as in the example below, means
volumes created from this storage pool can be accessed as /dev/sdb1, /dev/sdb2,
/dev/sdb3.
<f o rmat t yp e= ' gpt' />
The format parameter specifies the partition table type. This example uses the gpt
in the example below, to match the GPT disk label type created in the previous step.
Create the XML file for the storage pool device with a text editor.

Examp le 12.2. D isk b ased st o rag e d evice st o rag e p o o l

<pool type='disk'>
<name>guest_images_disk</name>
<source>
<device path='/dev/sdb'/>
<format type='gpt'/>
</source>
<target>
<path>/dev</path>
</target>
</pool>

3. At t ach t h e d evice
Add the storage pool definition using the vi rsh po o l -d efi ne command with the XML
configuration file created in the previous step.
# virsh pool-define ~/guest_images_disk.xml
Pool guest_images_disk defined from /root/guest_images_disk.xml
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images_disk
inactive
no
4. St art t h e st o rag e p o o l
Start the storage pool with the vi rsh po o l -start command. Verify the pool is started with
the vi rsh po o l -l i st --al l command.
# virsh pool-start guest_images_disk
Pool guest_images_disk started
# virsh pool-list --all
Name
State
Autostart

109

Virt ualiz at ion Administ rat ion G uide

----------------------------------------default
active
yes
guest_images_disk
active
no
5. T u rn o n au t o st art
Turn on autostart for the storage pool. Autostart configures the l i bvi rtd service to start
the storage pool when the service starts.
# virsh pool-autostart guest_images_disk
Pool guest_images_disk marked as autostarted
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images_disk
active
yes
6. Verif y t h e st o rag e p o o l co n f ig u rat io n
Verify the storage pool was created correctly, the sizes reported correctly, and the state
reports as runni ng .
# virsh pool-info guest_images_disk
Name:
guest_images_disk
UUID:
551a67c8-5f2a-012c-3844-df29b167431c
State:
running
Capacity:
465.76 GB
Allocation:
0.00
Available:
465.76 GB
# ls -la /dev/sdb
brw-rw----. 1 root disk 8, 16 May 30 14:08 /dev/sdb
# virsh vol-list guest_images_disk
Name
Path
----------------------------------------7. O p t io n al: R emo ve t h e t emp o rary co n f ig u rat io n f ile
Remove the temporary storage pool XML configuration file if it is not needed.
# rm ~/guest_images_disk.xml
A disk based storage pool is now available.

12.1.2. Delet ing a St orage Pool Using virsh

The following demonstrates how to delete a storage pool using virsh:
1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it.
# virsh pool-destroy guest_images_disk
2. Remove the storage pool's definition

110

Chapt er 1 2 . St orage Pools

# virsh pool-undefine guest_images_disk

12.2. Part it ion-based St orage Pools

This section covers using a pre-formatted block device, a partition, as a storage pool.
For the following examples, a host physical machine has a 500GB hard drive (/d ev/sd c)
partitioned into one 500GB, ext4 formatted partition (/d ev/sd c1). We set up a storage pool for it
using the procedure below.

12.2.1. Creat ing a Part it ion-based St orage Pool Using virt -manager
This procedure creates a new storage pool using a partition of a storage device.
Pro ced u re 12.1. C reat in g a p art it io n - b ased st o rag e p o o l wit h virt - man ag er
1. O p en t h e st o rag e p o o l set t in g s
a. In the vi rt-manag er graphical interface, select the host physical machine from the
main window.
Open the Ed i t menu and select C o nnecti o n D etai l s

Fig u re 12.1. C o n n ect io n D et ails

b. Click on the Sto rag e tab of the C o nnecti o n D etai l s window.

111

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.2. St o rag e t ab

2. C reat e t h e n ew st o rag e p o o l
a. Ad d a n ew p o o l ( p art 1)
Press the + button (the add pool button). The Ad d a New Sto rag e P o o l wizard
appears.
Choose a Name for the storage pool. This example uses the name guest_images_fs.
Change the T ype to fs: P re-Fo rmatted Bl o ck D evi ce.

112

Chapt er 1 2 . St orage Pools

Fig u re 12.3. St o rag e p o o l n ame an d t yp e

Press the Fo rward button to continue.
b. Ad d a n ew p o o l ( p art 2)
Change the T arg et P ath, Fo rmat, and So urce P ath fields.

Fig u re 12.4 . St o rag e p o o l p at h an d f o rmat

T arg et Pat h
Enter the location to mount the source device for the storage pool in the
T arg et P ath field. If the location does not already exist, vi rt-manag er
will create the directory.
Fo rmat
Select a format from the Fo rmat list. The device is formatted with the selected
format.
This example uses the ext4 file system, the default Red Hat Enterprise Linux
file system.
So u rce Pat h
Enter the device in the So urce P ath field.
This example uses the /dev/sdc1 device.
Verify the details and press the Fi ni sh button to create the storage pool.
3. Verif y t h e n ew st o rag e p o o l

113

Virt ualiz at ion Administ rat ion G uide

The new storage pool appears in the storage list on the left after a few seconds. Verify the size
is reported as expected, 458.20 GB Free in this example. Verify the State field reports the new
storage pool as Active.
Select the storage pool. In the Auto start field, click the O n Bo o t checkbox. This will make
sure the storage device starts whenever the l i bvi rtd service starts.

Fig u re 12.5. St o rag e list co n f irmat io n

The storage pool is now created, close the C o nnecti o n D etai l s window.

12.2.2. Delet ing a St orage Pool Using virt -manager

This procedure demonstrates how to delete a storage pool.
1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it. To do this, select the storage pool
you want to stop and click the red X icon at the bottom of the Storage window.

114

Chapt er 1 2 . St orage Pools

Fig u re 12.6 . St o p Ico n

2. D elete the storage pool by clicking the Trash can icon. This icon is only enabled if you stop
the storage pool first.

12.2.3. Creat ing a Part it ion-based St orage Pool Using virsh

This section covers creating a partition-based storage pool with the vi rsh command.

Warning
D o not use this procedure to assign an entire disk as a storage pool (for example,
/d ev/sd b). Guests should not be given write access to whole disks or block devices. Only
use this method to assign partitions (for example, /d ev/sd b1) to storage pools.

Pro ced u re 12.2. C reat in g p re- f o rmat t ed b lo ck d evice st o rag e p o o ls u sin g virsh
1. C reat e t h e st o rag e p o o l d ef in it io n
Use the virsh po o l -d efi ne-as command to create a new storage pool definition. There are
three options that must be provided to define a pre-formatted disk as a storage pool:
Part it io n n ame
The name parameter determines the name of the storage pool. This example uses
the name guest_images_fs in the example below.
d evice

115

Virt ualiz at ion Administ rat ion G uide

The device parameter with the path attribute specifies the device path of the
storage device. This example uses the partition /dev/sdc1.
mo u n t p o in t
The mountpoint on the local file system where the formatted device will be
mounted. If the mount point directory does not exist, the vi rsh command can create
the directory.
The directory /guest_images is used in this example.
# virsh pool-define-as guest_images_fs fs - - /dev/sdc1 "/guest_images"
Pool guest_images_fs defined
The new pool and mount points are now created.
2. Verif y t h e n ew p o o l
List the present storage pools.
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images_fs
inactive
no
3. C reat e t h e mo u n t p o in t
Use the vi rsh po o l -bui l d command to create a mount point for a pre-formatted file
system storage pool.
# virsh pool-build guest_images_fs
Pool guest_images_fs built
# ls -la /guest_images
total 8
drwx------. 2 root root 4096 May 31 19:38 .
dr-xr-xr-x. 25 root root 4096 May 31 19:38 ..
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images_fs
inactive
no
4. St art t h e st o rag e p o o l
Use the vi rsh po o l -start command to mount the file system onto the mount point and
make the pool available for use.
# virsh pool-start guest_images_fs
Pool guest_images_fs started
# virsh pool-list --all
Name
State
Autostart

116

Chapt er 1 2 . St orage Pools

----------------------------------------default
active
yes
guest_images_fs
active
no
5. T u rn o n au t o st art
By default, a storage pool defined with vi rsh, is not set to automatically start each time
l i bvi rtd starts. To remedy this, enable the automatic start with the vi rsh po o l auto start command. The storage pool is now automatically started each time l i bvi rtd
starts.
# virsh pool-autostart guest_images_fs
Pool guest_images_fs marked as autostarted
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images_fs
active
yes
6. Verif y t h e st o rag e p o o l
Verify the storage pool was created correctly, the sizes reported are as expected, and the state
is reported as runni ng . Verify there is a " lost+found" directory in the mount point on the file
system, indicating the device is mounted.
# virsh pool-info guest_images_fs
Name:
guest_images_fs
UUID:
c7466869-e82a-a66c-2187-dc9d6f0877d0
State:
running
Persistent:
yes
Autostart:
yes
Capacity:
458.39 GB
Allocation:
197.91 MB
Available:
458.20 GB
# mount | grep /guest_images
/dev/sdc1 on /guest_images type ext4 (rw)
# ls -la /guest_images
total 24
drwxr-xr-x. 3 root root 4096 May 31 19:47 .
dr-xr-xr-x. 25 root root 4096 May 31 19:38 ..
drwx------. 2 root root 16384 May 31 14:18 lost+found

12.2.4 . Delet ing a St orage Pool Using virsh

1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it.
# virsh pool-destroy guest_images_disk
2. Optionally, if you want to remove the directory where the storage pool resides use the
following command:
# virsh pool-delete guest_images_disk

117

Virt ualiz at ion Administ rat ion G uide

3. Remove the storage pool's definition

# virsh pool-undefine guest_images_disk

12.3. Direct ory-based St orage Pools

This section covers storing guest virtual machines in a directory on the host physical machine.
D irectory-based storage pools can be created with vi rt-manag er or the vi rsh command line
tools.

12.3.1. Creat ing a Direct ory-based St orage Pool wit h virt -manager
1. C reat e t h e lo cal d irect o ry
a. O p t io n al: C reat e a n ew d irect o ry f o r t h e st o rag e p o o l
Create the directory on the host physical machine for the storage pool. This example
uses a directory named /guest virtual machine_images.
# mkdir /guest_images
b. Set d irect o ry o wn ersh ip
Change the user and group ownership of the directory. The directory must be owned
by the root user.
# chown root:root /guest_images
c. Set d irect o ry p ermissio n s
Change the file permissions of the directory.
# chmod 700 /guest_images
d. Verif y t h e ch an g es
Verify the permissions were modified. The output shows a correctly configured empty
directory.
# ls -la /guest_images
total 8
drwx------. 2 root root 4096 May 28 13:57 .
dr-xr-xr-x. 26 root root 4096 May 28 13:57 ..
2. C o n f ig u re SELin u x f ile co n t ext s
Configure the correct SELinux context for the new directory. Note that the name of the pool
and the directory do not have to match. However, when you shutdown the guest virtual
machine, libvirt has to set the context back to a default value. The context of the directory
determines what this default value is. It is worth explicitly labeling the directory virt_image_t,

118

Chapt er 1 2 . St orage Pools

so that when the guest virtual machine is shutdown, the images get labeled 'virt_image_t' and
are thus isolated from other processes running on the host physical machine.
# semanage fcontext -a -t virt_image_t '/guest_images(/.*)?'
# restorecon -R /guest_images
3. O p en t h e st o rag e p o o l set t in g s
a. In the vi rt-manag er graphical interface, select the host physical machine from the
main window.
Open the Ed i t menu and select C o nnecti o n D etai l s

Fig u re 12.7. C o n n ect io n d et ails win d o w

b. Click on the Sto rag e tab of the C o nnecti o n D etai l s window.

119

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.8. St o rag e t ab

4. C reat e t h e n ew st o rag e p o o l
a. Ad d a n ew p o o l ( p art 1)
Press the + button (the add pool button). The Ad d a New Sto rag e P o o l wizard
appears.
Choose a Name for the storage pool. This example uses the name guest_images.
Change the T ype to d i r: Fi l esystem D i recto ry.

Fig u re 12.9 . N ame t h e st o rag e p o o l

120

Chapt er 1 2 . St orage Pools

Press the Fo rward button to continue.

b. Ad d a n ew p o o l ( p art 2)
Change the T arg et P ath field. For example, /guest_images.
Verify the details and press the Fi ni sh button to create the storage pool.
5. Verif y t h e n ew st o rag e p o o l
The new storage pool appears in the storage list on the left after a few seconds. Verify the size
is reported as expected, 36.41 GB Free in this example. Verify the State field reports the new
storage pool as Active.
Select the storage pool. In the Auto start field, confirm that the O n Bo o t checkbox is
checked. This will make sure the storage pool starts whenever the l i bvi rtd service starts.

Fig u re 12.10. Verif y t h e st o rag e p o o l in f o rmat io n

The storage pool is now created, close the C o nnecti o n D etai l s window.

12.3.2. Delet ing a St orage Pool Using virt -manager

This procedure demonstrates how to delete a storage pool.
1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it. To do this, select the storage pool
you want to stop and click the red X icon at the bottom of the Storage window.

121

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.11. St o p Ico n

2. D elete the storage pool by clicking the Trash can icon. This icon is only enabled if you stop
the storage pool first.

12.3.3. Creat ing a Direct ory-based St orage Pool wit h virsh

1. C reat e t h e st o rag e p o o l d ef in it io n
Use the vi rsh po o l -d efi ne-as command to define a new storage pool. There are two
options required for creating directory-based storage pools:
The name of the storage pool.
This example uses the name guest_images. All further vi rsh commands used in this
example use this name.
The path to a file system directory for storing guest image files. If this directory does not
exist, vi rsh will create it.
This example uses the /guest_images directory.
# virsh pool-define-as guest_images dir - - - - "/guest_images"
Pool guest_images defined
2. Verif y t h e st o rag e p o o l is list ed
Verify the storage pool object is created correctly and the state reports it as i nacti ve.
# virsh pool-list --all
Name
State

122

Autostart

Chapt er 1 2 . St orage Pools

----------------------------------------default
active
yes
guest_images
inactive
no
3. C reat e t h e lo cal d irect o ry
Use the vi rsh po o l -bui l d command to build the directory-based storage pool for the
directory guest_images (for example), as shown:
# virsh pool-build guest_images
Pool guest_images built
# ls -la /guest_images
total 8
drwx------. 2 root root 4096 May 30 02:44 .
dr-xr-xr-x. 26 root root 4096 May 30 02:44 ..
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images
inactive
no
4. St art t h e st o rag e p o o l
Use the virsh command po o l -start to enable a directory storage pool, thereby allowing
allowing volumes of the pool to be used as guest disk images.
# virsh pool-start guest_images
Pool guest_images started
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images
active
no
5. T u rn o n au t o st art
Turn on autostart for the storage pool. Autostart configures the l i bvi rtd service to start
the storage pool when the service starts.
# virsh pool-autostart guest_images
Pool guest_images marked as autostarted
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images
active
yes
6. Verif y t h e st o rag e p o o l co n f ig u rat io n
Verify the storage pool was created correctly, the size is reported correctly, and the state is
reported as runni ng . If you want the pool to be accessible even if the guest virtual machine
is not running, make sure that P ersi stent is reported as yes. If you want the pool to start
automatically when the service starts, make sure that Auto start is reported as yes.

123

Virt ualiz at ion Administ rat ion G uide

# virsh pool-info guest_images

Name:
guest_images
UUID:
779081bf-7a82-107b-2874-a19a9c51d24c
State:
running
Persistent:
yes
Autostart:
yes
Capacity:
49.22 GB
Allocation:
12.80 GB
Available:
36.41 GB
# ls -la /guest_images
total 8
drwx------. 2 root root 4096 May 30 02:44 .
dr-xr-xr-x. 26 root root 4096 May 30 02:44 ..
#
A directory-based storage pool is now available.

12.3.4 . Delet ing a St orage Pool Using virsh

The following demonstrates how to delete a storage pool using virsh:
1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it.
# virsh pool-destroy guest_images_disk
2. Optionally, if you want to remove the directory where the storage pool resides use the
following command:
# virsh pool-delete guest_images_disk
3. Remove the storage pool's definition
# virsh pool-undefine guest_images_disk

12.4 . LVM-based St orage Pools

This chapter covers using LVM volume groups as storage pools.
LVM-based storage groups provide the full flexibility of LVM.

Note
Thin provisioning is currently not possible with LVM based storage pools.

124

Chapt er 1 2 . St orage Pools

Note
Please refer to the Red Hat Enterprise Linux Storage Administration Guide for more details on LVM.

Warning
LVM-based storage pools require a full disk partition. If activating a new partition/device with
these procedures, the partition will be formatted and all data will be erased. If using the host's
existing Volume Group (VG) nothing will be erased. It is recommended to back up the storage
device before commencing the following procedure.

12.4 .1. Creat ing an LVM-based St orage Pool wit h virt -manager
LVM-based storage pools can use existing LVM volume groups or create new LVM volume groups on
a blank partition.
1. O p t io n al: C reat e n ew p art it io n f o r LVM vo lu mes
These steps describe how to create a new partition and LVM volume group on a new hard
disk drive.

Warning
This procedure will remove all data from the selected storage device.

a. C reat e a n ew p art it io n
Use the fd i sk command to create a new disk partition from the command line. The
following example creates a new partition that uses the entire disk on the storage
device /d ev/sd b.
# fdisk /dev/sdb
Command (m for help):
Press n for a new partition.
b. Press p for a primary partition.
Command action
e
extended
p
primary partition (1-4)
c. Choose an available partition number. In this example the first partition is chosen by
entering 1.
Partition number (1-4): 1
d. Enter the default first cylinder by pressing Enter.

125

Virt ualiz at ion Administ rat ion G uide

First cylinder (1-400, default 1):

e. Select the size of the partition. In this example the entire disk is allocated by pressing
Enter.
Last cylinder or +size or +sizeM or +sizeK (2-400, default
400):
f. Set the type of partition by pressing t.
Command (m for help): t
g. Choose the partition you created in the previous steps. In this example, the partition
number is 1.
Partition number (1-4): 1
h. Enter 8e for a Linux LVM partition.
Hex code (type L to list codes): 8e
i. write changes to disk and quit.
Command (m for help): w
Command (m for help): q
j. C reat e a n ew LVM vo lu me g ro u p
Create a new LVM volume group with the vg create command. This example creates
a volume group named guest_images_lvm.
# vgcreate guest_images_lvm /dev/sdb1
Physical volume "/dev/vdb1" successfully created
Volume group "guest_images_lvm" successfully created
The new LVM volume group, guest_images_lvm, can now be used for an LVM-based storage
pool.
2. O p en t h e st o rag e p o o l set t in g s
a. In the vi rt-manag er graphical interface, select the host from the main window.
Open the Ed i t menu and select C o nnecti o n D etai l s

126

Chapt er 1 2 . St orage Pools

Fig u re 12.12. C o n n ect io n d et ails

b. Click on the Sto rag e tab.

Fig u re 12.13. St o rag e t ab

3. C reat e t h e n ew st o rag e p o o l
a. St art t h e Wiz ard
Press the + button (the add pool button). The Ad d a New Sto rag e P o o l wizard
appears.
Choose a Name for the storage pool. We use guest_images_lvm for this example. Then
change the T ype to l o g i cal : LVM Vo l ume G ro up, and

127

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.14 . Ad d LVM st o rag e p o o l

Press the Fo rward button to continue.
b. Ad d a n ew p o o l ( p art 2)
Change the T arg et P ath field. This example uses /guest_images.
Now fill in the T arg et P ath and So urce P ath fields, then tick the Bui l d P o o l
check box.
Use the T arg et P ath field to either select an existing LVM volume group or as the
name for a new volume group. The default format is /d ev/storage_pool_name.
This example uses a new volume group named /dev/guest_images_lvm.
The So urce P ath field is optional if an existing LVM volume group is used in the
T arg et P ath.
For new LVM volume groups, input the location of a storage device in the So urce
P ath field. This example uses a blank partition /dev/sdc.
The Bui l d P o o l checkbox instructs vi rt-manag er to create a new LVM
volume group. If you are using an existing volume group you should not select the
Bui l d P o o l checkbox.
This example is using a blank partition to create a new volume group so the
Bui l d P o o l checkbox must be selected.

128

Chapt er 1 2 . St orage Pools

Fig u re 12.15. Ad d t arg et an d so u rce

Verify the details and press the Fi ni sh button format the LVM volume group and
create the storage pool.
c. C o n f irm t h e d evice t o b e f o rmat t ed
A warning message appears.

Fig u re 12.16 . Warn in g messag e

Press the Y es button to proceed to erase all data on the storage device and create
the storage pool.
4. Verif y t h e n ew st o rag e p o o l
The new storage pool will appear in the list on the left after a few seconds. Verify the details
are what you expect, 465.76 GB Free in our example. Also verify the State field reports the
new storage pool as Active.
It is generally a good idea to have the Auto start check box enabled, to ensure the storage
pool starts automatically with libvirtd.

129

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.17. C o n f irm LVM st o rag e p o o l d et ails

Close the Host D etails dialog, as the task is now complete.

12.4 .2. Delet ing a St orage Pool Using virt -manager

This procedure demonstrates how to delete a storage pool.
1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it. To do this, select the storage pool
you want to stop and click the red X icon at the bottom of the Storage window.

130

Chapt er 1 2 . St orage Pools

Fig u re 12.18. St o p Ico n

2. D elete the storage pool by clicking the Trash can icon. This icon is only enabled if you stop
the storage pool first.

12.4 .3. Creat ing an LVM-based St orage Pool wit h virsh

This section outlines the steps required to create an LVM-based storage pool with the vi rsh
command. It uses the example of a pool named g u est _imag es_lvm from a single drive
(/d ev/sd c). This is only an example and your settings should be substituted as appropriate.
Pro ced u re 12.3. C reat in g an LVM- b ased st o rag e p o o l wit h virsh
1. D efine the pool name g u est _imag es_lvm.
# virsh pool-define-as guest_images_lvm logical - - /dev/sdc
libvirt_lvm \ /dev/libvirt_lvm
Pool guest_images_lvm defined
2. Build the pool according to the specified name. If you are using an already existing volume
group, skip this step.
# virsh pool-build guest_images_lvm
Pool guest_images_lvm built
3. Initialize the new pool.

131

Virt ualiz at ion Administ rat ion G uide

# virsh pool-start guest_images_lvm

Pool guest_images_lvm started
4. Show the volume group information with the vg s command.
# vgs
VG
#PV #LV #SN Attr
VSize
VFree
libvirt_lvm
1
0
0 wz--n- 465.76g 465.76g
5. Set the pool to start automatically.
# virsh pool-autostart guest_images_lvm
Pool guest_images_lvm marked as autostarted
6. List the available pools with the vi rsh command.
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
guest_images_lvm
active
yes
7. The following commands demonstrate the creation of three volumes (volume1, volume2 and
volume3) within this pool.
# virsh vol-create-as guest_images_lvm volume1 8G
Vol volume1 created
# virsh vol-create-as guest_images_lvm volume2 8G
Vol volume2 created
# virsh vol-create-as guest_images_lvm volume3 8G
Vol volume3 created
8. List the available volumes in this pool with the vi rsh command.
# virsh vol-list guest_images_lvm
Name
Path
----------------------------------------volume1
/dev/libvirt_lvm/volume1
volume2
/dev/libvirt_lvm/volume2
volume3
/dev/libvirt_lvm/volume3
9. The following two commands (l vscan and l vs) display further information about the newly
created volumes.
# lvscan
ACTIVE
ACTIVE
ACTIVE
# lvs

132

'/dev/libvirt_lvm/volume1' [8.00 GiB] inherit

'/dev/libvirt_lvm/volume2' [8.00 GiB] inherit
'/dev/libvirt_lvm/volume3' [8.00 GiB] inherit

Chapt er 1 2 . St orage Pools

LV
VG
Copy% Convert
volume1 libvirt_lvm
volume2 libvirt_lvm
volume3 libvirt_lvm

Attr

LSize

-wi-a-wi-a-wi-a-

8.00g
8.00g
8.00g

Pool Origin Data%

Move Log

12.4 .4 . Delet ing a St orage Pool Using virsh

The following demonstrates how to delete a storage pool using virsh:
1. To avoid any issues with other guests using the same pool, it is best to stop the storage pool
and release any resources in use by it.
# virsh pool-destroy guest_images_disk
2. Optionally, if you want to remove the directory where the storage pool resides use the
following command:
# virsh pool-delete guest_images_disk
3. Remove the storage pool's definition
# virsh pool-undefine guest_images_disk

12.5. iSCSI-based St orage Pools

This section covers using iSCSI-based devices to store guest virtual machines.
iSCSI (Internet Small Computer System Interface) is a network protocol for sharing storage devices.
iSCSI connects initiators (storage clients) to targets (storage servers) using SCSI instructions over
the IP layer.

12.5.1. Configuring a Soft ware iSCSI T arget

The scsi-target-utils package provides a tool for creating software-backed iSCSI targets.
Pro ced u re 12.4 . C reat in g an iSC SI t arg et
1. In st all t h e req u ired p ackag es
Install the scsi-target-utils package and all dependencies
# yum install scsi-target-utils
2. St art t h e t g t d service
The tg td service host physical machines SCSI targets and uses the iSCSI protocol to host
physical machine targets. Start the tg td service and make the service persistent after
restarting with the chkco nfi g command.
# service tgtd start
# chkconfig tgtd on

133

Virt ualiz at ion Administ rat ion G uide

3. O p t io n al: C reat e LVM vo lu mes

LVM volumes are useful for iSCSI backing images. LVM snapshots and resizing can be
beneficial for guest virtual machines. This example creates an LVM image named virtimage1
on a new volume group named virtstore on a RAID 5 array for hosting guest virtual machines
with iSCSI.
a. C reat e t h e R AID array
Creating software RAID 5 arrays is covered by the Red Hat Enterprise Linux Deployment
Guide.
b. C reat e t h e LVM vo lu me g ro u p
Create a volume group named virtstore with the vg create command.
# vgcreate virtstore /dev/md1
c. C reat e a LVM lo g ical vo lu me
Create a logical volume group named virtimage1 on the virtstore volume group with a
size of 20GB using the l vcreate command.
# lvcreate --size 20G -n virtimage1 virtstore
The new logical volume, virtimage1, is ready to use for iSCSI.
4. O p t io n al: C reat e f ile- b ased imag es
File-based storage is sufficient for testing but is not recommended for production
environments or any significant I/O activity. This optional procedure creates a file based
imaged named virtimage2.img for an iSCSI target.
a. C reat e a n ew d irect o ry f o r t h e imag e
Create a new directory to store the image. The directory must have the correct SELinux
contexts.
# mkdir -p /var/lib/tgtd/virtualization
b. C reat e t h e imag e f ile
Create an image named virtimage2.img with a size of 10GB.
# dd if=/dev/zero
of=/var/lib/tgtd/virtualization/virtimage2.img bs=1M
seek=10000 count=0
c. C o n f ig u re SELin u x f ile co n t ext s
Configure the correct SELinux context for the new image and directory.
# restorecon -R /var/lib/tgtd

134

Chapt er 1 2 . St orage Pools

The new file-based image, virtimage2.img, is ready to use for iSCSI.

5. C reat e t arg et s
Targets can be created by adding a XML entry to the /etc/tg t/targ ets. co nf file. The
targ et attribute requires an iSCSI Qualified Name (IQN). The IQN is in the format:
iqn.yyyy-mm.reversed domain name:optional identifier text
Where:
yyyy-mm represents the year and month the device was started (for example: 2010-05);
reversed domain name is the host physical machines domain name in reverse (for example
server1.example.com in an IQN would be com.example.server1); and
optional identifier text is any text string, without spaces, that assists the administrator in
identifying devices or hardware.
This example creates iSCSI targets for the two types of images created in the optional steps
on server1.example.com with an optional identifier trial. Add the following to the
/etc/tg t/targ ets. co nf file.
<target iqn.2010-05.com.example.server1:iscsirhel6guest>
backing-store /dev/virtstore/virtimage1 #LUN 1
backing-store /var/lib/tgtd/virtualization/virtimage2.img
2
write-cache off
</target>

#LUN

Ensure that the /etc/tg t/targ ets. co nf file contains the d efaul t-d ri ver i scsi line
to set the driver type as iSCSI. The driver uses iSCSI by default.

Important
This example creates a globally accessible target without access control. Refer to the
scsi-target-utils for information on implementing secure access.

6. R est art t h e t g t d service

Restart the tg td service to reload the configuration changes.
# service tgtd restart
7. ip t ab les co n f ig u rat io n
Open port 3260 for iSCSI access with i ptabl es.
# iptables -I INPUT -p tcp -m tcp --dport 3260 -j ACCEPT
# service iptables save
# service iptables restart
8. Verif y t h e n ew t arg et s

135

Virt ualiz at ion Administ rat ion G uide

View the new targets to ensure the setup was successful with the tg t-ad mi n --sho w
command.
# tgt-admin --show
Target 1: iqn.2010-05.com.example.server1:iscsirhel6guest
System information:
Driver: iscsi
State: ready
I_T nexus information:
LUN information:
LUN: 0
Type: controller
SCSI ID: IET
00010000
SCSI SN: beaf10
Size: 0 MB
Online: Yes
Removable media: No
Backing store type: rdwr
Backing store path: None
LUN: 1
Type: disk
SCSI ID: IET
00010001
SCSI SN: beaf11
Size: 20000 MB
Online: Yes
Removable media: No
Backing store type: rdwr
Backing store path: /dev/virtstore/virtimage1
LUN: 2
Type: disk
SCSI ID: IET
00010002
SCSI SN: beaf12
Size: 10000 MB
Online: Yes
Removable media: No
Backing store type: rdwr
Backing store path: /var/lib/tgtd/virtualization/virtimage2.img
Account information:
ACL information:
ALL

Warning
The ACL list is set to all. This allows all systems on the local network to access this
device. It is recommended to set host physical machine access ACLs for production
environments.

9. O p t io n al: T est d isco very

Test whether the new iSCSI device is discoverable.

136

Chapt er 1 2 . St orage Pools

# iscsiadm --mode discovery --type sendtargets --portal

server1.example.com
127.0.0.1:3260,1 iqn.2010-05.com.example.server1:iscsirhel6guest
10. O p t io n al: T est at t ach in g t h e d evice
Attach the new device (iqn.2010-05.com.example.server1:iscsirhel6guest) to determine whether
the device can be attached.
# iscsiadm -d2 -m node --login
scsiadm: Max file limits 1024 1024
Logging in to [iface: default, target: iqn.201005.com.example.server1:iscsirhel6guest, portal: 10.0.0.1,3260]
Login to [iface: default, target: iqn.201005.com.example.server1:iscsirhel6guest, portal: 10.0.0.1,3260]
successful.
D etach the device.
# iscsiadm -d2 -m node --logout
scsiadm: Max file limits 1024 1024
Logging out of session [sid: 2, target: iqn.201005.com.example.server1:iscsirhel6guest, portal: 10.0.0.1,3260
Logout of [sid: 2, target: iqn.201005.com.example.server1:iscsirhel6guest, portal: 10.0.0.1,3260]
successful.
An iSCSI device is now ready to use for virtualization.

12.5.2. Adding an iSCSI T arget t o virt -manager

This procedure covers creating a storage pool with an iSCSI target in vi rt-manag er.
Pro ced u re 12.5. Ad d in g an iSC SI d evice t o virt - man ag er
1. O p en t h e h o st p h ysical mach in e' s st o rag e t ab
Open the Sto rag e tab in the C o nnecti o n D etai l s window.
a. Open vi rt-manag er.
b. Select a host physical machine from the main vi rt-manag er window. Click Ed i t
menu and select C o nnecti o n D etai l s.

137

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.19 . C o n n ect io n d et ails

c. Click on the Sto rag e tab.

Fig u re 12.20. St o rag e men u

2. Ad d a n ew p o o l ( p art 1)
Press the + button (the add pool button). The Ad d a New Sto rag e P o o l wizard appears.

138

Chapt er 1 2 . St orage Pools

Fig u re 12.21. Ad d an iscsi st o rag e p o o l n ame an d t yp e

Choose a name for the storage pool, change the Type to iscsi, and press Fo rward to
continue.
3. Ad d a n ew p o o l ( p art 2)
You will need the information you used in Section 12.5, iSCSI-based Storage Pools and
Procedure 12.4, Creating an iSCSI target to complete the fields in this menu.
a. Enter the iSCSI source and target. The Fo rmat option is not available as formatting is
handled by the guest virtual machines. It is not advised to edit the T arg et P ath. The
default target path value, /d ev/d i sk/by-path/, adds the drive path to that
directory. The target path should be the same on all host physical machines for
migration.
b. Enter the hostname or IP address of the iSCSI target. This example uses
ho st1. exampl e. co m.
c. In the So urce P athfield, enter the iSCSI target IQN. If you look at Procedure 12.4,
Creating an iSCSI target in Section 12.5, iSCSI-based Storage Pools , this is the
information you added in the /etc/tg t/targ ets. co nf fi l e. This example uses
i q n. 20 10 -0 5. co m. exampl e. server1: i scsi rhel 6 g uest.
d. Check the IQ N checkbox to enter the IQN for the initiator. This example uses
i q n. 20 10 -0 5. co m. exampl e. ho st1: i scsi rhel 6 .
e. Click Fi ni sh to create the new storage pool.

139

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.22. C reat e an iscsi st o rag e p o o l

12.5.3. Delet ing a St orage Pool Using virt -manager

This procedure demonstrates how to delete a storage pool.
1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it. To do this, select the storage pool
you want to stop and click the red X icon at the bottom of the Storage window.

14 0

Chapt er 1 2 . St orage Pools

Fig u re 12.23. St o p Ico n

2. D elete the storage pool by clicking the Trash can icon. This icon is only enabled if you stop
the storage pool first.

12.5.4 . Creat ing an iSCSI-based St orage Pool wit h virsh

1. U se p o o l- d ef in e- as t o d ef in e t h e p o o l f ro m t h e co mman d lin e
Storage pool definitions can be created with the vi rsh command line tool. Creating storage
pools with vi rsh is useful for systems administrators using scripts to create multiple storage
pools.
The vi rsh po o l -d efi ne-as command has several parameters which are accepted in the
following format:
virsh pool-define-as name type source-host source-path source-dev
source-name target
The parameters are explained as follows:
t yp e
defines this pool as a particular type, iscsi for example
n ame
must be unique and sets the name for the storage pool
so u rce- h o st an d so u rce- p at h
the hostname and iSCSI IQN respectively

14 1

Virt ualiz at ion Administ rat ion G uide

so u rce- d ev an d so u rce- n ame

these parameters are not required for iSCSI-based pools, use a - character to leave
the field blank.
t arg et
defines the location for mounting the iSCSI device on the host physical machine
The example below creates the same iSCSI-based storage pool as the previous step.
#

virsh pool-define-as --name scsirhel6guest --type iscsi \

--source-host server1.example.com \
--source-dev iqn.2010-05.com.example.server1:iscsirhel6guest
--target /dev/disk/by-path
Pool iscsirhel6guest defined
2. Verif y t h e st o rag e p o o l is list ed
Verify the storage pool object is created correctly and the state reports as i nacti ve.
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
iscsirhel6guest
inactive
no
3. St art t h e st o rag e p o o l
Use the virsh command po o l -start for this. po o l -start enables a directory storage pool,
allowing it to be used for volumes and guest virtual machines.
# virsh pool-start guest_images_disk
Pool guest_images_disk started
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
iscsirhel6guest
active
no
4. T u rn o n au t o st art
Turn on autostart for the storage pool. Autostart configures the l i bvi rtd service to start
the storage pool when the service starts.
# virsh pool-autostart iscsirhel6guest
Pool iscsirhel6guest marked as autostarted
Verify that the iscsirhel6guest pool has autostart set:
# virsh pool-list --all
Name
State
Autostart
----------------------------------------default
active
yes
iscsirhel6guest
active
yes

14 2

Chapt er 1 2 . St orage Pools

5. Verif y t h e st o rag e p o o l co n f ig u rat io n

Verify the storage pool was created correctly, the sizes reported correctly, and the state
reports as runni ng .
# virsh pool-info iscsirhel6guest
Name:
iscsirhel6guest
UUID:
afcc5367-6770-e151-bcb3-847bc36c5e28
State:
running
Persistent:
unknown
Autostart:
yes
Capacity:
100.31 GB
Allocation:
0.00
Available:
100.31 GB
An iSCSI-based storage pool is now available.

12.5.5. Delet ing a St orage Pool Using virsh

The following demonstrates how to delete a storage pool using virsh:
1. To avoid any issues with other guest virtual machines using the same pool, it is best to stop
the storage pool and release any resources in use by it.
# virsh pool-destroy guest_images_disk
2. Remove the storage pool's definition
# virsh pool-undefine guest_images_disk

12.6. NFS-based St orage Pools

This procedure covers creating a storage pool with a NFS mount point in vi rt-manag er.

12.6.1. Creat ing an NFS-based St orage Pool wit h virt -manager

1. O p en t h e h o st p h ysical mach in e' s st o rag e t ab
Open the Sto rag e tab in the Ho st D etai l s window.
a. Open vi rt-manag er.
b. Select a host physical machine from the main vi rt-manag er window. Click Ed i t
menu and select C o nnecti o n D etai l s.

14 3

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.24 . C o n n ect io n d et ails

c. Click on the Storage tab.

Fig u re 12.25. St o rag e t ab

2. C reat e a n ew p o o l ( p art 1)
Press the + button (the add pool button). The Ad d a New Sto rag e P o o l wizard appears.

14 4

Chapt er 1 2 . St orage Pools

Fig u re 12.26 . Ad d an N FS n ame an d t yp e

Choose a name for the storage pool and press Fo rward to continue.
3. C reat e a n ew p o o l ( p art 2)
Enter the target path for the device, the hostname and the NFS share path. Set the Fo rmat
option to NFS or auto (to detect the type). The target path must be identical on all host
physical machines for migration.
Enter the hostname or IP address of the NFS server. This example uses
server1. exampl e. co m.
Enter the NFS path. This example uses /nfstri al .

14 5

Virt ualiz at ion Administ rat ion G uide

Fig u re 12.27. C reat e an N FS st o rag e p o o l

Press Fi ni sh to create the new storage pool.

12.6.2. Delet ing a St orage Pool Using virt -manager

This procedure demonstrates how to delete a storage pool.
1. To avoid any issues with other guests using the same pool, it is best to stop the storage pool
and release any resources in use by it. To do this, select the storage pool you want to stop
and click the red X icon at the bottom of the Storage window.

14 6

Chapt er 1 2 . St orage Pools

Fig u re 12.28. St o p Ico n

2. D elete the storage pool by clicking the Trash can icon. This icon is only enabled if you stop
the storage pool first.

12.7. Glust erFS St orage Pools

GlusterFS is a userspace file system that uses FUSE. When enabled in a guest virtual machine it
enables a KVM host physical machine to boot guest virtual machine images from one or more
GlusterFS storage volumes, and to use images from a GlusterFS storage volume as data disks for
guest virtual machines.

Important
Refer to the Red Hat Storage Administration Guide for additonal information.

12.8. Using an NPIV Virt ual Adapt er (vHBA) wit h SCSI Devices
NPIV (N_Port ID Virtualization) is a software technology that allows sharing of a single physical
Fibre Channel host bus adapter (HBA).
This allows multiple guests to see the same storage from multiple physical hosts, and thus allows for
easier migration paths for the storage. As a result, there is no need for the migration to create or copy
storage, as long as the correct storage path is specified.

14 7

Virt ualiz at ion Administ rat ion G uide

In virtualization, the virtual host bus adapter, or vHBA, controls the LUNs for virtual machines. Each
vHBA is identified by its own WWNN (World Wide Node Name) and WWPN (World Wide Port Name).
The path to the storage is determined by the WWNN and WWPN values.
This section provides instructions for configuring a vHBA on a virtual machine. Note that Red Hat
Enterprise Linux 6 does not support persistent vHBA configuration across host reboots; verify any
vHBA-related settings following a host reboot.

12.8.1. Creat ing a vHBA

Pro ced u re 12.6 . C reat in g a vH B A
1. Lo cat e H B As o n t h e h o st syst em
To locate the HBAs on your host system, examine the SCSI devices on the host system to
locate a scsi _ho st with vpo rt capability.
Run the following command to retrieve a scsi _ho st list:
# virsh nodedev-list --cap scsi_host
scsi_host0
scsi_host1
scsi_host2
scsi_host3
scsi_host4
For each scsi _ho st, run the following command to examine the device XML for the line
<capabi l i ty type= ' vpo rt_o ps' >, which indicates a scsi _ho st with vpo rt capability.
# virsh nodedev-dumpxml scsi_hostN
2. C h eck t h e H B A' s d et ails
Use the vi rsh no d ed ev-d umpxml HBA_device command to see the HBA's details.
The XML output from the vi rsh no d ed ev-d umpxml command will list the fields <name>,
<wwnn>, and <wwpn>, which are used to create a vHBA. The <max_vpo rts> value shows the
maximum number of supported vHBAs.
# virsh nodedev-dumpxml scsi_host3
<device>
<name>scsi_host3</name>
<path>/sys/devices/pci0000:00/0000:00:04.0/0000:10:00.0/host3</path
>
<parent>pci_0000_10_00_0</parent>
<capability type='scsi_host'>
<host>3</host>
<capability type='fc_host'>
<wwnn>20000000c9848140</wwnn>
<wwpn>10000000c9848140</wwpn>
<fabric_wwn>2002000573de9a81</fabric_wwn>
</capability>
<capability type='vport_ops'>
<max_vports>127</max_vports>

14 8

Chapt er 1 2 . St orage Pools

<vports>0</vports>
</capability>
</capability>
</device>
In this example, the <max_vpo rts> value shows there are a total 127 virtual ports available
for use in the HBA configuration. The <vpo rts> value shows the number of virtual ports
currently being used. These values update after creating a vHBA.
3. C reat e a vH B A h o st d evice
Create an XML file similar to the following (in this example, named vhba_host3.xml) for the
vHBA host.
# cat vhba_host3.xml
<device>
<parent>scsi_host3</parent>
<capability type='scsi_host'>
<capability type='fc_host'>
</capability>
</capability>
</device>
The <parent> field specifies the HBA device to associate with this vHBA device. The details
in the <d evi ce> tag are used in the next step to create a new vHBA device for the host. See
http://libvirt.org/formatnode.html for more information on the no d ed ev XML format.
4. C reat e a n ew vH B A o n t h e vH B A h o st d evice
To create a vHBA on vhba_host3, use the vi rsh no d ed ev-create command:
# virsh nodedev-create vhba_host3.xml
Node device scsi_host5 created from vhba_host3.xml
5. Verif y t h e vH B A
Verify the new vHBA's details (scsi _ho st5) with the vi rsh no d ed ev-d umpxml command:
# virsh nodedev-dumpxml scsi_host5
<device>
<name>scsi_host5</name>
<path>/sys/devices/pci0000:00/0000:00:04.0/0000:10:00.0/host3/vport
-3:0-0/host5</path>
<parent>scsi_host3</parent>
<capability type='scsi_host'>
<host>5</host>
<capability type='fc_host'>
<wwnn>5001a4a93526d0a1</wwnn>
<wwpn>5001a4ace3ee047d</wwpn>
<fabric_wwn>2002000573de9a81</fabric_wwn>
</capability>
</capability>
</device>

14 9

Virt ualiz at ion Administ rat ion G uide

12.8.2. Creat ing a St orage Pool Using t he vHBA

It is recommended to define a libvirt storage pool based on the vHBA in order to preserve the vHBA
configuration.
Using a storage pool has two primary advantages:
the libvirt code can easily find the LUN's path via virsh command output, and
virtual machine migration requires only defining and starting a storage pool with the same vHBA
name on the target machine. To do this, the vHBA LUN, libvirt storage pool and volume name must
be specified in the virtual machine's XML configuration. Refer to Section 12.8.3, Configuring the
Virtual Machine to Use a vHBA LUN for an example.
1. C reat e a SC SI st o rag e p o o l
To create a vHBA configuration, first create a libvirt ' scsi ' storage pool XML file based on
the vHBA using the format below.

Note
Ensure you use the vHBA created in Procedure 12.6, Creating a vHBA as the host
name, modifying the vHBA name scsi_hostN to hostN for the storage pool configuration.
In this example, the vHBA is named scsi_host5, which is specified as <ad apter
name= ' ho st5' /> in a Red Hat Enterprise Linux 6 libvirt storage pool.
It is recommended to use a stable location for the <path> value, such as one of the
/d ev/d i sk/by-{path| i d | uui d | l abel } locations on your system. More information on
<path> and the elements within <targ et> can be found at
http://libvirt.org/formatstorage.html.
In this example, the ' scsi ' storage pool is named vhbapool_host3.xml:
<pool type='scsi'>
<name>vhbapool_host3</name>
<uuid>e9392370-2917-565e-692b-d057f46512d6</uuid>
<capacity unit='bytes'>0</capacity>
<allocation unit='bytes'>0</allocation>
<available unit='bytes'>0</available>
<source>
<adapter name='host5'/>
</source>
<target>
<path>/dev/disk/by-path</path>
<permissions>
<mode>0700</mode>
<owner>0</owner>
<group>0</group>
</permissions>
</target>
</pool>
2. D ef in e t h e p o o l

150

Chapt er 1 2 . St orage Pools

To define the storage pool (named vhbapool_host3 in this example), use the vi rsh po o l d efi ne command:
# virsh pool-define vhbapool_host3.xml
Pool vhbapool_host3 defined from vhbapool_host3.xml
3. St art t h e p o o l
Start the storage pool with the following command:
# virsh pool-start vhbapool_host3
Pool vhbapool_host3 started
4. En ab le au t o st art
Finally, to ensure that subsequent host reboots will automatically define vHBAs for use in
virtual machines, set the storage pool autostart feature (in this example, for a pool named
vhbapool_host3):
# virsh pool-autostart vhbapool_host3

12.8.3. Configuring t he Virt ual Machine t o Use a vHBA LUN

After a storage pool is created for a vHBA, add the vHBA LUN to the virtual machine configuration.
1. Fin d availab le LU N s
First, use the vi rsh vo l -l i st command in order to generate a list of available LUNs on
the vHBA. For example:
# virsh vol-list vhbapool_host3
Name
Path
----------------------------------------------------------------------------unit:0:4:0
/dev/disk/by-path/pci-0000:10:00.0-fc0x5006016844602198-lun-0
unit:0:5:0
/dev/disk/by-path/pci-0000:10:00.0-fc0x5006016044602198-lun-0
The list of LUN names displayed will be available for use as disk volumes in virtual machine
configurations.
2. Ad d t h e vH B A LU N t o t h e virt u al mach in e
Add the vHBA LUN to the virtual machine by specifying in the virtual machine's XML:
the device type as l un or d i sk in the <d i sk> parameter, and
the source device in the <so urce> parameter. Note this can be entered as /d ev/sd aN, or
as a symbolic link generated by udev in /d ev/d i sk/by-path| by-i d | by-uui d | byl abel , which can be found by running the vi rsh vo l -l i st pool command.
For example:

151

Virt ualiz at ion Administ rat ion G uide

<disk type='block' device='lun'>

<driver name='qemu' type='raw'/>
<source dev='/dev/disk/by-path/pci-0000\:04\:00.1-fc0x203400a0b85ad1d7-lun-0'/>
<target dev='sda' bus='scsi'/>
</disk>

12.8.4 . Dest roying t he vHBA St orage Pool

A vHBA storage pool can be destroyed by the vi rsh po o l -d estro y command:
# virsh pool-destroy vhbapool_host3
D elete the vHBA with the following command
# virsh nodedev-destroy scsi_host5
To verify the pool and vHBA have been destroyed, run:
# virsh nodedev-list --cap scsi_host
scsi _ho st5 will no longer appear in the list of results.

152

Chapt er 1 3. Volumes

Chapter 13. Volumes

13.1. Creat ing Volumes
This section shows how to create disk volumes inside a block based storage pool. In the example
below, the vi rsh vo l -create-as command will create a storage volume with a specific size in GB
within the guest_images_disk storage pool. As this command is repeated per volume needed, three
volumes are created as shown in the example.
# virsh vol-create-as guest_images_disk volume1 8G
Vol volume1 created
# virsh vol-create-as guest_images_disk volume2 8G
Vol volume2 created
# virsh vol-create-as guest_images_disk volume3 8G
Vol volume3 created
# virsh vol-list guest_images_disk
Name
Path
----------------------------------------volume1
/dev/sdb1
volume2
/dev/sdb2
volume3
/dev/sdb3
# parted -s /dev/sdb pri nt
Model: ATA ST3500418AS (scsi)
Disk /dev/sdb: 500GB
Sector size (logical/physical): 512B/512B
Partition Table: gpt
Number Start
2
17.4kB
3
8590MB
1
21.5GB

End
8590MB
17.2GB
30.1GB

Size
8590MB
8590MB
8590MB

File system

Name
primary
primary
primary

Flags

13.2. Cloning Volumes

The new volume will be allocated from storage in the same storage pool as the volume being cloned.
The vi rsh vo l -cl o ne must have the --po o l argument which dictates the name of the storage
pool that contains the volume to be cloned. The rest of the command names the volume to be cloned
(volume3) and the name of the new volume that was cloned (clone1). The vi rsh vo l -l i st
command lists the volumes that are present in the storage pool (guest_images_disk).
# virsh vol-clone --pool guest_images_disk volume3 clone1
Vol clone1 cloned from volume3
# vi rsh vo l -l i st guest_images_disk
Name
Path
----------------------------------------volume1
/dev/sdb1
volume2
/dev/sdb2

153

Virt ualiz at ion Administ rat ion G uide

volume3
clone1

/dev/sdb3
/dev/sdb4

# parted -s /dev/sdb pri nt

Model: ATA ST3500418AS (scsi)
Disk /dev/sdb: 500GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Number
1
2
3
4

Start
4211MB
12.8GB
21.4GB
30.0GB

End
Size
File system
12.8GB 8595MB primary
21.4GB 8595MB primary
30.0GB 8595MB primary
38.6GB 8595MB primary

Name

Flags

13.3. Adding St orage Devices t o Guest s

This section covers adding storage devices to a guest. Additional storage can only be added as
needed.

13.3.1. Adding File-based St orage t o a Guest

File-based storage is a collection of files that are stored on the host physical machines file system
that act as virtualized hard drives for guests. To add file-based storage, perform the following steps:
Pro ced u re 13.1. Ad d in g f ile- b ased st o rag e
1. Create a storage file or use an existing file (such as an IMG file). Note that both of the
following commands create a 4GB file which can be used as additional storage for a guest:
Pre-allocated files are recommended for file-based storage images. Create a pre-allocated
file using the following d d command as shown:
# dd if=/dev/zero of=/var/lib/libvirt/images/FileName.img bs=1M
count=4096
Alternatively, create a sparse file instead of a pre-allocated file. Sparse files are created
much faster and can be used for testing, but are not recommended for production
environments due to data integrity and performance issues.
# dd if=/dev/zero of=/var/lib/libvirt/images/FileName.img bs=1M
seek=4096 count=0
2. Create the additional storage by writing a <disk> element in a new file. In this example, this file
will be known as NewSto rag e. xml .
A <d i sk> element describes the source of the disk, and a device name for the virtual block
device. The device name should be unique across all devices in the guest, and identifies the
bus on which the guest will find the virtual block device. The following example defines a
virtio block device whose source is a file-based storage container named Fi l eName. i mg :
<disk type='file' device='disk'>

154

Chapt er 1 3. Volumes

<driver name='qemu' type='raw' cache='none'/>

<source file='/var/lib/libvirt/images/FileName.img'/>
<target dev='vdb'/>
</disk>
D evice names can also start with " hd" or " sd" , identifying respectively an ID E and a SCSI
disk. The configuration file can also contain an <ad d ress> sub-element that specifies the
position on the bus for the new device. In the case of virtio block devices, this should be a
PCI address. Omitting the <ad d ress> sub-element lets libvirt locate and assign the next
available PCI slot.
3. Attach the CD -ROM as follows:
<disk type='file' device='cdrom'>
<driver name='qemu' type='raw' cache='none'/>
<source file='/var/lib/libvirt/images/FileName.img'/>
<readonly/>
<target dev='hdc'/>
</disk >
4. Add the device defined in NewSto rag e. xml with your guest (G uest1):
# virsh attach-device --config Guest1 ~/NewStorage.xml

Note
This change will only apply after the guest has been destroyed and restarted. In
addition, persistent devices can only be added to a persistent domain, that is a
domain whose configuration has been saved with vi rsh d efi ne command.
If the guest is running, and you want the new device to be added temporarily until the guest is
destroyed, omit the --co nfi g option:
# virsh attach-device Guest1 ~/NewStorage.xml

Note
The vi rsh command allows for an attach-d i sk command that can set a limited
number of parameters with a simpler syntax and without the need to create an XML file.
The attach-d i sk command is used in a similar manner to the attach-d evi ce
command mentioned previously, as shown:
# virsh attach-disk Guest1
/var/lib/libvirt/images/FileName.img vdb --cache none
Note that the vi rsh attach-d i sk command also accepts the --co nfi g option.

5. Start the guest machine (if it is currently not running):

155

Virt ualiz at ion Administ rat ion G uide

# virsh start Guest1

Note
The following steps are Linux guest specific. Other operating systems handle new
storage devices in different ways. For other systems, refer to that operating system's
documentation.

6. Part it io n in g t h e d isk d rive

The guest now has a hard disk device called /d ev/vd b. If required, partition this disk drive
and format the partitions. If you do not see the device that you added, then it indicates that
there is an issue with the disk hotplug in your guest's operating system.
a. Start fd i sk for the new device:
# fdisk /dev/vdb
Command (m for help):
b. Type n for a new partition.
c. The following appears:
Command action
e
extended
p
primary partition (1-4)
Type p for a primary partition.
d. Choose an available partition number. In this example, the first partition is chosen by
entering 1.
Partition number (1-4): 1
e. Enter the default first cylinder by pressing Enter.
First cylinder (1-400, default 1):
f. Select the size of the partition. In this example the entire disk is allocated by pressing
Enter.
Last cylinder or +size or +sizeM or +sizeK (2-400, default
400):
g. Enter t to configure the partition type.
Command (m for help): t
h. Select the partition you created in the previous steps. In this example, the partition
number is 1 as there was only one partition created and fdisk automatically selected
partition 1.

156

Chapt er 1 3. Volumes

Partition number (1-4): 1

i. Enter 83 for a Linux partition.
Hex code (type L to list codes): 83
j. Enter w to write changes and quit.
Command (m for help): w
k. Format the new partition with the ext3 file system.
# mke2fs -j /dev/vdb1
7. Create a mount directory, and mount the disk on the guest. In this example, the directory is
located in myfiles.
# mkdir /myfiles
# mount /dev/vdb1 /myfiles
The guest now has an additional virtualized file-based storage device. Note however, that
this storage will not mount persistently across reboot unless defined in the guest's
/etc/fstab file:
/dev/vdb1

/myfiles

ext3

defaults

0 0

13.3.2. Adding Hard Drives and Ot her Block Devices t o a Guest

System administrators have the option to use additional hard drives to provide increased storage
space for a guest, or to separate system data from user data.
Pro ced u re 13.2. Ad d in g p h ysical b lo ck d evices t o g u est s
1. This procedure describes how to add a hard drive on the host physical machine to a guest. It
applies to all physical block devices, including CD -ROM, D VD and floppy devices.
Physically attach the hard disk device to the host physical machine. Configure the host
physical machine if the drive is not accessible by default.
2. D o one of the following:
a. Create the additional storage by writing a d i sk element in a new file. In this example,
this file will be known as NewSto rag e. xml . The following example is a configuration
file section which contains an additional device-based storage container for the host
physical machine partition /d ev/sr0 :
<disk type='block' device='disk'>
<driver name='qemu' type='raw' cache='none'/>
<source dev='/dev/sr0'/>
<target dev='vdc' bus='virtio'/>
</disk>

157

Virt ualiz at ion Administ rat ion G uide

b. Follow the instruction in the previous section to attach the device to the guest virtual
machine. Alternatively, you can use the virsh attach-disk command, as shown:
# virsh attach-disk Guest1 /dev/sr0 vdc
Note that the following options are available:
The vi rsh attach-d i sk command also accepts the --config, --type, and -mode options, as shown:
# vi rsh attach-d i sk G uest1 /d ev/sr0 vd c --co nfi g --type
cd ro m --mo d e read o nl y
Additionally, --type also accepts --type disk in cases where the device is a
hard drive.
3. The guest virtual machine now has a new hard disk device called /d ev/vd c on Linux (or
something similar, depending on what the guest virtual machine OS chooses) or D : d ri ve
(for example) on Windows. You can now initialize the disk from the guest virtual machine,
following the standard procedures for the guest virtual machine's operating system. Refer to
Procedure 13.1, Adding file-based storage and Procedure 13.1, Adding file-based storage
for an example.

Warning
The host physical machine should not use filesystem labels to identify file systems in
the fstab file, the i ni trd file or on the kernel command line. D oing so presents a
security risk if less privileged users, such as guest virtual machines, have write access
to whole partitions or LVM volumes, because a guest virtual machine could potentially
write a filesystem label belonging to the host physical machine, to its own block device
storage. Upon reboot of the host physical machine, the host physical machine could
then mistakenly use the guest virtual machine's disk as a system disk, which would
compromise the host physical machine system.
It is preferable to use the UUID of a device to identify it in the fstab file, the i ni trd file
or on the kernel command line. While using UUID s is still not completely secure on
certain file systems, a similar compromise with UUID is significantly less feasible.

Important
Guest virtual machines should not be given write access to whole disks or block
devices (for example, /d ev/sd b). Guest virtual machines with access to whole block
devices may be able to modify volume labels, which can be used to compromise the
host physical machine system. Use partitions (for example, /d ev/sd b1) or LVM
volumes to prevent this issue.

13.4 . Delet ing and Removing Volumes

This section shows how to delete a disk volume from a block based storage pool using the vi rsh
vo l -d el ete command. In this example, the volume is volume 1 and the storage pool is

158

Chapt er 1 3. Volumes

guest_images.
# virsh vol-delete --pool guest_images volume1
Vol volume1 deleted

159

Virt ualiz at ion Administ rat ion G uide

Chapter 14. Managing guest virtual machines with virsh

vi rsh is a command line interface tool for managing guest virtual machines and the hypervisor. The
vi rsh command-line tool is built on the l i bvi rt management API and operates as an alternative
to the q emu-kvm command and the graphical vi rt-manag er application. The vi rsh command
can be used in read-only mode by unprivileged users or, with root access, full administration
functionality. The vi rsh command is ideal for scripting virtualization administration.

14 .1. Generic Commands

The commands in this section are generic because they are not specific to any domain.

14 .1.1. help
$ vi rsh hel p [co mmand | g ro up] The help command can be used with or without options. When
used without options, all commands are listed, one per line. When used with an option, it is grouped
into categories, displaying the keyword for each group.
To display the commands that are only for a specific option, you need to give the keyword for that
group as an option. For example:
$ vi rsh hel p po o l
Storage Pool (help keyword 'pool'):
find-storage-pool-sources-as
find potential storage pool sources
find-storage-pool-sources
discover potential storage pool
sources
pool-autostart
autostart a pool
pool-build
build a pool
pool-create-as
create a pool from a set of args
pool-create
create a pool from an XML file
pool-define-as
define a pool from a set of args
pool-define
define (but don't start) a pool from
an XML file
pool-delete
delete a pool
pool-destroy
destroy (stop) a pool
pool-dumpxml
pool information in XML
pool-edit
edit XML configuration for a storage
pool
pool-info
storage pool information
pool-list
list pools
pool-name
convert a pool UUID to pool name
pool-refresh
refresh a pool
pool-start
start a (previously defined) inactive
pool
pool-undefine
undefine an inactive pool
pool-uuid
convert a pool name to pool UUID
Using the same command with a command option, gives the help information on that one specific
command. For example:
$ vi rsh hel p vo l -path
NAME
vol-path - returns the volume path for a given volume name or key

160

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

SYNOPSIS
vol-path <vol> [--pool <string>]
OPTIONS
[--vol] <string> volume name or key
--pool <string> pool name or uuid

14 .1.2. quit and exit

The quit command and the exit command will close the terminal. For example:
$ vi rsh exi t
$ vi rsh q ui t

14 .1.3. version
The version command displays the current libvirt version and displays information about where the
build is from. For example:
$ vi rsh versi o n
Compiled against library: libvirt 1.1.1
Using library: libvirt 1.1.1
Using API: QEMU 1.1.1
Running hypervisor: QEMU 1.5.3

14 .1.4 . Argument Display

The vi rsh echo [--shel l ][--xml ][arg ] command echos or displays the specified argument.
Each argument echoed will be separated by a space. by using the --shel l option, the output will
be single quoted where needed so that it is suitable for reusing in a shell command. If the --xml
option is used the output will be made suitable for use in an XML file. For example, the command
vi rsh echo --shel l "hel l o wo rl d " will send the output ' hel l o wo rl d ' .

14 .1.5. connect
Connects to a hypervisor session. When the shell is first started this command runs automatically
when the URI parameter is requested by the -c command. The URI specifies how to connect to the
hypervisor. The most commonly used URIs are:
xen: /// - connects to the local Xen hypervisor.
q emu: ///system - connects locally as root to the daemon supervising QEMU and KVM
domains.
xen: ///sessi o n - connects locally as a user to the user's set of QEMU and KVM domains.
l xc: /// - connects to a local Linux container.
Additional values are available on libvirt's website http://libvirt.org/uri.html.
The command can be run as follows:

161

Virt ualiz at ion Administ rat ion G uide

$ vi rsh co nnect {name|URI}

Where {name} is the machine name (hostname) or URL (the output of the vi rsh uri command) of
the hypervisor. To initiate a read-only connection, append the above command with --read o nl y.
For more information on URIs refer to Remote URIs. If you are unsure of the URI, the vi rsh uri
command will display it:
$ vi rsh uri
qemu:///session

14 .1.6. Displaying Basic Informat ion

The following commands may be used to display basic information:
$ ho stname - displays the hypervisor's hostname
$ sysi nfo - displays the XML representation of the hypervisor's system information, if available

14 .1.7. Inject ing NMI

The $ vi rsh i nject-nmi [d o mai n] injects NMI (non-maskable interrupt) message to the guest
virtual machine. This is used when response time is critical, such as non-recoverable hardware
errors. To run this command:
$ virsh inject-nmi guest-1

14 .2. At t aching and Updat ing a Device wit h virsh

For information on attaching storage devices refer to Section 13.3.1, Adding File-based Storage to a
Guest
Pro ced u re 14 .1. H o t p lu g g in g U SB d evices f o r u se b y t h e g u est virt u al mach in e
The following procedure demonstrates how to attach USB devices to the guest virtual machine. This
can be done while the guest virtual machine is running as a hotplug procedure or it can be done
while the guest is shutoff. The device you want to emulate needs to be attached to the host physical
machine.
1. Locate the USB device you want to attach with the following command:
# lsusb -v
idVendor
idProduct

0x17ef Lenovo
0x480f Integrated Webcam [R5U877]

2. Create an XML file and give it a logical name (usb_d evi ce. xml , for example). Make sure
you copy the vendor and procuct ID s exactly as was displayed in your search.

<hostdev mode='subsystem' type='usb' managed='yes'>

162

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

<source>
<vendor id='0x17ef'/>
<product id='0x480f'/>
</source>
</hostdev>
...

Fig u re 14 .1. U SB D evices XML Sn ip p et

3. Attach the device with the following command:
# virsh attach-device rhel6 --fi l e usb_d evi ce. xml > --co nfi g
In this example [rhel6] is the name of your guest virtual machine and [usb_device.xml] is the
file you created in the previous step. If you want to have the change take effect in the next
reboot, use the --co nfi g option. If you want this change to be persistent, use the -persi stent option. If you want the change to take effect on the current domain, use the -current option. See the Virsh man page for additional information.
4. If you want to detach the device (hot unplug), perform the following command:
# virsh detach-device rhel6 --fi l e usb_d evi ce. xml >
In this example [rhel6] is the name of your guest virtual machine and [usb_device.xml] is the
file you attached in the previous step

14 .3. At t aching Int erface Devices

The vi rsh attach-i nterfacedomain type source command can take the following options:
--l i ve - get value from running domain
--co nfi g - get value to be used on next boot
--current - get value according to current domain state
--persi stent - behaves like --co nfi g for an offline domain, and like --l i ve for a running
domain.
--targ et - indicates the target device in the guest virtual machine.
--mac - use this to specify the MAC address of the network interface
--scri pt - use this to specify a path to a script file handling a bridge instead of the default one.
--mo d el - use this to specify the model type.
--i nbo und - controls the inbound bandwidth of the interface. Acceptable values are averag e,
peak, and burst.
--o utbo und - controls the outbound bandwidth of the interface. Acceptable values are
averag e, peak, and burst.

163

Virt ualiz at ion Administ rat ion G uide

The type can be either netwo rk to indicate a physical network device, or bri d g e to indicate a
bridge to a device. source is the source of the device. To remove the attached device, use the vi rsh
d etach-d evi ce.

14 .4 . Changing t he Media of a CDROM

Changing the media of a CD ROM to another source or format
# chang e-med i a domain path source --eject --i nsert --upd ate --current -l i ve --co nfi g --fo rce
--path - A string containing a fully-qualified path or target of disk device
--so urce - A string containing the source of the media
--eject - Eject the media
--i nsert - Insert the media
--upd ate - Update the media
--current - can be either or both of --l i ve and --co nfi g , depends on implementation of
hypervisor driver
--l i ve - alter live configuration of running domain
--co nfi g - alter persistent configuration, effect observed on next boot
--fo rce - force media changing

14 .5. Domain Commands

A domain name is required for most of these commands as they manipulate the specified domain
directly. The domain may be given as a short integer (0,1,2...), a name, or a full UUID .

14 .5.1. Configuring a Domain t o be St art ed Aut omat ically at Boot

$ vi rsh auto start [--d i sabl e] d o mai n will automatically start the specified domain at boot.
Using the --d i sabl e option disables autostart.
# vi rsh auto start rhel6
In the example above, the rhel6 guest virtual machine will automatically start when the host physical
machine boots
# vi rsh auto start rhel6 --d i sabl e
In the example above, the autostart function is disabled and the guest virtual machine will no longer
start automatically when the host physical machine boots.

14 .5.2. Connect ing t he Serial Console for t he Guest Virt ual Machine
The $ vi rsh co nso l e <d o mai n> [--d evname <stri ng >] [--fo rce] [--safe]
command connects the virtual serial console for the guest virtual machine. The optional --devname

164

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

<string> parameter refers to the device alias of an alternate console, serial, or parallel device
configured for the guest virtual machine. If this parameter is omitted, the primary console will be
opened. The --fo rce option will force the console connection or when used with disconnect, will
disconnect connections. Using the --safe option will only allow the guest to connect if safe console
handling is supported.
$ virsh console virtual_machine --safe

14 .5.3. Defining a Domain wit h an XML File

The d efi ne <FILE> command defines a domain from an XML file. The domain definition in this
case is registered but not started. If the domain is already running, the changes will take effect on the
next boot.

14 .5.4 . Edit ing and Displaying a Descript ion and T it le of a Domain

The following command is used to show or modify the description and title of a domain, but does not
configure it:
# vi rsh d esc [domain-name] [[--l i ve] [--co nfi g ] | [--current]] [--ti tl e]
[--ed i t] [--new-d esc New description or title message]
These values are user fields that allow storage of arbitrary textual data to allow easy identification of
domains. Ideally, the title should be short, although this is not enforced by libvirt.
The options --l i ve or --co nfi g select whether this command works on live or persistent
definitions of the domain. If both --l i ve and --co nfi g are specified, the --co nfi g option will be
implemented first, where the description entered in the command becomes the new configuration
setting which is applied to both the live configuration and persistent configuration setting. The -current option will modify or get the current state configuration and will not be persistent. The -current option will be used if neither --l i ve nor --co nfi g , nor --current are specified. The -ed i t option specifies that an editor with the contents of current description or title should be opened
and the contents saved back afterwards. Using the --ti tl e option will show or modify the domain's
title field only and not include its description. In addition, if neither --ed i t nor --new-d esc are
used in the command, then only the description is displayed and cannot be modified.
For example, the following command changes the guest virtual machine's title from testvm to TestVM4F and will change the description to Guest VM on fourth floor:
$ vi rsh d esc testvm --current --ti tl e TestVM-4F --new-d esc Guest VM on
fourth floor

14 .5.5. Displaying Device Block St at ist ics

This command will display the block statistics for a running domain. You need to have both the
domain name and the device name (use the vi rsh d o mbl kl i st to list the devices.)In this case a
block device is the unique target name (<target dev='name'/>) or a source file (< source file
='name'/>). Note that not every hypervisor can display every field. To make sure that the output is
presented in its most legible form use the --human option, as shown:
# virsh domblklist rhel6
Target
Source
------------------------------------------------

165

Virt ualiz at ion Administ rat ion G uide

vda
hdc

/VirtualMachines/rhel6.img
-

# virsh domblkstat --human rhel6

Device: vda
number of read operations:
number of bytes read:
number of write operations:
number of bytes written:
number of flush operations:
total duration of reads (ns):
total duration of writes (ns):
total duration of flushes (ns):

vda
174670
3219440128
23897
164849664
11577
1005410244506
1085306686457
340645193294

14 .5.6. Ret rieving Net work St at ist ics

The d o mnetstat [d o mai n][i nterface-d evi ce] command displays the network interface
statistics for the specified device running on a given domain.
# domifstat rhel6 eth0

14 .5.7. Modifying t he Link St at e of a Domain's Virt ual Int erface

The following command can either configure a specified interface as up or down:
# d o mi f-setl i nk [domain][interface-device][state]{--co nfi g }
Using this modifies the status of the specified interface for the specified domain. Note that if you only
want the persistent configuration of the domain to be modified, you need to use the -co nfi g option. It should also be noted that for compatibility reasons, --persi stent is an alias of -co nfi g . The " interface device" can be the interface's target name or the MAC address.
# d o mi f-setl i nk rhel 6 eth0 up

14 .5.8. List ing t he Link St at e of a Domain's Virt ual Int erface

This command can be used to query the state of a specified interface on a given domain. Note that if
you only want the persistent configuration of the domain to be modified, you need to use the -co nfi g option. It should also be noted that for compatibility reasons, --persi stent is an alias of -co nfi g . The " interface device" can be the interface's target name or the MAC address.
# d o mi f-g etl i nk rhel 6 eth0 up

14 .5.9. Set t ing Net work Int erface Bandwidt h Paramet ers
d o mi ftune sets the guest virtual machine's network interface bandwidth parameters. The following
format should be used:
#vi rsh d o mi ftune d o mai n i nterface-d evi ce [[--co nfi g ] [--l i ve] | [-current]] [--i nbo und averag e,peak,burst] [--o utbo und averag e,peak,burst]

166

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

The only required parameter is the domain name and interface device of the guest virtual machine,
the --co nfi g , --l i ve, and --current functions the same as in Section 14.19, Setting Schedule
Parameters . If no limit is specified, it will query current network interface setting. Otherwise, alter the
limits with the following options:
<interface-device> This is mandatory and it will set or query the domains network interfaces
bandwidth parameters. i nterface-d evi ce can be the interfaces target name (<target
dev=name/>), or the MAC address.
If no --i nbo und or --o utbo und is specified, this command will query and show the bandwidth
settings. Otherwise, it will set the inbound or outbound bandwidth. average,peak,burst is the same
as in attach-i nterface command. Refer to Section 14.3, Attaching Interface D evices

14 .5.10. Ret rieving Memory St at ist ics for a Running Domain

This command may return varied results depending on the hypervisor you are using.
The d o mmemstat [d o mai n] [--peri o d (sec)][[--co nfi g ][--l i ve]| [--current]]
displays the memory statistics for a running domain. Using the --peri o d option requires a time
period in seconds. Setting this option to a value larger than 0 will allow the balloon driver to return
additional statistics which will be displayed by subsequent d o memstat commands. Setting the -peri o d option to 0, will stop the balloon driver collection but does not clear the statistics in the
balloon driver. You cannot use the --l i ve, --co nfi g , or --current options without also setting
the --peri o d option in order to also set the collection period for the balloon driver. If the --l i ve
option is specified, only the running guest's collection period is affected. If the --co nfi g option is
used, it will affect the next boot of a persistent guest. If --current option is used, it will affect the
current guest state
Both the --l i ve and --co nfi g options may be used but --current is exclusive. If no option is
specified, the behavior will be different depending on the guest's state.
#vi rsh d o memstat rhel6 --current

14 .5.11. Displaying Errors on Block Devices

This command is best used following a d o mstate that reports that a domain is paused due to an I/O
error. The d o mbl kerro r domain command shows all block devices that are in error state on a
given domain and it displays the error message that the device is reporting.
# vi rsh d o mbl kerro r rhel6

14 .5.12. Displaying t he Block Device Siz e

In this case a block device is the unique target name (<target dev='name'/>) or a source file (< source
file ='name'/>). To retrieve a list you can run d o mbl kl i st. This d o mbl ki nfo requires a domain
name.
# vi rsh d o mbl ki nfo rhel6

14 .5.13. Displaying t he Block Devices Associat ed wit h a Domain

The d o mbl kl i st domain --i nacti ve --d etai l s displays a table of all block devices that are
associated with the specified domain.

167

Virt ualiz at ion Administ rat ion G uide

If --i nacti ve is specified, the result will show the devices that are to be used at the next boot and
will not show those that are currently running in use by the running domain. If --d etai l s is
specified, the disk type and device value will be included in the table. The information displayed in
this table can be used with the d o mbl ki nfo and snapsho t-create.
#d o mbl kl i st rhel6 --d etai l s

14 .5.14 . Displaying Virt ual Int erfaces Associat ed wit h a Domain

Running the d o mi fl i st command results in a table that displays information of all the virtual
interfaces that are associated with a specified domain. The d o mi fl i st requires a domain name and
optionally can take the --i nacti ve option.
If --i nacti ve is specified, the result will show the devices that are to be used at the next boot and
will not show those that are currently running in use by the running domain.
Commands that require a MAC address of a virtual interface (such as d etach-i nterface or
d o mi f-setl i nk) will accept the output displayed by this command.

14 .5.15. Using blockcommit t o Short en a Backing Chain

This section demonstrates how to use vi rsh bl o ckco mmi t to shorten a backing chain. For more
background on backing chains, see Section 14.5.18, D isk Image Management with Live Block
Copy .
bl o ckco mmi t copies data from one part of the chain down into a backing file, allowing you to pivot
the rest of the chain in order to bypass the committed portions. For example, suppose this is the
current state:
base snap1 snap2 acti ve.
Using bl o ckco mmi t moves the contents of snap2 into snap1, allowing you to delete snap2 from the
chain, making backups much quicker.
Pro ced u re 14 .2. virsh b lo ckco mmit
Run the following command:
# vi rsh bl o ckco mmi t $d o m $d i sk -base snap1 -to p snap2 -wai t -verbo se
The contents of snap2 are moved into snap1, resulting in:
base snap1 acti ve. Snap2 is no longer valid and can be deleted

Warning
bl o ckco mmi t will corrupt any file that depends on the -base option (other than files that
depend on the -top option, as those files now point to the base). To prevent this, do not
commit changes into files shared by more than one guest. The -verbose option allows the
progress to be printed on the screen.

14 .5.16. Using blockpull t o Short en a Backing Chain

168

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

bl o ckpul l can be used in in the following applications:

Flattens an image by populating it with data from its backing image chain. This makes the image
file self-contained so that it no longer depends on backing images and looks like this:
Before: base.img Active
After: base.img is no longer used by the guest and Active contains all of the data.
Flattens part of the backing image chain. This can be used to flatten snapshots into the top-level
image and looks like this:
Before: base sn1 sn2 active
After: base.img active. Note that active now contains all data from sn1 and sn2 and neither
sn1 nor sn2 are used by the guest.
Moves the disk image to a new file system on the host. This is allows image files to be moved while
the guest is running and looks like this:
Before (The original image file): /fs1/base. vm. i mg
After: /fs2/acti ve. vm. q co w2 is now the new file system and /fs1/base. vm. i mg is no
longer used.
Useful in live migration with post-copy storage migration. The disk image is copied from the
source host to the destination host after live migration completes.
In short this is what happens: Before:/so urce-ho st/base. vm. i mg After:/d esti nati o nho st/acti ve. vm. q co w2./so urce-ho st/base. vm. i mg is no longer used.
Pro ced u re 14 .3. U sin g b lo ckp u ll t o Sh o rt en a B ackin g C h ain
1. It may be helpful to run this command prior to running bl o ckpul l :
# vi rsh snapsho t-create-as $d o m $name - d i sk-o nl y
2. If the chain looks like this: base snap1 snap2 acti ve run the following:
# vi rsh bl o ckpul l $d o m $d i sk snap1
This command makes 'snap1' the backing file of active, by pulling data from snap2 into
active resulting in: base snap1 active.
3. Once the bl o ckpul l is complete, the lib virt tracking of the snapshot that created the extra
image in the chain is no longer useful. D elete the tracking on the outdated snapshot with this
command:
# vi rsh snapsho t-d el ete $d o m $name - metad ata
Additional applications of bl o ckpul l can be done as follows:
To flatten a single image and populate it with data from its backing image chain:# vi rsh
bl o ckpul l example-domain vd a - wai t
To flatten part of the backing image chain:# vi rsh bl o ckpul l example-domain vd a base /path/to /base. i mg - wai t

169

Virt ualiz at ion Administ rat ion G uide

To move the disk image to a new file system on the host:# vi rsh snapsho t-create
example-domaine - xml fi l e /path/to /new. xml - d i sk-o nl y followed by # vi rsh
bl o ckpul l example-domain vd a - wai t
To use live migration with post-copy storage migration:
On the destination run:
# q emu-i mg create -f q co w2 -o backi ng _fi l e= /so urce-ho st/vm. i mg
/d esti nati o n-ho st/vm. q co w2
On the source run:
# vi rsh mi g rate example-domain
On the destination run:
# vi rsh bl o ckpul l example-domain vd a - wai t

14 .5.17. Using blockresiz e t o Change t he Siz e of a Domain Pat h

bl o ckresi ze can be used to re-size a block device of a domain while the domain is running, using
the absolute path of the block device which also corresponds to a unique target name (<targ et
d ev= "name"/>) or source file (<so urce fi l e= "name"/>). This can be applied to one of the disk
devices attached to domain (you can use the command d o mbl kl i st to print a table showing the
brief information of all block devices associated with a given domain).

Note
Live image re-sizing will always re-size the image, but may not immediately be picked up by
guests. With recent guest kernels, the size of virtio-blk devices is automatically updated (older
kernels require a guest reboot). With SCSI devices, it is required to manually trigger a re-scan
in the guest with the command, echo >
/sys/cl ass/scsi _d evi ce/0 : 0 : 0 : 0 /d evi ce/rescan. In addition, with ID E it is
required to reboot the guest before it picks up the new size.
Run the following command: bl o ckresi ze [d o mai n] [path si ze] where:
D omain is the unique target name or source file of the domain whose size you want to change
Path size is a scaled integer which defaults to KiB (blocks of 1024 bytes) if there is no suffix.
You must use a suffix of " B" to for bytes.

14 .5.18. Disk Image Management wit h Live Block Copy

170

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

Note
Live block copy is a feature that is not supported with the version of KVM that is supplied with
Red Hat Enterprise Linux. Live block copy is available with the version of KVM that is supplied
with Red Hat Virtualization. This version of KVM must be running on your physical host
machine in order for the feature to be supported. Contact your representative at Red Hat for
more details.
Live block copy allows you to copy an in use guest disk image to a destination image and switches
the guest disk image to the destination guest image while the guest is running. Whilst live migration
moves the memory and registry state of the host, the guest is kept in shared storage. Live block copy
allows you to move the entire guest contents to another host on the fly while the guest is running. Live
block copy may also be used for live migration without requiring permanent share storage. In this
method the disk image is copied to the destination host after migration, but while the guest is
running.
Live block copy is especially useful for the following applications:
moving the guest image from local storage to a central location
when maintenance is required, guests can be transferred to another location, with no loss of
performance
allows for management of guest images for speed and efficiency
image format conversions can be done without having to shut down the guest

Examp le 14 .1. Examp le u sin g live b lo ck co p y

This example shows what happens when live block copy is performed. The example has a
backing file (base) that is shared between a source and destination. It also has two overlays (sn1
and sn2) that are only present on the source and must be copied.
1. The backing file chain at the beginning looks like this:
base sn1 sn2
The components are as follows:
base - the original disk image
sn1 - the first snapshot that was taken of the base disk image
sn2 - the most current snapshot
active - the copy of the disk
2. When a copy of the image is created as a new image on top of sn2 the result is this:
base sn1 sn2 acti ve
3. At this point the read permissions are all in the correct order and are set automatically. To
make sure write permissions are set properly, a mirror mechanism redirects all writes to
both sn2 and active, so that sn2 and active read the same at any time (and this mirror
mechanism is the essential difference between live block copy and image streaming).

171

Virt ualiz at ion Administ rat ion G uide

4. A background task that loops over all disk clusters is executed. For each cluster, there are
the following possible cases and actions:
The cluster is already allocated in active and there is nothing to do.
Use bd rv_i s_al l o cated () to follow the backing file chain. If the cluster is read from
base (which is shared) there is nothing to do.
If bd rv_i s_al l o cated () variant is not feasible, rebase the image and compare the
read data with write data in base in order to decide if a copy is needed.
In all other cases, copy the cluster into acti ve
5. When the copy has completed, the backing file of active is switched to base (similar to
rebase)
To reduce the length of a backing chain after a series of snapshots, the following commands are
helpful: bl o ckco mmi t and bl o ckpul l . See Section 14.5.15, Using blockcommit to Shorten a
Backing Chain for more information.

14 .5.19. Displaying a URI for Connect ion t o a Graphical Display

Running the vi rsh d o md i spl ay command will output a URI which can then be used to connect to
the graphical display of the domain via VNC, SPICE, or RD P. If the --i ncl ud e-passwo rd option is
used, the SPICE channel password will be included in the URI.

14 .5.20. Domain Ret rieval Commands

The following commands will display different information about a given domain
vi rsh d o mho stname domain displays the hostname of the specified domain provided the
hypervisor can publish it.
vi rsh d o mi nfo domain displays basic information about a specified domain.
vi rsh d o mui d domain|ID converts a given domain name or ID into a UUID .
vi rsh d o mi d domain|ID converts a given domain name or UUID into an ID .
vi rsh d o mjo babo rt domain aborts the currently running job on the specified domain.
vi rsh d o mjo bi nfo domain displays information about jobs running on the specified domain,
including migration statistics
vi rsh d o mname domain ID|UUID converts a given domain ID or UUID into a domain name.
vi rsh d o mstate domain displays the state of the given domain. Using the --reaso n option
will also display the reason for the displayed state.
vi rsh d o mco ntro l domain displays the state of an interface to VMM that were used to control
a domain. For states that are not OK or Error, it will also print the number of seconds that have
elapsed since the control interface entered the displayed state.

Examp le 14 .2. Examp le o f st at ist ical f eed b ack

In order to get information about the domain, run the following command:

172

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

# virsh domjobinfo rhel6

Job type:
Unbounded
Time elapsed:
1603
ms
Data processed:
47.004 MiB
Data remaining:
658.633 MiB
Data total:
1.125 GiB
Memory processed: 47.004 MiB
Memory remaining: 658.633 MiB
Memory total:
1.125 GiB
Constant pages:
114382
Normal pages:
12005
Normal data:
46.895 MiB
Expected downtime: 0
ms
Compression cache: 64.000 MiB
Compressed data: 0.000 B
Compressed pages: 0
Compression cache misses: 12005
Compression overflows: 0

14 .5.21. Convert ing QEMU Argument s t o Domain XML

The vi rsh d o mxml -fro m-nati ve provides a way to convert an existing set of QEMU arguments
into a guest description using libvirt D omain XML that can then be used by libvirt. Please note that
this command is intended to be used only to convert existing qemu guests previously started from the
command line in order to allow them to be managed through libvirt. The method described here
should not be used to create new guests from scratch. New guests should be created using either
virsh or virt-manager. Additional information can be found here.
Suppose you have a QEMU guest with the following args file:
$ cat demo.args
LC_ALL=C
PATH=/bin
HOME=/home/test
USER=test
LOGNAME=test /usr/bin/qemu -S -M pc -m 214 -smp 1 -nographic -monitor pty
-no-acpi -boot c -hda /dev/HostVG/QEMUGuest1 -net none -serial none parallel none -usb
To convert this to a domain XML file so that the guest can be managed by libvirt, run:
$ vi rsh d o mxml -fro m-nati ve q emu-arg v d emo . arg s
This command turns the args file above, into this domain XML file:

<domain type='qemu'>
<uuid>00000000-0000-0000-0000-000000000000</uuid>
<memory>219136</memory>
<currentMemory>219136</currentMemory>
<vcpu>1</vcpu>
<os>
<type arch='i686' machine='pc'>hvm</type>

173

Virt ualiz at ion Administ rat ion G uide

<boot dev='hd'/>
</os>
<clock offset='utc'/>
<on_poweroff>destroy</on_poweroff>
<on_reboot>restart</on_reboot>
<on_crash>destroy</on_crash>
<devices>
<emulator>/usr/bin/qemu</emulator>
<disk type='block' device='disk'>
<source dev='/dev/HostVG/QEMUGuest1'/>
<target dev='hda' bus='ide'/>
</disk>
</devices>
</domain>

14 .5.22. Creat ing a Dump File of a Domain's Core

Sometimes it is necessary (especially in the cases of troubleshooting), to create a dump file
containing the core of the domain so that it can be analyzed. In this case, running vi rsh d ump
domain corefilepath --bypass-cache --l i ve | --crash | --reset --verbo se -memo ry-o nl y dumps the domain core to a file specified by the corefilepath Note that some
hypervisors may gave restrictions on this action and may require the user to manually ensure proper
permissions on the file and path specified in the corefilepath parameter. This command is supported
with SR-IOV devices as well as other passthrough devices. The following options are supported and
have the following effect:
--bypass-cache the file saved will not contain the file system cache. Note that selecting this
option may slow down dump operation.
--l i ve will save the file as the domain continues to run and will not pause or stop the domain.
--crash puts the domain in a crashed status rather than leaving it in a paused state while the
dump file is saved.
--reset once the dump file is successfully saved, the domain will reset.
--verbo se displays the progress of the dump process
--memo ry-o nl y the only information that will be saved in the dump file will be the domain's
memory and CPU common register file.
Note that the entire process can be monitored using the d o mjo bi nfo command and can be
canceled using the d o mjo babo rt command.

14 .5.23. Creat ing a Virt ual Machine XML Dump (Configurat ion File)
Output a guest virtual machine's XML configuration file with vi rsh:
# virsh dumpxml {guest-id, guestname or uuid}
This command outputs the guest virtual machine's XML configuration file to standard out (std o ut).
You can save the data by piping the output to a file. An example of piping the output to a file called
guest.xml:
# virsh dumpxml GuestID > guest.xml

174

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

This file g uest. xml can recreate the guest virtual machine (refer to Section 14.6, Editing a Guest
Virtual Machine's configuration file . You can edit this XML configuration file to configure additional
devices or to deploy additional guest virtual machines.
An example of vi rsh d umpxml output:
# virsh dumpxml guest1-rhel6-64
<domain type='kvm'>
<name>guest1-rhel6-64</name>
<uuid>b8d7388a-bbf2-db3a-e962-b97ca6e514bd</uuid>
<memory>2097152</memory>
<currentMemory>2097152</currentMemory>
<vcpu>2</vcpu>
<os>
<type arch='x86_64' machine='rhel6.2.0'>hvm</type>
<boot dev='hd'/>
</os>
<features>
<acpi/>
<apic/>
<pae/>
</features>
<clock offset='utc'/>
<on_poweroff>destroy</on_poweroff>
<on_reboot>restart</on_reboot>
<on_crash>restart</on_crash>
<devices>
<emulator>/usr/libexec/qemu-kvm</emulator>
<disk type='file' device='disk'>
<driver name='qemu' type='raw' cache='none' io='threads'/>
<source file='/home/guest-images/guest1-rhel6-64.img'/>
<target dev='vda' bus='virtio'/>
<shareable/<
<address type='pci' domain='0x0000' bus='0x00' slot='0x05'
function='0x0'/>
</disk>
<interface type='bridge'>
<mac address='52:54:00:b9:35:a9'/>
<source bridge='br0'/>
<model type='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x03'
function='0x0'/>
</interface>
<serial type='pty'>
<target port='0'/>
</serial>
<console type='pty'>
<target type='serial' port='0'/>
</console>
<input type='tablet' bus='usb'/>
<input type='mouse' bus='ps2'/>
<graphics type='vnc' port='-1' autoport='yes'/>
<sound model='ich6'>
<address type='pci' domain='0x0000' bus='0x00' slot='0x04'
function='0x0'/>
</sound>

175

Virt ualiz at ion Administ rat ion G uide

<video>
<model type='cirrus' vram='9216' heads='1'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x02'
function='0x0'/>
</video>
<memballoon model='virtio'>
<address type='pci' domain='0x0000' bus='0x00' slot='0x06'
function='0x0'/>
</memballoon>
</devices>
</domain>
Note that the <shareable/> flag is set. This indicates the device is expected to be shared between
domains (assuming the hypervisor and OS support this), which means that caching should be
deactivated for that device.

14 .5.24 . Creat ing a Guest Virt ual Machine from a Configurat ion File
Guest virtual machines can be created from XML configuration files. You can copy existing XML from
previously created guest virtual machines or use the d umpxml option (refer to Section 14.5.23,
Creating a Virtual Machine XML D ump (Configuration File) ). To create a guest virtual machine with
vi rsh from an XML file:
# virsh create configuration_file.xml

14 .6. Edit ing a Guest Virt ual Machine's configurat ion file
Instead of using the d umpxml option (refer to Section 14.5.23, Creating a Virtual Machine XML
D ump (Configuration File) ), guest virtual machines can be edited either while they are running or
while they are offline. The vi rsh ed i t command provides this functionality. For example, to edit the
guest virtual machine named rhel6:
# virsh edit rhel6
This opens a text editor. The default text editor is the $ED IT O R shell parameter (set to vi by default).

14 .6.1. Adding Mult ifunct ion PCI Devices t o KVM Guest Virt ual Machines
This section will demonstrate how to add multi-function PCI devices to KVM guest virtual machines.
1. Run the vi rsh ed i t [guestname] command to edit the XML configuration file for the
guest virtual machine.
2. In the address type tag, add a mul ti functi o n= ' o n' entry for functi o n= ' 0 x0 ' .
This enables the guest virtual machine to use the multifunction PCI devices.
<disk type='file' device='disk'>
<driver name='qemu' type='raw' cache='none'/>
<source file='/var/lib/libvirt/images/rhel62-1.img'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x05'
function='0x0' multifunction='on'/
</disk>

176

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

For a PCI device with two functions, amend the XML configuration file to include a second
device with the same slot number as the first device and a different function number, such as
functi o n= ' 0 x1' .
For Example:
<disk type='file' device='disk'>
<driver name='qemu' type='raw' cache='none'/>
<source file='/var/lib/libvirt/images/rhel62-1.img'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x05'
function='0x0' multifunction='on'/>
</disk>
<disk type='file' device='disk'>
<driver name='qemu' type='raw' cache='none'/>
<source file='/var/lib/libvirt/images/rhel62-2.img'/>
<target dev='vdb' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x05'
function='0x1'/>
</disk>
3. l spci output from the KVM guest virtual machine shows:
$ lspci
00:05.0 SCSI storage controller: Red Hat, Inc Virtio block device
00:05.1 SCSI storage controller: Red Hat, Inc Virtio block device

14 .6.2. St opping a Running Domain t o Rest art It Lat er

vi rsh manag ed save domain --bypass-cache --runni ng | --paused | --verbo se
saves and destroys (stops) a running domain so that it can be restarted from the same state at a later
time. When used with a vi rsh start command it is automatically started from this save point. If it is
used with the --bypass-cache option the save will avoid the filesystem cache. Note that this option
may slow down the save process speed.
--verbo se displays the progress of the dump process
Under normal conditions, the managed save will decide between using the running or paused state
as determined by the state the domain is in when the save is done. However, this can be overridden
by using the --runni ng option to indicate that it must be left in a running state or by using -paused option which indicates it is to be left in a paused state.
To remove the managed save state, use the vi rsh manag ed save-remo ve command which will
force the domain to do a full boot the next time it is started.
Note that the entire managed save process can be monitored using the d o mjo bi nfo command and
can also be canceled using the d o mjo babo rt command.

14 .6.3. Displaying CPU St at ist ics for a Specified Domain

The vi rsh cpu-stats domain --to tal start co unt command provides the CPU statistical
information on the specified domain. By default it shows the statistics for all CPUs, as well as a total.
The --to tal option will only display the total statistics.

177

Virt ualiz at ion Administ rat ion G uide

14 .6.4 . Saving a Screenshot

The vi rsh screensho t command takes a screenshot of a current domain console and stores it
into a file. If however the hypervisor supports more displays for a domain, using the --screen and
giving a screen ID will specify which screen to capture. In the case where there are multiple graphics
cards, where the heads are numerated before their devices, screen ID 5 addresses the second head
on the second card.

14 .6.5. Sending a Keyst roke Combinat ion t o a Specified Domain

Using the vi rsh send -key domain --co d eset --ho l d ti me keycode command you can
send a sequence as a keycode to a specific domain.
Each keycode can either be a numeric value or a symbolic name from the corresponding codeset. If
multiple keycodes are specified, thay are all sent simultaneously to the guest virtual machine and as
such may be received in random order. If you need distinct keycodes, you must send the send -key
command multiple times.
# vi rsh send -key rhel6 --ho l d ti me 10 0 0 0xf
If a --ho l d ti me is given, each keystroke will be held for the specified amount in milliseconds. The -co d eset allows you to specify a code set, the default being Linux, but the following options are
permitted:
l i nux - choosing this option causes the symbolic names to match the corresponding Linux key
constant macro names and the numeric values are those offered by the Linux generic input event
subsystems.
xt- this will send a value that is defined by the XT keyboard controller. No symbolic names are
provided.
atset1 - the numeric values are those that are defined by the AT keyboard controller, set1 (XT
compatible set). Extended keycodes from the atset1 may differ from extended keycodes in the XT
codeset. No symbolic names are provided.
atset2 - The numeric values are those defined by the AT keyboard controller, set 2. No symbolic
names are provided.
atset3 - The numeric values are those defined by the AT keyboard controller, set 3 (PS/2
compatible). No symbolic names are provided.
o s_x - The numeric values are those defined by the OS-X keyboard input subsystem. The
symbolic names match the corresponding OS-X key constant macro names.
xt_kbd - The numeric values are those defined by the Linux KBD device. These are a variant on
the original XT codeset, but often with different encoding for extended keycodes. No symbolic
names are provided.
wi n32 - The numeric values are those defined by the Win32 keyboard input subsystem. The
symbolic names match the corresponding Win32 key constant macro names.
usb - The numeric values are those defined by the USB HID specification for keyboard input. No
symbolic names are provided.
rfb - The numeric values are those defined by the RFB extension for sending raw keycodes.
These are a variant on the XT codeset, but extended keycodes have the low bit of the second bite
set, instead of the high bit of the first byte. No symbolic names are provided.

178

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

14 .6.6. Sending Process Signal Names t o Virt ual Processes

Using the vi rsh send -pro cess-si g nal domain-ID PID signame command sends the
specified signal (identified by its signame) to a process running in a virtual domain (specified by the
domain ID) and identified by its process ID (PID).
Either an integer signal constant number or a symbolic signal name can be sent this way. Thus, for
example, both of the following commands send the kill signal to process ID 187 on the rhel6 domain:
# vi rsh send -pro cess-si g nal rhel 6 187 ki l l
# vi rsh send -pro cess-si g nal rhel 6 187 9
For the full list of available signals and their uses, see the virsh(1) and signal(7) manual pages.

14 .6.7. Displaying t he IP Address and Port Number for t he VNC Display

The vi rsh vncd i spl ay will print the IP address and port number of the VNC display for the
specified domain. If the information is unavailable the exit code 1 will be displayed.
# virsh vncdisplay rhel6
127.0.0.1:0

14 .7. NUMA Node Management

This section contains the commands needed for NUMA node management.

14 .7.1. Displaying Node Informat ion

The no d ei nfo command displays basic information about the node, including the model number,
number of CPUs, type of CPU, and size of the physical memory. The output corresponds to
vi rNo d eInfo structure. Specifically, the " CPU socket(s)" field indicates the number of CPU sockets
per NUMA cell.
$ vi rsh no d ei nfo
CPU model:
CPU(s):
CPU frequency:
CPU socket(s):
Core(s) per socket:
Thread(s) per core:
NUMA cell(s):
Memory size:

x86_64
4
1199 MHz
1
2
2
1
3715908 KiB

14 .7.2. Set t ing NUMA Paramet ers

The vi rsh numatune can either set or retrieve the NUMA parameters for a specified domain. Within
the D omain XML file these parameters are nested within the <numatune> element. Without using
options, only the current settings are displayed. The numatune domain command requires a
specified domain and can take the following options:
--mo d e - The mode can be set to either stri ct, i nterl eave, or preferred . Running domains
cannot have their mode changed while live unless the domain was started within stri ct mode.

179

Virt ualiz at ion Administ rat ion G uide

--no d eset contains a list of NUMA nodes that are used by the host physical machine for
running the domain. The list contains nodes, each separated by a comma, with a dash - used for
node ranges and a caret ^ used for excluding a node.
Only one of the following three options can be used per instance:
--co nfi g will take effect on the next boot of a persistent guest virtual machine.
--l i ve will set the scheduler information of a running guest virtual machine.
--current will affect the current state of the guest virtual machine.

14 .7.3. Displaying t he Amount of Free Memory in a NUMA Cell

The vi rsh freecel l displays the available amount of memory on the machine within a specified
NUMA cell. This command can provide one of three different displays of available memory on the
machine depending on the options specified. If no options are used, the total free memory on the
machine is displayed. Using the --al l option, it displays the free memory in each cell and the total
free memory on the machine. By using a numeric argument or with --cel l no option along with a
cell number it will display the free memory for the specified cell.

14 .7.4 . Displaying a CPU List

The no d ecpumap command displays the number of CPUs that are available to the node, whether
they are online or not and it also lists the number that are currently online.
$ vi rsh no d ecpumap
CPUs present: 4
CPUs online: 1
CPU map: y

14 .7.5. Displaying CPU St at ist ics

The no d ecpustats command displays statistical information about the specified CPU, if the CPU is
given. If not, it will display the CPU status of the node. If a percent is specified, it will display the
percentage of each type of CPU statistics that were recorded over an one (1) second interval.
This example shows no CPU specified:
$ vi rsh no d ecpustats
user:
1056442260000000
system:
401675280000000
idle:
7549613380000000
iowait:
94593570000000
This example shows the statistical percentages for CPU number 2:
$ vi rsh no d ecpustats 2 --percent
usage:
2.0%
user:
1.0%
system:
1.0%
idle:
98.0%
iowait:
0.0%
You can control the behavior of the rebooting guest virtual machine by modifying the o n_rebo o t

180

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

element in the guest virtual machine's configuration file.

14 .7.6. Suspending t he Host Physical Machine

The no d esuspend command puts the host physical machine into a system-wide sleep state similar
to that of Suspend-to-RAM (s3), Suspend-to-D isk (s4), or Hybrid-Suspend and sets up a Real-TimeClock to wake up the node after the duration that is set has past. The --targ et option can be set to
either mem,disk, or hybrid. These options indicate to set the memory, disk, or combination of the
two to suspend. Setting the --d urati o n instructs the host physical machine to wake up after the set
duration time has run out. It is set in seconds. It is recommended that the duration time be longer than
60 seconds.
$ vi rsh no d esuspend d i sk 6 0

14 .7.7. Set t ing and Displaying t he Node Memory Paramet ers

The no d e-memo ry-tune [shm-pag es-to -scan] [shm-sl eep-mi l i secs] [shm-merg eacro ss-no d es] command displays and allows you to set the node memory parameters. There are
three parameters that may be set with this command:
shm-pag es-to -scan - sets the number of pages to scan before the shared memory service goes
to sleep.
shm-sl eep-mi l i secs - sets the number of miliseconds that the shared memory service will
sleep before the next scan
shm-merg e-acro ss-no d es - specifies if pages from different NUMA nodes can be merged.
Values allowed are 0 and 1. When set to 0, the only pages that can be merged are those that are
physically residing in the memory area of the same NUMA node. When set to 1, pages from all of
the NUMA nodes can be merged. The default setting is 1.

14 .7.8. Creat ing Devices on Host Nodes

The vi rsh no d ed ev-create file command allows you to create a device on a host node and
then assign it to a guest virtual machine. lib virt normally detects which host nodes are available for
use automatically, but this command allows for the registration of host hardware that lib virt did not
detect. The file should contain the XML for the top level <d evi ce> description of the node device.
To stop this device, use the no d ed ev-d estro y device command.

14 .7.9. Det aching a Node Device

The vi rsh no d ed ev-d etach detaches the nodedev from the host so it can be safely used by
guests via <ho std ev> passthrough. This action can be reversed with the no d ed ev-reattach
command but it is done automatically for managed services. This command also accepts no d ed evd ettach.
Note that different drivers expect the device to be bound to different dummy devices. Using the -d ri ver option allows you to specify the desired backend driver.

14 .7.10. Ret rieving a Device's Configurat ion Set t ings

The vi rsh no d ed ev-d umpxml [device] command dumps the XML configuration file for the
given node <d evi ce>. The XML configuration includes information such as: the device name, which
bus owns the device, the vendor, and product ID , etc. The argument device can either be a device

181

Virt ualiz at ion Administ rat ion G uide

name or a WWN pair in WWNN | WWPN format (HBA only).

14 .7.11. List ing Devices on a Node

The vi rsh no d ed ev-l i st cap --tree command lists all the devices available on the node that
are known by lib virt . cap is used to filter the list by capability types, each separated by a comma and
cannot be used with --tree. Using the --tree option, puts the output into a tree structure as shown:
# virsh nodedev-list --tree
computer
|
++++++++|
|

net_lo_00_00_00_00_00_00
net_macvtap0_52_54_00_12_fe_50
net_tun0
net_virbr0_nic_52_54_00_03_7d_cb
pci_0000_00_00_0
pci_0000_00_02_0
pci_0000_00_16_0
pci_0000_00_19_0
|
+- net_eth0_f0_de_f1_3a_35_4f

(this is a partial screen)

14 .7.12. T riggering a Reset for a Node

The no d ed ev-reset nodedev command triggers a device reset for the specified nodedev. Running
this command is useful prior to transferring a node device between guest virtual machine
passthrough and the host physical machine. libvirt will do this action implicitly when required, but this
command allows an explicit reset when needed.

14 .8. St art ing, Suspending, Resuming, Saving, and Rest oring a Guest
Virt ual Machine
This section provides information on starting, suspending, resuming, saving, and restoring guest
virtual machines.

14 .8.1. St art ing a Defined Domain

The vi rsh start domain --co nso l e --paused --auto d estro y --bypass-cache -fo rce-bo o t --pass-fd s command starts a inactive domain that was already defined but whose
state is inactive since its last managed save state or a fresh boot. The command can take the
following options:
--co nso l e - will boot the domain attaching to the console
--paused - If this is supported by the driver it will boot the domain and then put it into a paused
state
--auto d estro y - the guest virtual machine is automatically destroyed when the virsh session
closes or the connection to libvirt closes, or it otherwise exits

182

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

--bypass-cache - used if the domain is in the managedsave state. If this is used, it will restore
the guest virtual machine, avoiding the system cache. Note this will slow down the restore
process.
--fo rce-bo o t - discards any managedsave options and causes a fresh boot to occur
--pass-fd s - is a list of additional options separated by commas, which are passed onto the
guest virtual machine.

14 .8.2. Suspending a Guest Virt ual Machine

Suspend a guest virtual machine with vi rsh:
# virsh suspend {domain-id, domain-name or domain-uuid}
When a guest virtual machine is in a suspended state, it consumes system RAM but not processor
resources. D isk and network I/O does not occur while the guest virtual machine is suspended. This
operation is immediate and the guest virtual machine can be restarted with the resume
(Section 14.8.6, Resuming a Guest Virtual Machine ) option.

14 .8.3. Suspending a Running Domain

The vi rsh d o mpmsuspend domain --d urati o n --targ et command will take a running
domain and suspended it so it can be placed into one of three possible states (S3, S4, or a hybrid of
the two).
# vi rsh d o mpmsuspend rhel6 --d urati o n 10 0 --targ et mem
This command can take the following options:
--d urati o n - sets the duration for the state change in seconds
--targ et - can be either mem (suspend to R AM (S3))d i sk (suspend to d i sk (S4 )),
or hybri d (hybri d suspend )

14 .8.4 . Waking Up a Domain from a pmsuspend St at e

This command will inject a wake-up alert to a guest that is in a pmsuspend state, rather than waiting
for the duration time set to expire. This operation will not fail if the domain is running.
# d o mpmwakeup rhel6
This command requires the name of the domain, rhel6 for example as shown.

14 .8.5. Undefining a Domain

# vi rsh und efi ne domain --manag ed -save --snapsho ts-metad ata --sto rag e -remo ve-al l -sto rag e --wi pe-sto rag e
This command will undefine a domain. Although it can work on a running domain, it will convert the
running domain into a transient domain without stopping it. If the domain is inactive, the domain
configuration is removed.
The command can take the following options:

183

Virt ualiz at ion Administ rat ion G uide

--manag ed -save - this option guarantees that any managed save image is also cleaned up.
Without using this option, attempts to undefine a domain with a managed save image will fail.
--snapsho ts-metad ata - this option guarantees that any snapshots (as shown with
snapsho t-l i st) are also cleaned up when undefining an inactive domain. Note that any
attempts to undefine an inactive domain whose configuration file contains snapshot metadata will
fail. If this option is used and the domain is active, it is ignored.
--sto rag e - using this option requires a comma separated list of volume target names or source
paths of storage volumes to be removed along with the undefined domain. This action will
undefine the storage volume before it is removed. Note that this can only be done with inactive
domains. Note too that this will only work with storage volumes that are managed by libvirt.
--remo ve-al l -sto rag e - in addition to undefining the domain, all associated storage volumes
are deleted.
--wi pe-sto rag e - in addition to deleting the storage volume, the contents are wiped.

14 .8.6. Resuming a Guest Virt ual Machine

Restore a suspended guest virtual machine with vi rsh using the resume option:
# virsh resume {domain-id, domain-name or domain-uuid}
This operation is immediate and the guest virtual machine parameters are preserved for suspend
and resume operations.

14 .8.7. Save a Guest Virt ual Machine

Save the current state of a guest virtual machine to a file using the vi rsh command:
# vi rsh save {domain-name|domain-id|domain-uuid} state-file --bypasscache --xml --runni ng --paused --verbo se
This stops the guest virtual machine you specify and saves the data to a file, which may take some
time given the amount of memory in use by your guest virtual machine. You can restore the state of
the guest virtual machine with the resto re (Section 14.8.11, Restore a Guest Virtual Machine )
option. Save is similar to pause, instead of just pausing a guest virtual machine the present state of
the guest virtual machine is saved.
The vi rsh save command can take the following options:
--bypass-cache - causes the restore to avoid the file system cache but note that using this
option may slow down the restore operation.
--xml - this option must be used with an XML file name. Although this option is usually omitted, it
can be used to supply an alternative XML file for use on a restored guest virtual machine with
changes only in the host-specific portions of the domain XML. For example, it can be used to
account for the file naming differences in underlying storage due to disk snapshots taken after the
guest was saved.
--runni ng - overrides the state recorded in the save image to start the domain as running.
--paused - overrides the state recorded in the save image to start the domain as paused.
--verbo se - displays the progress of the save.

184

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

If you want to restore the guest virtual machine directly from the XML file, the vi rsh resto re
command will do just that. You can monitor the process with the d o mjo bi nfo and cancel it with the
d o mjo babo rt.

14 .8.8. Updat ing t he Domain XML File t hat will be Used for Rest oring t he
Guest
The vi rsh save-i mag e-d efi ne file xml --runni ng | --paused command will update the
domain XML file that will be used when the specified file is later used during the vi rsh resto re
command. The xml argument must be an XML file name containing the alternative XML with changes
only in the host physical machine specific portions of the domain XML. For example, it can be used to
account for the file naming differences resulting from creating disk snapshots of underlying storage
after the guest was saved. The save image records if the domain should be restored to a running or
paused state. Using the options --runni ng or --paused dictates the state that is to be used.

14 .8.9. Ext ract ing t he Domain XML File

save-i mag e-d umpxml file --securi ty-i nfo command will extract the domain XML file that
was in effect at the time the saved state file (used in the vi rsh save command) was referenced.
Using the --securi ty-i nfo option includes security sensitive information in the file.

14 .8.10. Edit Domain XML Configurat ion Files

save-i mag e-ed i t file --runni ng --paused command edits the XML configuration file that
is associated with a saved file that was created by the vi rsh save command.
Note that the save image records whether the domain should be restored to a --runni ng or -paused state. Without using these options the state is determined by the file itself. By selecting -runni ng or --paused you can overwrite the state that vi rsh resto re should use.

14 .8.11. Rest ore a Guest Virt ual Machine

Restore a guest virtual machine previously saved with the vi rsh save command (Section 14.8.7,
Save a Guest Virtual Machine ) using vi rsh:
# virsh restore state-file
This restarts the saved guest virtual machine, which may take some time. The guest virtual machine's
name and UUID are preserved but are allocated for a new id.
The vi rsh resto re state-file command can take the following options:
--bypass-cache - causes the restore to avoid the file system cache but note that using this
option may slow down the restore operation.
--xml - this option must be used with an XML file name. Although this option is usually omitted, it
can be used to supply an alternative XML file for use on a restored guest virtual machine with
changes only in the host-specific portions of the domain XML. For example, it can be used to
account for the file naming differences in underlying storage due to disk snapshots taken after the
guest was saved.
--runni ng - overrides the state recorded in the save image to start the domain as running.
--paused - overrides the state recorded in the save image to start the domain as paused.

185

Virt ualiz at ion Administ rat ion G uide

14 .9. Shut t ing Down, Reboot ing, and Forcing Shut down of a Guest
Virt ual Machine
This section provides information about shutting down, rebooting, and forcing shutdown of a guest
virtual machine.

14 .9.1. Shut t ing Down a Guest Virt ual Machine

Shut down a guest virtual machine using the vi rsh shutd o wn command:
# virsh shutdown {domain-id, domain-name or domain-uuid} [--mo d e method]
You can control the behavior of the rebooting guest virtual machine by modifying the o n_shutd o wn
parameter in the guest virtual machine's configuration file.

14 .9.2. Shut t ing Down Red Hat Ent erprise Linux 6 Guest s on a Red Hat
Ent erprise Linux 7 Host
Installing Red Hat Enterprise Linux 6 guest virtual machines with the Mi ni mal i nstal l ati o n
option does not install the acpid package. Red Hat Enterprise Linux 7 no longer requires this
package, as it has been taken over by systemd . However, Red Hat Enterprise Linux 6 guest virtual
machines running on a Red Hat Enterprise Linux 7 host still require it.
Without the acpid package, the Red Hat Enterprise Linux 6 guest virtual machine does not shut down
when the vi rsh shutd o wn command is executed. The vi rsh shutd o wn command is designed to
gracefully shut down guest virtual machines.
Using vi rsh shutd o wn is easier and safer for system administration. Without graceful shut down
with the vi rsh shutd o wn command a system administrator must log into a guest virtual machine
manually or send the C trl -Al t-D el key combination to each guest virtual machine.

Note
Other virtualized operating systems may be affected by this issue. The vi rsh shutd o wn
command requires that the guest virtual machine operating system is configured to handle
ACPI shut down requests. Many operating systems require additional configuration on the
guest virtual machine operating system to accept ACPI shut down requests.

Pro ced u re 14 .4 . Wo rkaro u n d f o r R ed H at En t erp rise Lin u x 6 g u est s

1. In st all t h e acp id p ackag e
The acpi d service listen and processes ACPI requests.
Log into the guest virtual machine and install the acpid package on the guest virtual machine:
# yum install acpid
2. En ab le t h e acp id service

186

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

Set the acpi d service to start during the guest virtual machine boot sequence and start the
service:
# chkconfig acpid on
# service acpid start
3. Prep are g u est d o main xml
Edit the domain XML file to include the following element. Replace the virtio serial port with
o rg . q emu. g uest_ag ent. 0 and use your guest's name instead of $guestname

<channel type='unix'>
<source mode='bind'
path='/var/lib/libvirt/qemu/{$guestname}.agent'/>
<target type='virtio' name='org.qemu.guest_agent.0'/>
</channel>

Fig u re 14 .2. G u est XML rep lacemen t

4. In st all t h e Q EMU g u est ag en t
Install the QEMU guest agent (QEMU-GA) and start the service as directed in Chapter 10,
QEMU-img and QEMU Guest Agent. If you are running a Windows guest there are instructions
in this chapter for that as well.
5. Sh u t d o wn t h e g u est
a. Run the following commands
# vi rsh l i st --al l - this command lists all of the known
domains
Id Name
State
---------------------------------rhel6
running
b. Shut down the guest virtual machine
# vi rsh shutd o wn rhel6
Domain rhel6 is being shutdown
c. Wait a few seconds for the guest virtual machine to shut down.
# vi rsh l i st --al l
Id Name
State
---------------------------------. rhel6
shut off
d. Start the domain named rhel6, with the XML file you edited.

187

Virt ualiz at ion Administ rat ion G uide

# vi rsh start rhel 6

e. Shut down the acpi in the rhel6 guest virtual machine.
# vi rsh shutd o wn --mo d e acpi rhel6
f. List all the domains again, rhel6 should still be on the list, and it should indicate it is
shut off.
# vi rsh l i st --al l
Id Name
State
---------------------------------rhel6
shut off
g. Start the domain named rhel6, with the XML file you edited.
# vi rsh start rhel6
h. Shut down the rhel6 guest virtual machine guest agent.
# vi rsh shutd o wn --mo d e ag ent rhel6
i. List the domains. rhel6 should still be on the list, and it should indicate it is shut off
# vi rsh l i st --al l
Id Name
State
---------------------------------rhel6
shut off
The guest virtual machine will shut down using the vi rsh shutd o wn command for the consecutive
shutdowns, without using the workaround described above.
In addition to the method described above, a guest can be automatically shutdown, by stopping the
l i bvi rt-g uest service. Refer to Section 14.9.3, Manipulating the libvirt-guests Configuration
Settings for more information on this method.

14 .9.3. Manipulat ing t he libvirt -guest s Configurat ion Set t ings

The l i bvi rt-g uests service has parameter settings that can be configured to assure that the
guest is shutdown properly. It is a package that is a part of the libvirt installation and is installed by
default. This service automatically saves guests to the disk when the host shuts down, and restores
them to their pre-shutdown state when the host reboots. By default, this setting is set to suspend the
guest. If you want the guest to be shutoff, you will need to change one of the parameters of the
l i bvi rt-g uests configuration file.
Pro ced u re 14 .5. C h an g in g t h e lib virt - g u est s service p aramet ers t o allo w f o r t h e
g racef u l sh u t d o wn o f g u est s
The procedure described here allows for the graceful shutdown of guest virtual machines when the
host physical machine is stuck, powered off, or needs to be restarted.
1. O p en t h e co n f ig u rat io n f ile

188

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

The configuration file is located in /etc/sysco nfi g /l i bvi rt-g uests. Edit the file,
remove the comment mark (#) and change the ON_SHUTDOWN=suspend to
ON_SHUTDOWN=shutdown. Remember to save the change.
$ vi /etc/sysconfig/libvirt-guests
# URIs to check for running guests
# example: URIS='default xen:/// vbox+tcp://host/system lxc:///'
#URIS=default
# action taken on host boot
# - start
all guests which were running on shutdown are started
on boot
#
regardless on their autostart settings

# - ignore

libvirt-guests init script won't start any guest on

boot, however,
#

guests marked as autostart will still be automatically

started by
#

libvirtd

#ON_BOOT=start

# Number of seconds to wait between each guest start. Set to 0 to

allow
# parallel startup.
#START_DELAY=0
# action taken on host shutdown
# - suspend
all running guests are suspended using virsh
managedsave
# - shutdown all running guests are asked to shutdown. Please be
careful with
#
this settings since there is no way to distinguish
between a
#
guest which is stuck or ignores shutdown requests and
a guest
#
which just needs a long time to shutdown. When
setting
#
ON_SHUTDOWN=shutdown, you must also set
SHUTDOWN_TIMEOUT to a
#
value suitable for your guests.
ON_SHUTDOWN=shutdown
# If set to non-zero, shutdown will suspend guests concurrently.
Number of
# guests on shutdown at any time will not exceed number set in this
variable.
#PARALLEL_SHUTDOWN=0

189

Virt ualiz at ion Administ rat ion G uide

# Number of seconds we're willing to wait for a guest to shut down.

If parallel
# shutdown is enabled, this timeout applies as a timeout for
shutting down all
# guests on a single URI defined in the variable URIS. If this is
0, then there
# is no time out (use with caution, as guests might not respond to
a shutdown
# request). The default value is 300 seconds (5 minutes).
#SHUTDOWN_TIMEOUT=300
# If non-zero, try to bypass the file system cache when saving and
# restoring guests, even though this may give slower operation for
# some file systems.
#BYPASS_CACHE=0
URIS - checks the specified connections for a running guest. The Default setting
functions in the same manner as vi rsh does when no explicit URI is set In addition,
one can explicitly set the URI from /etc/l i bvi rt/l i bvi rt. co nf. It should be
noted that when using the libvirt configuration file default setting, no probing will be
used.
ON_BOOT - specifies the action to be done to / on the guests when the host boots. The
start option starts all guests that were running prior to shutdown regardless on their
autostart settings. The ignore option will not start the formally running guest on boot,
however, any guest marked as autostart will still be automatically started by libvirtd.
The START_DELAY - sets a delay interval in between starting up the guests. This time
period is set in seconds. Use the 0 time setting to make sure there is no delay and that
all guests are started simultaneously.
ON_SHUTDOWN - specifies the action taken when a host shuts down. Options that can
be set include: suspend which suspends all running guests using vi rsh
manag ed save and shutdown which shuts down all running guests. It is best to be
careful with using the shutd o wn option as there is no way to distinguish between a
guest which is stuck or ignores shutdown requests and a guest that just needs a
longer time to shutdown. When setting the ON_SHUTDOWN=shutdown, you must also
set SHUTDOWN_TIMEOUT to a value suitable for the guests.
PARALLEL_SHUTDOWN D ictates that the number of guests on shutdown at any time will
not exceed number set in this variable and the guests will be suspended concurrently.
If set to 0 , then guests are not shutdown concurrently.
Number of seconds to wait for a guest to shut down. If SHUTDOWN_TIMEOUT is enabled,
this timeout applies as a timeout for shutting down all guests on a single URI defined in
the variable URIS. If SHUTDOWN_TIMEOUT is set to 0, then there is no time out (use with
caution, as guests might not respond to a shutdown request). The default value is 300
seconds (5 minutes).
BYPASS_CACHE can have 2 values, 0 to disable and 1 to enable. If enabled it will bypass the file system cache when guests are restored. Note that setting this may effect
performance and may cause slower operation for some file systems.
2. St art lib virt - g u est s service
If you have not started the service, start the libvirt-guests service. D o not restart the service as
this will cause all running domains to shutdown.

14 .9.4 . Reboot ing a Guest Virt ual Machine

190

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

Use the vi rsh rebo o t command to reboot a guest virtual machine. The prompt will return once the
reboot has executed. Note that there may be a time lapse until the guest virtual machine returns.
#virsh reboot {domain-id, domain-name or domain-uuid} [--mo d e method]
You can control the behavior of the rebooting guest virtual machine by modifying the <o n_rebo o t>
element in the guest virtual machine's configuration file. Refer to Section 20.12, Events
Configuration for more information.
By default, the hypervisor will try to pick a suitable shutdown method. To specify an alternative
method, the --mo d e option can specify a comma separated list which includes i ni tctl , acpi ,
ag ent, and si g nal . The order in which drivers will try each mode is not related to the order
specified in the command. For strict control over ordering, use a single mode at a time and repeat the
command.

14 .9.5. Forcing a Guest Virt ual Machine t o St op

Force a guest virtual machine to stop with the vi rsh d estro y command:
# virsh destroy {domain-id, domain-name or domain-uuid} [--g raceful ]
This command does an immediate ungraceful shutdown and stops the specified guest virtual
machine. Using vi rsh d estro y can corrupt guest virtual machine file systems. Use the d estro y
option only when the guest virtual machine is unresponsive. If you want to initiate a graceful
shutdown, use the vi rsh d estro y --g raceful command.

14 .9.6. Reset t ing a Virt ual Machine

vi rsh reset domain resets the domain immediately without any guest shutdown. A reset emulates
the power reset button on a machine, where all guest hardware sees the RST line and re-initializes
the internal state. Note that without any guest virtual machine OS shutdown, there are risks for data
loss.

14 .10. Ret rieving Guest Virt ual Machine Informat ion

This section provides information on retrieving guest virtual machine information.

14 .10.1. Get t ing t he Domain ID of a Guest Virt ual Machine

To get the domain ID of a guest virtual machine:
# virsh domid {domain-name or domain-uuid}

14 .10.2. Get t ing t he Domain Name of a Guest Virt ual Machine

To get the domain name of a guest virtual machine:
# virsh domname {domain-id or domain-uuid}

14 .10.3. Get t ing t he UUID of a Guest Virt ual Machine

191

Virt ualiz at ion Administ rat ion G uide

To get the Universally Unique Identifier (UUID ) for a guest virtual machine:
# virsh domuuid {domain-id or domain-name}
An example of vi rsh d o muui d output:
# virsh domuuid r5b2-mySQL01
4a4c59a7-ee3f-c781-96e4-288f2862f011

14 .10.4 . Displaying Guest Virt ual Machine Informat ion

Using vi rsh with the guest virtual machine's domain ID , domain name or UUID you can display
information on the specified guest virtual machine:
# virsh dominfo {domain-id, domain-name or domain-uuid}
This is an example of vi rsh d o mi nfo output:
# virsh dominfo
Id:
Name:
UUID:
OS Type:
State:
CPU(s):
CPU time:
Max memory:
Used memory:
Persistent:
Autostart:
Security model:
Security DOI:
Security label:

vr-rhel6u1-x86_64-kvm
9
vr-rhel6u1-x86_64-kvm
a03093a1-5da6-a2a2-3baf-a845db2f10b9
hvm
running
1
21.6s
2097152 kB
1025000 kB
yes
disable
selinux
0
system_u:system_r:svirt_t:s0:c612,c921 (permissive)

14 .11. St orage Pool Commands

The following commands manipulate storage pools. Using libvirt you can manage various storage
solutions, including files, raw partitions, and domain-specific formats, used to provide the storage
volumes visible as devices within virtual machines. For more detailed information about this feature,
see more information at libvirt.org. Many of the commands for storage pools are similar to the ones
used for domains.

14 .11.1. Searching for a St orage Pool XML

The fi nd -sto rag e-po o l -so urces type srcSpec command displays the XML describing all
storage pools of a given type that could be found. If srcSpec is provided, it is a file that contains XML
to further restrict the query for pools.
The fi nd -sto rag e-po o l -so urces-as type host port initiator displays the XML
describing all storage pools of a given type that could be found. If host, port, or initiator are provided,
they control where the query is performed.

192

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

The po o l -i nfo pool-or-uuid command will list the basic information about the specified
storage pool object. This command requires the name or UUID of the storage pool. To retrieve this
information, use the following coomand:
po o l -l i st [--i nacti ve] [--al l ] [--persi stent] [--transi ent] [-auto start] [--no -auto start] [--d etai l s] type
This lists all storage pool objects known to libvirt. By default, only active pools are listed; but using
the --i nacti ve option lists just the inactive pools, and using the --al l option lists all of the
storage pools.
In addition to those options there are several sets of filtering options that can be used to filter the
content of the list. --persi stent restricts the list to persistent pools, --transi ent restricts the list
to transient pools, --auto start restricts the list to autostarting pools and finally --no -auto start
restricts the list to the storage pools that have autostarting disabled.
For all storage pool commands which require a type, the pool types must be separated by comma.
The valid pool types include: d i r, fs, netfs, l o g i cal , d i sk, i scsi , scsi , mpath, rbd , and
sheepd o g .
The --d etai l s option instructs vi rsh to additionally display pool persistence and capacity
related information where available.

Note
When this command is used with older servers, it is forced to use a series of API calls with an
inherent race, where a pool might not be listed or might appear more than once if it changed
its state between calls while the list was being collected. Newer servers however, do not have
this problem.
The po o l -refresh pool-or-uuid refreshes the list of volumes contained in pool.

14 .11.2. Creat ing, Defining, and St art ing St orage Pools

This section provides information about creating, defining, and starting storage pools.

1 4 .1 1 .2 .1 . Building a st o rage po o l
The po o l -bui l d pool-or-uuid --o verwri te --no -o verwri te command builds a pool with
a specified pool name or UUID. The options --o verwri te and --no -o verwri te can only be used
for a pool whose type is file system. If neither option is specified, and the pool is a file system type
pool, then the resulting build will only make the directory.
If --no -o verwri te is specified, it probes to determine if a file system already exists on the target
device, returning an error if it exists, or using mkfs to format the target device if it does not. If -o verwri te is specified, then the mkfs command is executed and any existing data on the target
device is overwritten.

1 4 .1 1 .2 .2 . Cre at ing and de fining a st o rage po o l fro m an XML file

The po o l -create file creates and starts a storage pool from its associated XML file.
The po o l -d efi ne file creates, but does not start, a storage pool object from the XML file.

193

Virt ualiz at ion Administ rat ion G uide

1 4 .1 1 .2 .3. Cre at ing and st art ing a st o rage po o l fro m raw param e t e rs
# po o l -create-as name --pri nt-xml type so urce-ho st so urce-path so urced ev so urce-name <targ et> --so urce-fo rmat format
This command creates and starts a pool object name from the raw parameters given.
If --pri nt-xml is specified, then it prints the XML of the storage pool object without creating the
pool. Otherwise, the pool requires a type in order to be built. For all storage pool commands which
require a type, the pool types must be separated by comma. The valid pool types include: d i r, fs,
netfs, l o g i cal , d i sk, i scsi , scsi , mpath, rbd , and sheepd o g .
In contrast, the following command creates, but does not start, a pool object name from the raw
parameters given:
# po o l -d efi ne-as name --pri nt-xml type so urce-ho st so urce-path so urced ev so urce-name <targ et> --so urce-fo rmat format
If --pri nt-xml is specified, then it prints the XML of the pool object without defining the pool.
Otherwise, the pool has to have a specified type. For all storage pool commands which require a
type, the pool types must be separated by comma. The valid pool types include: d i r, fs, netfs,
l o g i cal , d i sk, i scsi , scsi , mpath, rbd , and sheepd o g .
The po o l -start pool-or-uuid starts the specified storage pool, which was previously defined
but inactive.

1 4 .1 1 .2 .4 . Aut o -st art ing a st o rage po o l

The po o l -auto start pool-or-uuid --d i sabl e command enables or disables a storage pool
to automatically start at boot. This command requires the pool name or UUID . To disable the po o l auto start command use the --d i sabl e option.

14 .11.3. St opping and Delet ing St orage Pools

The po o l -d estro y pool-or-uuid stops a storage pool. Once stopped, libvirt will no longer
manage the pool but the raw data contained in the pool is not changed, and can be later recovered
with the po o l -create command.
The po o l -d el ete pool-or-uuid destroys the resources used by the specified storage pool. It is
important to note that this operation is non-recoverable and non-reversible. However, the pool
structure will still exist after this command, ready to accept the creation of new storage volumes.
The po o l -und efi ne pool-or-uuid command undefines the configuration for an inactive pool.

14 .11.4 . Creat ing an XML Dump File for a St orage Pool

The po o l -d umpxml --i nacti ve pool-or-uuid command returns the XML information about
the specified storage pool object. Using --i nacti ve dumps the configuration that will be used on
next start of the pool as opposed to the current pool configuration.

14 .11.5. Edit ing t he St orage Pool's Configurat ion File

The po o l -ed i t pool-or-uuid opens the specified storage pool's XML configuration file for
editing.

194

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

This method is the only method that should be used to edit an XML configuration file as it does error
checking before applying.

14 .11.6. Convert ing St orage Pools

The po o l -name uuid command converts the specified UUID to a pool name.
The po o l -uui d pool command returns the UUID of the specified pool.

14 .12. St orage Volume Commands

This section covers all commands for creating, deleting, and managing storage volumes. It is best to
do this once you have created a storage pool as the storage pool name or UUID will be required. For
information on storage pools refer to Chapter 12, Storage Pools. For information on storage volumes
refer to, Chapter 13, Volumes .

14 .12.1. Creat ing St orage Volumes

The vo l -create-fro m pool-or-uuid file --i nputpo o l pool-or-uuid vol-name-orkey-or-path command creates a storage volume, using another storage volume as a template for
its contents. This command requires a pool-or-uuid which is the name or UUID of the storage pool to
create the volume in.
The file argument specifies the XML file and path containing the volume definition. The -i nputpo o l pool-or-uuid option specifies the name or uuid of the storage pool the source
volume is in. The vol-name-or-key-or-path argument specifies the name or key or path of the source
volume. For some examples, refer to Section 13.1, Creating Volumes .
The vo l -create-as command creates a volume from a set of arguments. The pool-or-uuid argument
contains the name or UUID of the storage pool to create the volume in.
vo l -create-as pool-or-uuid name capacity --al l o cati o n <si ze> --fo rmat
<stri ng > --backi ng -vo l <vo l -name-o r-key-o r-path> --backi ng -vo l -fo rmat
<stri ng >
name is the name of the new volume. capacity is the size of the volume to be created, as a scaled
integer, defaulting to bytes if there is no suffix. --al l o cati o n <si ze> is the initial size to be
allocated in the volume, also as a scaled integer defaulting to bytes. --fo rmat <stri ng > is used
in file based storage pools to specify the volume file format which is a string of acceptable formats
separated by a comma. Acceptable formats include raw, bo chs, q co w, q co w2, vmd k, --backi ng vo l vol-name-or-key-or-path is the source backing volume to be used if taking a snapshot of
an existing volume. --backi ng -vo l -fo rmat string is the format of the snapshot backing
volume which is a string of formats separated by a comma. Accepted values include: raw, bo chs,
q co w, q co w2, , vmd k, and ho st_d evi ce. These are, however, only meant for file based storage
pools.

1 4 .1 2 .1 .1 . Cre at ing a st o rage vo lum e fro m an XML file

The vo l -create pool-or-uuid file creates a storage volume from a saved XML file. This
command also requires the pool-or-uuid, which is the name or UUID of the storage pool in which the
volume will be created. The file argument contains the path with the volume definition's XML file. An
easy way to create the XML file is to use the vo l -d umpxml command to obtain the definition of a
pre-existing volume, modify it and then save it and then run the vo l -create.

195

Virt ualiz at ion Administ rat ion G uide

virsh vol-dumpxml --pool storagepool1 appvolume1 > newvolume.xml

virsh edit newvolume.xml
virsh vol-create differentstoragepool newvolume.xml
Other options available include:
The --i nacti ve option lists the inactive guest virtual machines (that is, guest virtual machines
that have been defined but are not currently active).
The --al l option lists all guest virtual machines.

1 4 .1 2 .1 .2 . Clo ning a st o rage vo lum e

The vo l -cl o ne --po o l pool-or-uuid vol-name-or-key-or-path name command
clones an existing storage volume. Although the vo l -create-fro m may also be used, it is not the
recommended way to clone a storage volume. The --po o l pool-or-uuid option is the name or
UUID of the storage pool to create the volume in. The vol-name-or-key-or-path argument is the name or
key or path of the source volume. Using a name argument refers to the name of the new volume.

14 .12.2. Delet ing St orage Volumes

The vo l -d el ete --po o l pool-or-uuid vol-name-or-key-or-path command deletes a
given volume. The command requires a specific --po o l pool-or-uuid which is the name or UUID
of the storage pool the volume is in. The vol-name-or-key-or-path option specifies the name or
key or path of the volume to delete.
The vo l -wi pe --po o l pool-or-uuid --al g o ri thm algorithm vol-name-or-key-orpath command wipes a volume, to ensure data previously on the volume is not accessible to future
reads. The command requires a --po o l pool-or-uuid, which is the name or UUID of the storage
pool the volume is in. The vol-name-or-key-or-path contains the name or key or path of the volume to
wipe. Note it is possible to choose different wiping algorithms instead of the default (where every
sector of the storage volume is written with value " 0" ). To specify a wiping algorithm, use the -al g o ri thm option with one of the following supported algorithm types:
zero - 1-pass all zeroes
nnsa - 4-pass NNSA Policy Letter NAP-14.1-C (XVI-8) for sanitizing removable and non-removable
hard disks: random x2, 0x00, verify.
d o d - 4-pass D oD 5220.22-M section 8-306 procedure for sanitizing removable and nonremovable rigid disks: random, 0x00, 0xff, verify.
bsi - 9-pass method recommended by the German Center of Security in Information
Technologies (http://www.bsi.bund.de): 0xff, 0xfe, 0xfd, 0xfb, 0xf7, 0xef, 0xdf, 0xbf, 0x7f.
g utmann - The canonical 35-pass sequence described in Gutmanns paper.
schnei er - 7-pass method described by Bruce Schneier in " Applied Cryptography" (1996):
0x00, 0xff, random x5.
pfi tzner7 - Roy Pfitzners 7-random-pass method: random x7
pfi tzner33 - Roy Pfitzners 33-random-pass method: random x33.
rand o m - 1-pass pattern: random.

196

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

Note
The version of the scrub binary installed on the host will limit the algorithms that are available.

14 .12.3. Dumping St orage Volume Informat ion t o an XML File

vo l -d umpxml --po o l pool-or-uuid vol-name-or-key-or-path command takes the
volume information as an XML dump to a specified file.
This command requires a --po o l pool-or-uuid, which is the name or UUID of the storage pool
the volume is in. vol-name-or-key-or-path is the name or key or path of the volume to place the
resulting XML file.

14 .12.4 . List ing Volume Informat ion

The vo l -i nfo --po o l pool-or-uuid vol-name-or-key-or-path command lists basic
information about the given storage volume --po o l , where pool-or-uuid is the name or UUID of the
storage pool the volume is in. vol-name-or-key-or-path is the name or key or path of the volume to
return information for.
The vo l -l i st --po o l pool-or-uuid --d etai l s lists all of volumes in the specified storage
pool. This command requires --po o l pool-or-uuid which is the name or UUID of the storage
pool. The --d etai l s option instructs virsh to additionally display volume type and capacity related
information where available.

14 .12.5. Ret rieving St orage Volume Informat ion

The vo l -po o l --uui d vol-key-or-path command returns the pool name or UUID for a given
volume. By default, the pool name is returned. If the --uui d option is given, the pool UUID is
returned instead. The command requires the vol-key-or-path which is the key or path of the volume for
which to return the requested information.
The vo l -path --po o l pool-or-uuid vol-name-or-key command returns the path for a
given volume. The command requires --po o l pool-or-uuid, which is the name or UUID of the
storage pool the volume is in. It also requires vol-name-or-key which is the name or key of the volume
for which the path has been requested.
The vo l -name vol-key-or-path command returns the name for a given volume, where vol-keyor-path is the key or path of the volume to return the name for.
The vo l -key --po o l pool-or-uuid vol-name-or-path command returns the volume key for
a given volume where --po o l pool-or-uuid is the name or UUID of the storage pool the volume
is in and vol-name-or-path is the name or path of the volume to return the volume key for.

14 .12.6. Uploading and Downloading St orage Volumes

This section will instruct how to upload and download information to and from storage volumes.

1 4 .1 2 .6 .1 . Uplo ading co nt e nt s t o a st o rage vo lum e

The vo l -upl o ad --po o l pool-or-uuid --o ffset bytes --l eng th bytes vol-nameor-key-or-path local-file command uploads the contents of specified local-file to a storage
volume. The command requires --po o l pool-or-uuid which is the name or UUID of the storage

197

Virt ualiz at ion Administ rat ion G uide

pool the volume is in. It also requires vol-name-or-key-or-path which is the name or key or path of the
volume to wipe. The --o ffset option is the position in the storage volume at which to start writing
the data. --l eng th length dictates an upper limit for the amount of data to be uploaded. An error
will occur if the local-file is greater than the specified --l eng th.

1 4 .1 2 .6 .2 . Do wnlo ading t he co nt e nt s fro m a st o rage vo lum e

# vo l -d o wnl o ad --po o l pool-or-uuid --o ffset bytes --l eng th bytes volname-or-key-or-path local-file
This command command downloads the contents of local-file from a storage volume. It requires a -po o l pool-or-uuid which is the name or UUID of the storage pool that the volume is in. It also
requires vol-name-or-key-or-path which is the name or key or path of the volume to wipe. Using the
option --o ffset dictates the position in the storage volume at which to start reading the data. -l eng th length dictates an upper limit for the amount of data to be downloaded.

14 .12.7. Re-siz ing St orage Volumes

# vo l -resi ze --po o l pool-or-uuid vol-name-or-path pool-or-uuid capacity
--al l o cate --d el ta --shri nk
This command command re-sizes the capacity of the given volume, in bytes. The command requires
--po o l pool-or-uuid which is the name or UUID of the storage pool the volume is in. This
command also requires vol-name-or-key-or-path is the name or key or path of the volume to re-size.
The new capacity may create a sparse file unless the --al l o cate option is specified. Normally,
capacity is the new size, but if --d el ta is present, then it is added to the existing size. Attempts to
shrink the volume will fail unless the --shri nk option is present.
Note that capacity cannot be negative unless the --shri nk option is provided and a negative sign
is not necessary. capacity is a scaled integer which defaults to bytes if there is no suffix. Note too that
this command is only safe for storage volumes not in use by an active guest. Refer to Section 14.5.17,
Using blockresize to Change the Size of a D omain Path for live re-sizing.

14 .13. Displaying Per-guest Virt ual Machine Informat ion

This section provides information about displaying virtual machine information for each guest.

14 .13.1. Displaying t he Guest Virt ual Machines

To display the guest virtual machine list and their current states with vi rsh:
# virsh list
Other options available include:
--i nacti ve option lists the inactive guest virtual machines (that is, guest virtual machines that
have been defined but are not currently active)
--al l option lists all guest virtual machines. For example:
# virsh list --all
Id Name

198

State

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

---------------------------------0 Domain-0
running
1 Domain202
paused
2 Domain010
inactive
3 Domain9600
crashed
There are seven states that can be visible using this command:
Running - The runni ng state refers to guest virtual machines which are currently active on a
CPU.
Idle - The i d l e state indicates that the domain is idle, and may not be running or able to run.
This can be caused because the domain is waiting on IO (a traditional wait state) or has gone
to sleep because there was nothing else for it to do.
Paused - The paused state lists domains that are paused. This occurs if an administrator
uses the paused button in vi rt-manag er or vi rsh suspend . When a guest virtual machine
is paused it consumes memory and other resources but it is ineligible for scheduling and CPU
resources from the hypervisor.
Shutdown - The shutd o wn state is for guest virtual machines in the process of shutting down.
The guest virtual machine is sent a shutdown signal and should be in the process of stopping
its operations gracefully. This may not work with all guest virtual machine operating systems;
some operating systems do not respond to these signals.
Shut off - The shut o ff state indicates that the domain is not running. This can be caused
when a domain completely shuts down or has not been started.
Crashed - The crashed state indicates that the domain has crashed and can only occur if the
guest virtual machine has been configured not to restart on crash.
D ying - D omains in the d yi ng state are in is in process of dying, which is a state where the
domain has not completely shut-down or crashed.
--manag ed -save Although this option alone does not filter the domains, it will list the domains
that have managed save state enabled. In order to actually list the domains seperately you will
need to use the --i nacti ve option as well.
--name is specified domain names are printed in a list. If --uui d is specified the donain's UUID
is printed instead. Using the option --tabl e specifies that a table style output should be used.
All three commands are mutually exclusive
--ti tl e This command must be used with --tabl e output. --ti tl ewill cause an extra column
to be created in the table with the short domain description (title).
--persi stentincludes persistent domains in a list. Use the --transi ent option.
--wi th-manag ed -save lists the domains that have been configured with managed save. To list
the commands without it, use the command --wi tho ut-manag ed -save
--state-runni ng filters out for the domains that are running, --state-paused for paused
domains, --state-shuto ff for domains that are turned off, and --state-o ther lists all states
as a fallback.
--auto start this option will cause the auto-starting domains to be listed. To list domains with
this feature disabled, use the option --no -auto start.
--wi th-snapsho t will list the domains whose snapshot images can be listed. To filter for the
domains without a snapshot, use the option --wi tho ut-snapsho t

199

Virt ualiz at ion Administ rat ion G uide

$ virsh list --title --name

Id
Title
0
Mailserver1
2

Name

State

Domain-0

running

rhelvm

paused

For an example of vi rsh vcpui nfo output, refer to Section 14.13.2, D isplaying Virtual CPU
Information

14 .13.2. Displaying Virt ual CPU Informat ion

To display virtual CPU information from a guest virtual machine with vi rsh:
# virsh vcpuinfo {domain-id, domain-name or domain-uuid}
An example of vi rsh vcpui nfo output:
# virsh vcpuinfo rhel6
VCPU:
0
CPU:
2
State:
running
CPU time:
7152.4s
CPU Affinity:
yyyy
VCPU:
CPU:
State:
CPU time:
CPU Affinity:

1
2
running
10889.1s
yyyy

14 .13.3. Configuring Virt ual CPU Affinit y

To configure the affinity of virtual CPUs with physical CPUs, refer to Example 14.3, Pinning vCPU to
a host physical machine's CPU .

Examp le 14 .3. Pin n in g vC PU t o a h o st p h ysical mach in e' s C PU

The vi rsh vcpupi n assigns a virtual CPU to a physical one.
# virsh vcpupin rhel6
VCPU: CPU Affinity
---------------------------------0: 0-3
1: 0-3
The vcpupi n can take the following options:
--vcpu requires the vcpu number
[--cpul i st] >stri ng < lists the host physical machine's CPU number(s) to set, or omit an
optional query

200

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

--co nfi g affects next boot

--l i ve affects the running domain
--current affects the current domain

14 .13.4 . Displaying Informat ion about t he Virt ual CPU Count s of a Domain
vi rsh vcpuco unt requires a domain name or a domain ID . For example:
# virsh vcpucount rhel6
maximum
config
maximum
live
current
config
current
live

2
2
2
2

The vcpuco unt can take the following options:

--maxi mum displays the maximum number of vCPUs available
--acti ve displays the number of currently active vCPUs
--l i ve displays the value from the running domain
--co nfi g displays the value to be configured on guest virtual machine's next boot
--current displays the value according to current domain state
--g uest displays the count that is returned is from the perspective of the guest

14 .13.5. Configuring Virt ual CPU Affinit y

To configure the affinity of virtual CPUs with physical CPUs:
# virsh vcpupin domain-id vcpu cpulist
The d o mai n-i d parameter is the guest virtual machine's ID number or name.
The vcpu parameter denotes the number of virtualized CPUs allocated to the guest virtual
machine.The vcpu parameter must be provided.
The cpul i st parameter is a list of physical CPU identifier numbers separated by commas. The
cpul i st parameter determines which physical CPUs the VCPUs can run on.
Additional parameters such as --co nfi g affect the next boot, whereas --l i ve affects the running
domain, and --current affects the current domain.

14 .13.6. Configuring Virt ual CPU Count

To modify the number of CPUs assigned to a guest virtual machine, use the vi rsh setvcpus
command:
# vi rsh setvcpus {domain-name, domain-id or domain-uuid} count [[-config] [--live] | [--current] [--guest]

201

Virt ualiz at ion Administ rat ion G uide

The following parameters may be set for the vi rsh setvcpus command:
{domain-name, domain-id or domain-uuid} - Specifies the virtual machine.
count - Specifies the number of virtual CPUs to set.

Note
The count value cannot exceed the number of CPUs that were assigned to the guest
virtual machine when it was created. It may also be limited by the host or the hypervisor.
For Xen, you can only adjust the virtual CPUs of a running domain if the domain is
paravirtualized.
--live - The default option, used if none are specified. The configuration change takes effect on
the running guest virtual machine. This is referred to as a hot plug if the number of vCPUs is
increased, and hot unplug if it is reduced.

Important
The vCPU hot unplug feature is a Technology Preview. Therefore, it is not supported and
not recommended for use in high-value deployments.
--config - The configuration change takes effect on the next reboot of the guest. Both the -co nfi g and --l i ve options may be specified together if supported by the hypervisor.
--current - Configuration change takes effect on the current state of the guest virtual machine. If
used on a running guest, it acts as --live, if used on a shut-down guest, it acts as --config.
--maximum - Sets a maximum vCPU limit that can be hot-plugged on the next reboot of the guest.
As such, it must only be used with the --co nfi g option, and not with the --l i ve option.
--guest - Instead of a hot plug or a hot unplug, the QEMU guest agent modifies the vCPU count
directly in the running guest by enabling or disabling vCPUs. This option cannot be used with
count value higher than the current number of vCPUs in the gueet, and configurations set with -guest are reset when a guest is rebooted.

Note
For information on increasing vCPU performance by using multi-queue, refer to the Red Hat
Enterprise Linux Virtualization Tuning and Optimization Guide.

Examp le 14 .4 . vC PU h o t p lu g an d h o t u n p lu g
To hot-plug a vCPU, run the following command on a guest with a single vCPU:
vi rsh setvcpus guestVM1 2 --live
This increases the number of vCPUs for guestVM1 to two. The change is performed while guestVM1
is running, as indicated by the --live option.

202

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

To hot-unplug one vCPU from the same running guest, run the following:
vi rsh setvcpus guestVM1 1 --live
Be aware, however, that currently, using vCPU hot unplug can lead to problems with further
modifications of the vCPU count.

14 .13.7. Configuring Memory Allocat ion

To modify a guest virtual machine's memory allocation with vi rsh:
# vi rsh setmem {domain-id or domain-name} co unt
# vi rsh setmem vr-rhel6u1-x86_64-kvm --kilobytes 1025000
You must specify the co unt in kilobytes. The new count value cannot exceed the amount you
specified when you created the guest virtual machine. Values lower than 64 MB are unlikely to work
with most guest virtual machine operating systems. A higher maximum memory value does not affect
active guest virtual machines. If the new value is lower than the available memory, it will shrink
possibly causing the guest virtual machine to crash.
This command has the following options:
[--domain] <string> domain name, id or uuid
[--size] <number> new memory size, as scaled integer (default KiB)
Valid memory units include:
b or bytes for bytes
KB for kilobytes (10 3 or blocks of 1,000 bytes)
k or KiB for kibibytes (2 10 or blocks of 1024 bytes)
MB for megabytes (10 6 or blocks of 1,000,000 bytes)
M or MiB for mebibytes (2 20 or blocks of 1,048,576 bytes)
GB for gigabytes (10 9 or blocks of 1,000,000,000 bytes)
G or GiB for gibibytes (2 30 or blocks of 1,073,741,824 bytes)
TB for terabytes (10 12 or blocks of 1,000,000,000,000 bytes)
T or TiB for tebibytes (2 40 or blocks of 1,099,511,627,776 bytes)
Note that all values will be rounded up to the nearest kibibyte by libvirt, and may be further
rounded to the granularity supported by the hypervisor. Some hypervisors also enforce a
minimum, such as 4000KiB (or 4000 x 2 10 or 4,096,000 bytes). The units for this value are
determined by the optional attribute memory unit, which defaults to the kibibytes (KiB) as a unit
of measure where the value given is multiplied by 2 10 or blocks of 1024 bytes.
--config takes affect next boot

203

Virt ualiz at ion Administ rat ion G uide

--live controls the memory of the running domain

--current controls the memory on the current domain

14 .13.8. Changing t he Memory Allocat ion for t he Domain

The vi rsh setmaxmem domain size --co nfi g --l i ve --current allows the setting of the
maximum memory allocation for a guest virtual machine as shown:
vi rsh setmaxmem rhel6 1024 --current
The size that can be given for the maximum memory is a scaled integer that by default is expressed in
kibibytes, unless a supported suffix is provided. The following options can be used with this
command:
--co nfi g - takes affect next boot
--l i ve - controls the memory of the running domain, providing the hypervisor supports this
action as not all hypervisors allow live changes of the maximum memory limit.
--current - controls the memory on the current domain

14 .13.9. Displaying Guest Virt ual Machine Block Device Informat ion
Use vi rsh d o mbl kstat to display block device statistics for a running guest virtual machine.
# virsh domblkstat GuestName block-device

14 .13.10. Displaying Guest Virt ual Machine Net work Device Informat ion
Use vi rsh d o mi fstat to display network interface statistics for a running guest virtual machine.
# virsh domifstat GuestName interface-device

14 .14 . Managing Virt ual Net works

This section covers managing virtual networks with the vi rsh command. To list virtual networks:
# virsh net-list
This command generates output similar to:
# virsh net-list
Name
State
Autostart
----------------------------------------default
active
yes
vnet1
active
yes
vnet2
active
yes
To view network information for a specific virtual network:
# virsh net-dumpxml NetworkName

204

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

This displays information about a specified virtual network in XML format:

# virsh net-dumpxml vnet1
<network>
<name>vnet1</name>
<uuid>98361b46-1581-acb7-1643-85a412626e70</uuid>
<forward dev='eth0'/>
<bridge name='vnet0' stp='on' forwardDelay='0' />
<ip address='192.168.100.1' netmask='255.255.255.0'>
<dhcp>
<range start='192.168.100.128' end='192.168.100.254' />
</dhcp>
</ip>
</network>
Other vi rsh commands used in managing virtual networks are:
vi rsh net-auto start network-name Autostart a network specified as network-name.
vi rsh net-create XMLfile generates and starts a new network using an existing XML
file.
vi rsh net-d efi ne XMLfile generates a new network device from an existing XML file
without starting it.
vi rsh net-d estro y network-name destroy a network specified as network-name.
vi rsh net-name networkUUID convert a specified networkUUID to a network name.
vi rsh net-uui d network-name convert a specified network-name to a network UUID .
vi rsh net-start nameOfInactiveNetwork starts an inactive network.
vi rsh net-und efi ne nameOfInactiveNetwork removes the definition of an inactive
network.

14 .15. Migrat ing Guest Virt ual Machines wit h virsh

Information on migration using virsh is located in the section entitled Live KVM Migration with virsh
Refer to Section 4.4, Live KVM Migration with virsh

14 .15.1. Int erface Commands

The following commands manipulate host interfaces and as such should not be run from the guest
virtual machine. These commands should be run from a terminal on the host physical machine.

Warning
The commands in this section are only supported if the machine has the NetworkManager
service disabled, and is using the netwo rk service instead.
Often, these host interfaces can then be used by name within domain <i nterface> elements (such
as a system-created bridge interface), but there is no requirement that host interfaces be tied to any
particular guest configuration XML at all. Many of the commands for host interfaces are similar to the

205

Virt ualiz at ion Administ rat ion G uide

ones used for domains, and the way to name an interface is either by its name or its MAC address.
However, using a MAC address for an i face option only works when that address is unique (if an
interface and a bridge share the same MAC address, which is often the case, then using that MAC
address results in an error due to ambiguity, and you must resort to a name instead).

1 4 .1 5 .1 .1 . De fining and st art ing a ho st physical m achine int e rface via an XML file
The vi rsh i face-d efi ne file command define a host interface from an XML file. This command
will only define the interface and will not start it.
vi rsh i face-d efi ne iface.xml
To start an interface which has already been defined, run i face-start interface, where interface
is the interface name.

1 4 .1 5 .1 .2 . Edit ing t he XML co nfigurat io n file fo r t he ho st int e rface

The command i face-ed i t interface edits the XML configuration file for a host interface. This is
the o n ly recommended way to edit the XML configuration file. (Refer to Chapter 20, Manipulating the
Domain XML for more information about these files.)

1 4 .1 5 .1 .3. List ing act ive ho st int e rface s

The i face-l i st --i nacti ve --al l displays a list of active host interfaces. If --al l is
specified, this list will also include interfaces that are defined but are inactive. If --i nacti ve is
specified only the inactive interfaces will be listed.

1 4 .1 5 .1 .4 . Co nve rt ing a MAC addre ss int o an int e rface nam e

The i face-name interface command converts a host interface MAC to an interface name,
provided the MAC address is unique among the hosts interfaces. This command requires interface
which is the interface's MAC address.
The i face-mac interface command will convert a host's interface name to MAC address where
in this case interface, is the interface name.

1 4 .1 5 .1 .5 . St o pping a spe cific ho st physical m achine int e rface

The vi rsh i face-d estro y interface command destroys (stops) a given host interface, which
is the same as running i f-d o wn on the host. This command will disable that interface from active
use and takes effect immediately.
To undefine the interface, use the i face-und efi ne interface command along with the interface
name.

1 4 .1 5 .1 .6 . Displaying t he ho st co nfigurat io n file

vi rsh i face-d umpxml interface --i nacti ve displays the host interface information as an
XML dump to stdout. If the --i nacti ve option is specified, then the output reflects the persistent
state of the interface that will be used the next time it is started.

1 4 .1 5 .1 .7 . Cre at ing bridge de vice s

The i face-bri d g e creates a bridge device named bridge, and attaches the existing network device
interface to the new bridge, which starts working immediately, with STP enabled and a delay of 0.

206

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

# vi rsh i face-bri d g e interface bridge --no -stp delay --no -start

Note that these settings can be altered with --no-stp, --no-start, and an integer number of seconds for
delay. All IP address configuration of interface will be moved to the new bridge device. Refer to
Section 14.15.1.8, Tearing down a bridge device for information on tearing down the bridge.

1 4 .1 5 .1 .8 . T e aring do wn a bridge de vice

The i face-unbri d g e bridge --no -start command tears down a specified bridge device
named bridge, releases its underlying interface back to normal usage, and moves all IP address
configuration from the bridge device to the underlying device. The underlying interface is restarted
unless --no -start option is used, but keep in mind not restarting is generally not recommended.
Refer to Section 14.15.1.7, Creating bridge devices for the command to use to create a bridge.

1 4 .1 5 .1 .9 . Manipulat ing int e rface snapsho t s

The i face-beg i n command creates a snapshot of current host interface settings, which can later
be committed (with i face-co mmi t) or restored (i face-ro l l back). If a snapshot already exists,
then this command will fail until the previous snapshot has been committed or restored. Undefined
behavior will result if any external changes are made to host interfaces outside of the libvirt API
between the time of the creation of a snapshot and its eventual commit or rollback.
Use the i face-co mmi t command to declare all changes made since the last i face-beg i n as
working, and then delete the rollback point. If no interface snapshot has already been started via
i face-beg i n, then this command will fail.
Use the i face-ro l l back to revert all host interface settings back to the state that recorded the last
time the i face-beg i n command was executed. If i face-beg i n command had not been
previously executed, then i face-ro l l back will fail. Note that rebooting the host physical machine
also serves as an implicit rollback point.

14 .15.2. Managing Snapshot s

The sections that follow describe actions that can be done in order to manipulate domain snapshots.
Snapshots take the disk, memory, and device state of a domain at a specified point-in-time, and save
it for future use. Snapshots have many uses, from saving a " clean" copy of an OS image to saving a
domains state before what may be a potentially destructive operation. Snapshots are identified with
a unique name. See the libvirt website for documentation of the XML format used to represent
properties of snapshots.

1 4 .1 5 .2 .1 . Cre at ing Snapsho t s

The vi rsh snapsho t-create command creates a snapshot for domain with the properties
specified in the domain XML file (such as <name> and <description> elements, as well as <disks>).
To create a snapshot, run:
# snapsho t-create <d o mai n> <xml fi l e> [--red efi ne] [--current] [--no metad ata] [--reuse-external ]
The domain name, ID , or UID may be used as the domain requirement. The XML requirement is a
string must contain the <name>, <description> and <disks> elements.

207

Virt ualiz at ion Administ rat ion G uide

Note
Live snapshots are not supported in Red Hat Enterprise Linux. There are additional options
available with the vi rsh snapsho t-create command for use with live snapshots which are
visible in libvirt, but not supported in Red Hat Enterprise Linux 6.
The options available in Red Hat Enterprise Linux include:
--red efi ne specifies that if all XML elements produced by snapsho t-d umpxml are valid; it can
be used to migrate snapshot hierarchy from one machine to another, to recreate hierarchy for the
case of a transient domain that goes away and is later recreated with the same name and UUID ,
or to make slight alterations in the snapshot metadata (such as host-specific aspects of the
domain XML embedded in the snapshot). When this option is supplied, the xml fi l e argument is
mandatory, and the domains current snapshot will not be altered unless the --current option is
also given.
--no -metad ata creates the snapshot, but any metadata is immediately discarded (that is, libvirt
does not treat the snapshot as current, and cannot revert to the snapshot unless --red efi ne is
later used to teach libvirt about the metadata again).
--reuse-external , if used, this option specifies the location of an existing external XML
snapshot to use. If an existing external snapshot does not already exist, the command will fail to
take a snapshot to avoid losing contents of the existing files.

1 4 .1 5 .2 .2 . Cre at ing a snapsho t fo r t he curre nt do m ain

The vi rsh snapsho t-create-as d o mai n command creates a snapshot for the domain with the
properties specified in the domain XML file (such as <name> and <description> elements). If these
values are not included in the XML string, libvirt will choose a value. To create a snapshot run:
# vi rsh snapsho t-create-as d o mai n {[--pri nt-xml ] | [--no -metad ata] [-reuse-external ]} [name] [d escri pti o n] [--d i skspec] d i skspec]
The remaining options are as follows:
--pri nt-xml creates appropriate XML for snapsho t-create as output, rather than actually
creating a snapshot.
--d i skspec option can be used to control how --d i sk-o nl y and external checkpoints create
external files. This option can occur multiple times, according to the number of <disk> elements in
the domain XML. Each <diskspec> is in the form disk[,snapsho t= type][,d ri ver= type]
[,fi l e= name]. To include a literal comma in disk or in fi l e= name, escape it with a second
comma. A literal --d i skspec must precede each diskspec unless all three of <domain>,
<name>, and <description> are also present. For example, a diskspec of
vd a,snapsho t= external ,fi l e= /path/to ,,new results in the following XML:

<disk name=vda snapshot=external>

<source file=/path/to,new/>
</disk>
--reuse-external creates an external snapshot reusing an existing file as the destination
(meaning this file is overwritten). If this destination does not exist, the snapshot request will be
refused to avoid losing contents of the existing files.

208

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

--no -metad ata creates snapshot data but any metadata is immediately discarded (that is, libvirt
does not treat the snapshot as current, and cannot revert to the snapshot unless snapshot-create
is later used to teach libvirt about the metadata again). This option is incompatible with -pri nt-xml .

1 4 .1 5 .2 .3. T aking a snapsho t o f t he curre nt do m ain

This command is used to query which snapshot is currently in use. To use, run:
# vi rsh snapsho t-current d o mai n {[--name] | [--securi ty-i nfo ] |
[snapsho tname]}
If snapsho tname is not used, snapshot XML for the domains current snapshot (if there is one) will
be displayed as output. If --name is specified, just the current snapshot name instead of the full XML
will be sent as output. If --securi ty-i nfo is supplied, security sensitive information will be
included in the XML. Using snapsho tname, libvirt generates a request to make the existing named
snapshot become the current snapshot, without reverting it to the domain.

1 4 .1 5 .2 .4 . snapsho t -e dit -do m ain

This command is used to edit the snapshot that is currently in use. To use, run:
#vi rsh snapsho t-ed i t d o mai n [snapsho tname] [--current] {[--rename] [-cl o ne]}
If both snapsho tname and --current are specified, it forces the edited snapshot to become the
current snapshot. If snapsho tname is omitted, then --current must be supplied, in order to edit the
current snapshot.
This is equivalent to the following command sequence below, but it also includes some error
checking:
# virsh snapshot-dumpxml dom name > snapshot.xml
# vi snapshot.xml [note - this can be any editor]
# virsh snapshot-create dom snapshot.xml --redefine [--current]
If --rename is specified, then the resulting edited file gets saved in a different file name. If --cl o ne
is specified, then changing the snapshot name will create a clone of the snapshot metadata. If neither
is specified, then the edits will not change the snapshot name. Note that changing a snapshot name
must be done with care, since the contents of some snapshots, such as internal snapshots within a
single qcow2 file, are accessible only from the original snapshot filename.

1 4 .1 5 .2 .5 . snapsho t -info -do m ain

snapsho t-i nfo -d o mai n displays information about the snapshots. To use, run:
# snapsho t-i nfo d o mai n {snapsho t | --current}
Outputs basic information about a specified snapsho t , or the current snapshot with --current.

1 4 .1 5 .2 .6 . snapsho t -list -do m ain

List all of the available snapshots for the given domain, defaulting to show columns for the snapshot
name, creation time, and domain state. To use, run:

209

Virt ualiz at ion Administ rat ion G uide

#vi rsh snapsho t-l i st d o mai n [{--parent | --ro o ts | --tree}] [{[--fro m]

snapsho t | --current} [--d escend ants]] [--metad ata] [--no -metad ata] [-l eaves] [--no -l eaves] [--i nacti ve] [--acti ve] [--i nternal ] [--external ]
The remaining optional options are as follows:
--parent adds a column to the output table giving the name of the parent of each snapshot. This
option may not be used with --ro o ts or --tree.
--ro o ts filters the list to show only the snapshots that have no parents. This option may not be
used with --parent or --tree.
--tree displays output in a tree format, listing just snapshot names. These three options are
mutually exclusive. This option may not be used with --ro o ts or --parent.
--fro m filters the list to snapshots which are children of the given snapshot; or if --current is
provided, will cause the list to start at the current snapshot. When used in isolation or with -parent, the list is limited to direct children unless --d escend ants is also present. When used
with --tree, the use of --d escend ants is implied. This option is not compatible with --ro o ts.
Note that the starting point of --fro m or --current is not included in the list unless the --tree
option is also present.
--l eaves is specified, the list will be filtered to just snapshots that have no children. Likewise, if -no -l eaves is specified, the list will be filtered to just snapshots with children. (Note that omitting
both options does no filtering, while providing both options will either produce the same list or
error out depending on whether the server recognizes the options) Filtering options are not
compatible with --tree..
--metad ata is specified, the list will be filtered to just snapshots that involve libvirt metadata, and
thus would prevent the undefining of a persistent domain, or be lost on destroy of a transient
domain. Likewise, if --no -metad ata is specified, the list will be filtered to just snapshots that
exist without the need for libvirt metadata.
--i nacti ve is specified, the list will be filtered to snapshots that were taken when the domain
was shut off. If --acti ve is specified, the list will be filtered to snapshots that were taken when the
domain was running, and where the snapshot includes the memory state to revert to that running
state. If --d i sk-o nl y is specified, the list will be filtered to snapshots that were taken when the
domain was running, but where the snapshot includes only disk state.
--i nternal is specified, the list will be filtered to snapshots that use internal storage of existing
disk images. If --external is specified, the list will be filtered to snapshots that use external files for
disk images or memory state.

1 4 .1 5 .2 .7 . snapsho t -dum pxm l do m ain snapsho t

vi rsh snapsho t-d umpxml d o mai n snapsho t outputs the snapshot XML for the domains
snapshot named snapshot. To use, run:
# vi rsh snapsho t-d umpxml d o mai n snapsho t [--securi ty-i nfo ]
The --securi ty-i nfo option will also include security sensitive information. Use snapsho tcurrent to easily access the XML of the current snapshot.

1 4 .1 5 .2 .8 . snapsho t -pare nt do m ain

210

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

Outputs the name of the parent snapshot, if any, for the given snapshot, or for the current snapshot
with --current. To use, run:
#vi rsh snapsho t-parent d o mai n {snapsho t | --current}

1 4 .1 5 .2 .9 . snapsho t -re ve rt do m ain

Reverts the given domain to the snapshot specified by snapsho t, or to the current snapshot with -current.

Warning
Be aware that this is a destructive action; any changes in the domain since the last snapshot
was taken will be lost. Also note that the state of the domain after snapsho t-revert is
complete will be the state of the domain at the time the original snapshot was taken.
To revert the snapshot, run
# snapsho t-revert d o mai n {snapsho t | --current} [{--runni ng | --paused }]
[--fo rce]
Normally, reverting to a snapshot leaves the domain in the state it was at the time the snapshot was
created, except that a disk snapshot with no guest virtual machine state leaves the domain in an
inactive state. Passing either the --runni ng or --paused option will perform additional state
changes (such as booting an inactive domain, or pausing a running domain). Since transient
domains cannot be inactive, it is required to use one of these options when reverting to a disk
snapshot of a transient domain.
There are two cases where a snapsho t revert involves extra risk, which requires the use of -fo rce to proceed. One is the case of a snapshot that lacks full domain information for reverting
configuration; since libvirt cannot prove that the current configuration matches what was in use at
the time of the snapshot, supplying --fo rce assures libvirt that the snapshot is compatible with the
current configuration (and if it is not, the domain will likely fail to run). The other is the case of
reverting from a running domain to an active state where a new hypervisor has to be created rather
than reusing the existing hypervisor, because it implies drawbacks such as breaking any existing
VNC or Spice connections; this condition happens with an active snapshot that uses a provably
incompatible configuration, as well as with an inactive snapshot that is combined with the --start
or --pause option.

1 4 .1 5 .2 .1 0 . snapsho t -de le t e do m ain

snapsho t-d el ete d o mai n deletes the snapshot for the specified domain. To do this, run:
# vi rsh snapsho t-d el ete d o mai n {snapsho t | --current} [--metad ata] [{-chi l d ren | --chi l d ren-o nl y}]
This command D eletes the snapshot for the domain named snapsho t, or the current snapshot with
--current. If this snapshot has child snapshots, changes from this snapshot will be merged into the
children. If the option --chi l d ren is used, then it will delete this snapshot and any children of this
snapshot. If --chi l d ren-o nl y is used, then it will delete any children of this snapshot, but leave
this snapshot intact. These two options are mutually exclusive.

211

Virt ualiz at ion Administ rat ion G uide

The --metad ata is used it will delete the snapshot's metadata maintained by libvirt, while leaving the
snapshot contents intact for access by external tools; otherwise deleting a snapshot also removes its
data contents from that point in time.

14 .16. Guest Virt ual Machine CPU Model Configurat ion

This section provides information about guest virtual machine CPU model configuration.

14 .16.1. Int roduct ion

Every hypervisor has its own policy for what a guest virtual machine will see for its CPUs by default.
Whereas some hypervisors decide which CPU host physical machine features will be available for
the guest virtual machine, QEMU/KVM presents the guest virtual machine with a generic model named
q emu32 or q emu6 4 . These hypervisors perform more advanced filtering, classifying all physical
CPUs into a handful of groups and have one baseline CPU model for each group that is presented to
the guest virtual machine. Such behavior enables the safe migration of guest virtual machines
between host physical machines, provided they all have physical CPUs that classify into the same
group. libvirt does not typically enforce policy itself, rather it provides the mechanism on which the
higher layers define their own desired policy. Understanding how to obtain CPU model information
and define a suitable guest virtual machine CPU model is critical to ensure guest virtual machine
migration is successful between host physical machines. Note that a hypervisor can only emulate
features that it is aware of and features that were created after the hypervisor was released may not
be emulated.

14 .16.2. Learning about t he Host Physical Machine CPU Model

The vi rsh capabi l i ti es command displays an XML document describing the capabilities of the
hypervisor connection and host physical machine. The XML schema displayed has been extended to
provide information about the host physical machine CPU model. One of the big challenges in
describing a CPU model is that every architecture has a different approach to exposing their
capabilities. On x86, the capabilities of a modern CPU are exposed via the CPUID instruction.
Essentially this comes down to a set of 32-bit integers with each bit given a specific meaning.
Fortunately AMD and Intel agree on common semantics for these bits. Other hypervisors expose the
notion of CPUID masks directly in their guest virtual machine configuration format. However,
QEMU/KVM supports far more than just the x86 architecture, so CPUID is clearly not suitable as the
canonical configuration format. QEMU ended up using a scheme which combines a CPU model
name string, with a set of named options. On x86, the CPU model maps to a baseline CPUID mask,
and the options can be used to then toggle bits in the mask on or off. libvirt decided to follow this
lead and uses a combination of a model name and options.
It is not practical to have a database listing all known CPU models, so libvirt has a small list of
baseline CPU model names. It chooses the one that shares the greatest number of CPUID bits with
the actual host physical machine CPU and then lists the remaining bits as named features. Notice
that libvirt does not display which features the baseline CPU contains. This might seem like a flaw at
first, but as will be explained in this section, it is not actually necessary to know this information.

14 .16.3. Det ermining a Compat ible CPU Model t o Suit a Pool of Host Physical
Machines
Now that it is possible to find out what CPU capabilities a single host physical machine has, the next
step is to determine what CPU capabilities are best to expose to the guest virtual machine. If it is
known that the guest virtual machine will never need to be migrated to another host physical
machine, the host physical machine CPU model can be passed straight through unmodified. A
virtualized data center may have a set of configurations that can guarantee all servers will have

212

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

100% identical CPUs. Again the host physical machine CPU model can be passed straight through
unmodified. The more common case, though, is where there is variation in CPUs between host
physical machines. In this mixed CPU environment, the lowest common denominator CPU must be
determined. This is not entirely straightforward, so libvirt provides an API for exactly this task. If libvirt
is provided a list of XML documents, each describing a CPU model for a host physical machine,
libvirt will internally convert these to CPUID masks, calculate their intersection, and convert the
CPUID mask result back into an XML CPU description.
Here is an example of what libvirt reports as the capabilities on a basic workstation, when the vi rsh
capabi l i ti esis executed:

<capabilities>
<host>
<cpu>
<arch>i686</arch>
<model>pentium3</model>
<topology sockets='1' cores='2' threads='1'/>
<feature name='lahf_lm'/>
<feature name='lm'/>
<feature name='xtpr'/>
<feature name='cx16'/>
<feature name='ssse3'/>
<feature name='tm2'/>
<feature name='est'/>
<feature name='vmx'/>
<feature name='ds_cpl'/>
<feature name='monitor'/>
<feature name='pni'/>
<feature name='pbe'/>
<feature name='tm'/>
<feature name='ht'/>
<feature name='ss'/>
<feature name='sse2'/>
<feature name='acpi'/>
<feature name='ds'/>
<feature name='clflush'/>
<feature name='apic'/>
</cpu>
</host>
</capabilities>
Fig u re 14 .3. Pu llin g h o st p h ysical mach in e' s C PU mo d el in f o rmat io n
Now compare that to any random server, with the same vi rsh capabi l i ti es command:

<capabilities>
<host>
<cpu>
<arch>x86_64</arch>

213

Virt ualiz at ion Administ rat ion G uide

<model>phenom</model>
<topology sockets='2' cores='4' threads='1'/>
<feature name='osvw'/>
<feature name='3dnowprefetch'/>
<feature name='misalignsse'/>
<feature name='sse4a'/>
<feature name='abm'/>
<feature name='cr8legacy'/>
<feature name='extapic'/>
<feature name='cmp_legacy'/>
<feature name='lahf_lm'/>
<feature name='rdtscp'/>
<feature name='pdpe1gb'/>
<feature name='popcnt'/>
<feature name='cx16'/>
<feature name='ht'/>
<feature name='vme'/>
</cpu>
...snip...
Fig u re 14 .4 . G en erat e C PU d escrip t io n f ro m a ran d o m server
To see if this CPU description is compatible with the previous workstation CPU description, use the
vi rsh cpu-co mpare command.
The reduced content was stored in a file named vi rsh-caps-wo rkstati o n-cpu-o nl y. xml and
the vi rsh cpu-co mpare command can be executed on this file:
# virsh cpu-compare virsh-caps-workstation-cpu-only.xml
Host physical machine CPU is a superset of CPU described in virsh-capsworkstation-cpu-only.xml
As seen in this output, libvirt is correctly reporting that the CPUs are not strictly compatible. This is
because there are several features in the server CPU that are missing in the client CPU. To be able to
migrate between the client and the server, it will be necessary to open the XML file and comment out
some features. To determine which features need to be removed, run the vi rsh cpu-basel i ne
command, on the bo th-cpus. xml which contains the CPU information for both machines. Running
# vi rsh cpu-basel i ne bo th-cpus. xml , results in:

<cpu match='exact'>
<model>pentium3</model>
<feature policy='require'
<feature policy='require'
<feature policy='require'
<feature policy='require'
<feature policy='require'
<feature policy='require'
<feature policy='require'
<feature policy='require'
<feature policy='require'
</cpu>

214

name='lahf_lm'/>
name='lm'/>
name='cx16'/>
name='monitor'/>
name='pni'/>
name='ht'/>
name='sse2'/>
name='clflush'/>
name='apic'/>

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

Fig u re 14 .5. C o mp o sit e C PU b aselin e

This composite file shows which elements are in common. Everything that is not in common should
be commented out.

14 .17. Configuring t he Guest Virt ual Machine CPU Model

For simple defaults, the guest virtual machine CPU configuration accepts the same basic XML
representation as the host physical machine capabilities XML exposes. In other words, the XML from
the cpu-basel i ne virsh command can now be copied directly into the guest virtual machine XML at
the top level under the <domain> element. In the previous XML snippet, there are a few extra attributes
available when describing a CPU in the guest virtual machine XML. These can mostly be ignored, but
for the curious here is a quick description of what they do. The top level <cpu> element has an
attribute called match with possible values of:
match='minimum' - the host physical machine CPU must have at least the CPU features described
in the guest virtual machine XML. If the host physical machine has additional features beyond the
guest virtual machine configuration, these will also be exposed to the guest virtual machine.
match='exact' - the host physical machine CPU must have at least the CPU features described in
the guest virtual machine XML. If the host physical machine has additional features beyond the
guest virtual machine configuration, these will be masked out from the guest virtual machine.
match='strict' - the host physical machine CPU must have exactly the same CPU features
described in the guest virtual machine XML.
The next enhancement is that the <feature> elements can each have an extra 'policy' attribute with
possible values of:
policy='force' - expose the feature to the guest virtual machine even if the host physical machine
does not have it. This is usually only useful in the case of software emulation.
policy='require' - expose the feature to the guest virtual machine and fail if the host physical
machine does not have it. This is the sensible default.
policy='optional' - expose the feature to the guest virtual machine if it happens to support it.
policy='disable' - if the host physical machine has this feature, then hide it from the guest virtual
machine.
policy='forbid' - if the host physical machine has this feature, then fail and refuse to start the guest
virtual machine.
The 'forbid' policy is for a niche scenario where an incorrectly functioning application will try to use a
feature even if it is not in the CPUID mask, and you wish to prevent accidentally running the guest
virtual machine on a host physical machine with that feature. The 'optional' policy has special
behavior with respect to migration. When the guest virtual machine is initially started the parameter is
optional, but when the guest virtual machine is live migrated, this policy turns into 'require', since you
cannot have features disappearing across migration.

14 .18. Managing Resources for Guest Virt ual Machines

virsh allows the grouping and allocation of resources on a per guest virtual machine basis. This is
managed by the libvirt daemon, which creates cgroups and manages them on behalf of the guest
virtual machine. The only thing that is left for the system administrator to do is to either query or set
tunables against specified guest virtual machines. The following tunables may used:

215

Virt ualiz at ion Administ rat ion G uide

memo ry - The memory controller allows for setting limits on RAM and swap usage and querying
cumulative usage of all processes in the group
cpuset - The CPU set controller binds processes within a group to a set of CPUs and controls
migration between CPUs.
cpuacct - The CPU accounting controller provides information about CPU usage for a group of
processes.
cpu -The CPU scheduler controller controls the prioritization of processes in the group. This is
similar to granting ni ce level privileges.
d evi ces - The devices controller grants access control lists on character and block devices.
freezer - The freezer controller pauses and resumes execution of processes in the group. This is
similar to SIG ST O P for the whole group.
net_cl s - The network class controller manages network utilization by associating processes
with a tc network class.
In creating a group hierarchy cgroup will leave mount point and directory setup entirely to the
administrators discretion and is more complex than just adding some mount points to /etc/fstab.
It is necessary to setup the directory hierarchy and decide how processes get placed within it. This
can be done with the following virsh commands:
sched i nfo - described in Section 14.19, Setting Schedule Parameters
bl ki o tune- described in Section 14.20, D isplay or Set Block I/O Parameters
d o mi ftune- described in Section 14.5.9, Setting Network Interface Bandwidth Parameters
memtune - described in Section 14.21, Configuring Memory Tuning

14 .19. Set t ing Schedule Paramet ers

sched i nfo allows scheduler parameters to be passed to guest virtual machines. The following
command format should be used:
#vi rsh sched i nfo domain --set --wei g ht --cap --current --co nfi g --l i ve
Each parameter is explained below:
d o mai n - this is the guest virtual machine domain
--set - the string placed here is the controller or action that is to be called. Additional parameters
or values if required should be added as well.
--current - when used with --set, will use the specified set string as the current scheduler
information. When used without will display the current scheduler information.
--co nfi g - - when used with --set, will use the specified set string on the next reboot. When
used without will display the scheduler information that is saved in the configuration file.
--l i ve - when used with --set, will use the specified set string on a guest virtual machine that
is currently running. When used without will display the configuration setting currently used by
the running virtual machine

216

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

The scheduler can be set with any of the following parameters: cpu_shares, vcpu_period and
vcpu_quota.

Examp le 14 .5. sch ed in f o sh o w

This example shows the shell guest virtual machine's schedule information
# virsh schedinfo shell
Scheduler
: posix
cpu_shares
: 1024
vcpu_period
: 100000
vcpu_quota
: -1

Examp le 14 .6 . sch ed in f o set

In this example, the cpu_shares is changed to 2046. This effects the current state and not the
configuration file.
# virsh schedinfo --set cpu_shares=2046 shell
Scheduler
: posix
cpu_shares
: 2046
vcpu_period
: 100000
vcpu_quota
: -1

14 .20. Display or Set Block I/O Paramet ers

bl ki o tune sets and or displays the I/O parameters for a specified guest virtual machine. The
following format should be used:
# vi rsh bl ki o tune d o mai n [--wei g ht wei g ht] [--d evi ce-wei g hts d evi cewei g hts] [[--co nfi g ] [--l i ve] | [--current]]
More information on this command can be found in the Virtualization Tuning and Optimization Guide

14 .21. Configuring Memory T uning

The vi rsh memtune vi rtual _machi ne --parameter si ze is covered in the Virtualization
Tuning and Optimization Guide.

14 .22. Virt ual Net working Commands

The following commands manipulate virtual networks. libvirt has the capability to define virtual
networks which can then be used by domains and linked to actual network devices. For more
detailed information about this feature see the documentation at libvirt's website . Many of the
commands for virtual networks are similar to the ones used for domains, but the way to name a
virtual network is either by its name or UUID .

14 .22.1. Aut ost art ing a Virt ual Net work

217

Virt ualiz at ion Administ rat ion G uide

This command will configure a virtual network to be started automatically when the guest virtual
machine boots. To run this command:
# vi rsh net-auto start network [--d i sabl e]
This command accepts the --d i sabl e option which disables the autostart command.

14 .22.2. Creat ing a Virt ual Net work from an XML File
This command creates a virtual network from an XML file. Refer to libvirt's website to get a description
of the XML network format used by libvirt. In this command file is the path to the XML file. To create the
virtual network from an XML file, run:
# vi rsh net-create file

14 .22.3. Defining a Virt ual Net work from an XML File

This command defines a virtual network from an XML file, the network is just defined but not
instantiated. To define the virtual network, run:
# net-d efi ne file

14 .22.4 . St opping a Virt ual Net work

This command destroys (stops) a given virtual network specified by its name or UUID . This takes
effect immediately. To stop the specified network network is required.
# net-d estro y network

14 .22.5. Creat ing a Dump File

This command outputs the virtual network information as an XML dump to stdout for the specified
virtual network. If --i nacti ve is specified, then physical functions are not expanded into their
associated virtual functions. To create the dump file, run:
# vi rsh net-d umpxml network [--i nacti ve]

14 .22.6. Edit ing a Virt ual Net work's XML Configurat ion File
The following command edits the XML configuration file for a network. This is equivalent to:
#vi rsh net-d umpxml --i nacti ve network > network.xml
vi network.xml (or make changes with your other text editor)
virsh net-define network.xml
except that it does some error checking. The editor used can be supplied by the $VISUAL or
$ED ITOR environment variables, and defaults to " vi" . To edit the network, run:
#vi rsh net-ed i t network

218

Chapt er 1 4 . Managing guest virt ual machines wit h virsh

14 .22.7. Get t ing Informat ion about a Virt ual Net work
This command returns basic information about the network object. To get the network information,
run:
# vi rsh net-i nfo network

14 .22.8. List ing Informat ion about a Virt ual Net work
Returns the list of active networks, if --al l is specified this will also include defined but inactive
networks, if --i nacti ve is specified only the inactive ones will be listed. You may also want to filter
the returned networks by --persi stent to list the persitent ones, --transi ent to list the transient
ones, --auto start to list the ones with autostart enabled, and --no -auto start to list the ones
with autostart disabled.
Note: When talking to older servers, this command is forced to use a series of API calls with an
inherent race, where a pool might not be listed or might appear more than once if it changed state
between calls while the list was being collected. Newer servers do not have this problem.
To list the virtual networks, run:
# net-l i st [--i nacti ve | --al l ] [--persi stent] [<--transi ent>] [-auto start] [<--no -auto start>]

14 .22.9. Convert ing a Net work UUID t o Net work Name

This command converts a network UUID to network name. To do this run:
# vi rsh net-name network-UUID

14 .22.10. St art ing a (Previously Defined) Inact ive Net work

This command starts a (previously defined) inactive network. To do this, run:
# vi rsh net-start network

14 .22.11. Undefining t he Configurat ion for an Inact ive Net work

This command undefines the configuration for an inactive network. To do this, run:
# net-und efi ne network

14 .22.12. Convert ing a Net work Name t o Net work UUID

This command converts a network name to network UUID . To do this, run:
# vi rsh net-uui d network-name

14 .22.13. Updat ing an Exist ing Net work Definit ion File

219

Virt ualiz at ion Administ rat ion G uide

This command updates the given section of an existing network definition, taking effect immediately,
without needing to destroy and re-start the network. This command is one of " add-first" , " add-last" ,
" add" (a synonym for add-last), " delete" , or " modify" . section is one of " " bridge" , " domain" , " ip" , " ipdhcp-host" , " ip-dhcp-range" , " forward" , " forward-interface" , " forward-pf" , " portgroup" , " dns-host" ,
" dns-txt" , or " dns-srv" , each section being named by a concatenation of the xml element hierarchy
leading to the element being changed. For example, " ip-dhcp-host" will change a <ho st> element
that is contained inside a <d hcp> element inside an <i p> element of the network. xml is either the
text of a complete xml element of the type being changed (e.g. " <host mac=" 00:11:22:33:44:55
ip=192.0.2.1/>" , or the name of a file that contains a complete xml element. D isambiguation is done
by looking at the first character of the provided text - if the first character is " <" , it is xml text, if the first
character is not " >" , it is the name of a file that contains the xml text to be used. The --parent-index
option is used to specify which of several parent elements the requested element is in (0-based). For
example, a dhcp <ho st> element could be in any one of multiple <i p> elements in the network; if a
parent-index isnt provided, the " most appropriate" <i p> element will be selected (usually the only
one that already has a <d hcp> element), but if --parent-index is given, that particular instance of
<i p> will get the modification. If --live is specified, affect a running network. If --config is
specified, affect the next startup of a persistent network. If -- current is specified, affect the current
network state. Both --live and --config options may be given, but --current is exclusive. Not
specifying any option is the same as specifying --current.
To update the configuration file, run:
# vi rsh net-upd ate network command section xml [--parent-index index]
[[--live] [--config] | [--current]]

220

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Chapter 15. Managing Guests with the Virtual Machine Manager

(virt-manager)
This section describes the Virtual Machine Manager (vi rt-manag er) windows, dialog boxes, and
various GUI controls.
vi rt-manag er provides a graphical view of hypervisors and guests on your host system and on
remote host systems. vi rt-manag er can perform virtualization management tasks, including:
defining and creating guests,
assigning memory,
assigning virtual CPUs,
monitoring operational performance,
saving and restoring, pausing and resuming, and shutting down and starting guests,
links to the textual and graphical consoles, and
live and offline migrations.

15.1. St art ing virt -manager

To start vi rt-manag er session open the Ap p licat io n s menu, then the Syst em T o o ls menu and
select Virt u al Mach in e Man ag er (vi rt-manag er).
The vi rt-manag er main window appears.

Fig u re 15.1. St art in g vi rt-manag er

Alternatively, vi rt-manag er can be started remotely using ssh as demonstrated in the following
command:
ssh -X host's address
[remotehost]# virt-manager

221

Virt ualiz at ion Administ rat ion G uide

Using ssh to manage virtual machines and hosts is discussed further in Section 5.1, Remote
Management with SSH .

15.2. T he Virt ual Machine Manager Main Window

This main window displays all the running guests and resources used by guests. Select a guest by
double clicking the guest's name.

Fig u re 15.2. Virt u al Mach in e Man ag er main win d o w

15.3. T he Virt ual Hardware Det ails Window

The virtual hardware details window displays information about the virtual hardware configured for
the guest. Virtual hardware resources can be added, removed and modified in this window. To
access the virtual hardware details window, click on the icon in the toolbar.

222

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.3. T h e virt u al h ard ware d et ails ico n

Clicking the icon displays the virtual hardware details window.

223

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.4 . T h e virt u al h ard ware d et ails win d o w

15.3.1. At t aching USB Devices t o a Guest Virt ual Machine

Note
In order to attach the USB device to the guest virtual machine, you first must attach it to the
host physical machine and confirm that the device is working. If the guest is running, you need
to shut it down before proceeding.

Pro ced u re 15.1. At t ach in g U SB D evices u sin g Virt - Man ag er

1. Open the guest virtual machine's Virtual Machine D etails screen.
2. Click Ad d Hard ware

224

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.5. Ad d H ard ware B u t t o n

3. In the Ad d N ew Virt u al H ard ware popup, select U SB H o st D evice, select the device you
want to attach from the list and Click Fi ni sh.

225

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.6 . Ad d U SB D evice

4. To use the USB device in the guest virtual machine, start the guest virtual machine.

15.4 . Virt ual Machine Graphical Console

This window displays a guest's graphical console. Guests can use several different protocols to
export their graphical framebuffers: vi rt-manag er supports VN C and SPIC E. If your virtual
machine is set to require authentication, the Virtual Machine graphical console prompts you for a
password before the display appears.

226

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.7. G rap h ical co n so le win d o w

Note
VNC is considered insecure by many security experts, however, several changes have been
made to enable the secure usage of VNC for virtualization on Red Hat enterprise Linux. The
guest machines only listen to the local host's loopback address (127. 0 . 0 . 1). This ensures
only those with shell privileges on the host can access virt-manager and the virtual machine
through VNC. Although virt-manager is configured to listen to other public network interfaces
and alternative methods can be configured, it is not recommended.
Remote administration can be performed by tunneling over SSH which encrypts the traffic.
Although VNC can be configured to access remotely without tunneling over SSH, for security
reasons, it is not recommended. To remotely administer the guest follow the instructions in:
Chapter 5, Remote Management of Guests. TLS can provide enterprise level security for
managing guest and host systems.

227

Virt ualiz at ion Administ rat ion G uide

Your local desktop can intercept key combinations (for example, Ctrl+Alt+F1) to prevent them from
being sent to the guest machine. You can use the Sen d key menu option to send these sequences.
From the guest machine window, click the Sen d key menu and select the key sequence to send. In
addition, from this menu you can also capture the screen output.
SPICE is an alternative to VNC available for Red Hat Enterprise Linux.

15.5. Adding a Remot e Connect ion

This procedure covers how to set up a connection to a remote system using vi rt-manag er.
1. To create a new connection open the Fi l e menu and select the Ad d C o nnecti o n. . .
menu item.
2. The Ad d C o nnecti o n wizard appears. Select the hypervisor. For Red Hat Enterprise Linux
6 systems select Q EMU/KVM. Select Local for the local system or one of the remote connection
options and click C o nnect. This example uses Remote tunnel over SSH which works on
default installations. For more information on configuring remote connections refer to
Chapter 5, Remote Management of Guests

Fig u re 15.8. Ad d C o n n ect io n

3. Enter the root password for the selected host when prompted.
A remote host is now connected and appears in the main vi rt-manag er window.

228

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.9 . R emo t e h o st in t h e main virt - man ag er win d o w

15.6. Displaying Guest Det ails

You can use the Virtual Machine Monitor to view activity information for any virtual machines on your
system.
To view a virtual system's details:
1. In the Virtual Machine Manager main window, highlight the virtual machine that you want to
view.

Fig u re 15.10. Select in g a virt u al mach in e t o d isp lay

2. From the Virtual Machine Manager Ed i t menu, select Vi rtual Machi ne D etai l s.

229

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.11. D isp layin g t h e virt u al mach in e d et ails

When the Virtual Machine details window opens, there may be a console displayed. Should
this happen, click Vi ew and then select D etai l s. The Overview window opens first by
default. To go back to this window, select O vervi ew from the navigation pane on the left
hand side.
The O vervi ew view shows a summary of configuration details for the guest.

230

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.12. D isp layin g g u est d et ails o verview

3. Select P erfo rmance from the navigation pane on the left hand side.
The P erfo rmance view shows a summary of guest performance, including CPU and Memory
usage.

231

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.13. D isp layin g g u est p erf o rman ce d et ails

4. Select P ro cesso r from the navigation pane on the left hand side. The P ro cesso r view
allows you to view the current processor allocation, as well as to change it.
It is also possible to change the number of virtual CPUs (vCPUs) while the virtual machine is
running, which is referred to as hot plugging and hot unplugging.

Important
The hot unplugging feature is only available as a Technology Preview. Therefore, it is
not supported and not recommended for use in high-value deployments.

232

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.14 . Pro cesso r allo cat io n p an el

5. Select Memo ry from the navigation pane on the left hand side. The Memo ry view allows you
to view or change the current memory allocation.

233

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.15. D isp layin g memo ry allo cat io n

6. Each virtual disk attached to the virtual machine is displayed in the navigation pane. Click on
a virtual disk to modify or remove it.

234

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.16 . D isp layin g d isk co n f ig u rat io n

7. Each virtual network interface attached to the virtual machine is displayed in the navigation
pane. Click on a virtual network interface to modify or remove it.

235

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.17. D isp layin g n et wo rk co n f ig u rat io n

15.7. Performance Monit oring

Performance monitoring preferences can be modified with vi rt-manag er's preferences window.
To configure performance monitoring:
1. From the Ed i t menu, select P references.

Fig u re 15.18. Mo d if yin g g u est p ref eren ces

The P references window appears.
2. From the Stats tab specify the time in seconds or stats polling options.

236

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.19 . C o n f ig u rin g p erf o rman ce mo n it o rin g

15.8. Displaying CPU Usage for Guest s

To view the CPU usage for all guests on your system:
1. From the Vi ew menu, select G raph, then the G uest C P U Usag e check box.

237

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.20. En ab lin g g u est C PU u sag e st at ist ics g rap h in g

2. The Virtual Machine Manager shows a graph of CPU usage for all virtual machines on your
system.

Fig u re 15.21. G u est C PU u sag e g rap h

15.9. Displaying CPU Usage for Host s

To view the CPU usage for all hosts on your system:
1. From the Vi ew menu, select G raph, then the Ho st C P U Usag e check box.

Fig u re 15.22. En ab lin g h o st C PU u sag e st at ist ics g rap h in g

2. The Virtual Machine Manager shows a graph of host CPU usage on your system.

238

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.23. H o st C PU u sag e g rap h

15.10. Displaying Disk I/O

To view the disk I/O for all virtual machines on your system:
1. Make sure that the D isk I/O statistics collection is enabled. To do this, from the Ed i t menu,
select P references and click the Statstab.
2. Select the D i sk I/O checkbox.

239

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.24 . En ab lin g D isk I/O

3. To enable the D isk I.O display, from the Vi ew menu, select G raph, then the D i sk I/O check
box.

Fig u re 15.25. Select in g D isk I/O

4. The Virtual Machine Manager shows a graph of D isk I/O for all virtual machines on your
system.

Fig u re 15.26 . D isp layin g D isk I/O

15.11. Displaying Net work I/O

To view the network I/O for all virtual machines on your system:
1. Make sure that the Network I/O statistics collection is enabled. To do this, from the Ed i t
menu, select P references and click the Statstab.
2. Select the Netwo rk I/O checkbox.

24 0

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.27. En ab lin g N et wo rk I/O

3. To display the Network I/O statistics, from the Vi ew menu, select G raph, then the Netwo rk
I/O check box.

24 1

Virt ualiz at ion Administ rat ion G uide

Fig u re 15.28. Select in g N et wo rk I/O

4. The Virtual Machine Manager shows a graph of Network I/O for all virtual machines on your
system.

24 2

Chapt er 1 5. Managing G uest s wit h t he Virt ual Machine Manager (virt - manager)

Fig u re 15.29 . D isp layin g N et wo rk I/O

24 3

Virt ualiz at ion Administ rat ion G uide

Chapter 16. Guest Virtual Machine Disk Access with Offline

Tools
16.1. Int roduct ion
Red Hat Enterprise Linux 6 comes with tools to access, edit and create host physical machine disks
or other disk images. There are several uses for these tools, including:
Viewing or downloading files located on a host physical machine disk.
Editing or uploading files onto a host physical machine disk.
Reading or writing host physical machine configuration.
Reading or writing the Windows Registry in Windows host physical machines.
Preparing new disk images containing files, directories, file systems, partitions, logical volumes
and other options.
Rescuing and repairing host physical machines that fail to boot or those that need boot
configuration changes.
Monitoring disk usage of host physical machines.
Auditing compliance of host physical machines, for example to organizational security standards.
D eploying host physical machines by cloning and modifying templates.
Reading CD and D VD ISO and floppy disk images.

Warning
You must n ever use these tools to write to a host physical machine or disk image which is
attached to a running virtual machine, not even to open such a disk image in write mode.
D oing so will result in disk corruption of the guest virtual machine. The tools try to prevent you
from doing this, however do not catch all cases. If there is any suspicion that a guest virtual
machine might be running, it is strongly recommended that the tools not be used, or at least
always use the tools in read-only mode.

24 4

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

Note
Some virtualization commands in Red Hat Enterprise Linux 6 allow you to specify a remote
libvirt connection. For example:
virt-df -c qemu://remote/system -d Guest
However, libguestfs in Red Hat Enterprise Linux 6 cannot access remote guests, and
commands using remote URLs like this do not work as expected. This affects the following Red
Hat Enterprise Linux 6 commands:
guestfish
guestmount
virt-alignment-scan
virt-cat
virt-copy-in
virt-copy-out
virt-df
virt-edit
virt-filesystems
virt-inspector
virt-inspector2
virt-list-filesystems
virt-list-partitions
virt-ls
virt-rescue
virt-sysprep
virt-tar
virt-tar-in
virt-tar-out
virt-win-reg

16.2. T erminology
This section explains the terms used throughout this chapter.
lib g u est f s ( G U EST FileSyst em LIB rary) - the underlying C library that provides the basic
functionality for opening disk images, reading and writing files and so on. You can write C
programs directly to this API, but it is quite low level.
g u est f ish ( G U EST Filesyst em In t eract ive SH ell) is an interactive shell that you can use
from the command line or from shell scripts. It exposes all of the functionality of the libguestfs API.
Various virt tools are built on top of libguestfs, and these provide a way to perform specific single
tasks from the command line. Tools include virt - d f , virt - rescu e, virt - resiz e and virt - ed it .
h ivex and Au g eas are libraries for editing the Windows Registry and Linux configuration files
respectively. Although these are separate from libguestfs, much of the value of libguestfs comes
from the combination of these tools.
g u est mo u n t is an interface between libguestfs and FUSE. It is primarily used to mount file
systems from disk images on your host physical machine. This functionality is not necessary, but
can be useful.

24 5

Virt ualiz at ion Administ rat ion G uide

16.3. Inst allat ion

To install libguestfs, guestfish, the libguestfs tools, guestmount and support for Windows guest
virtual machines, subscribe to the RHEL V2WIN channel, go to the Red Hat Website and run the
following command:
# yum install libguestfs guestfish libguestfs-tools libguestfs-mount
libguestfs-winsupport
To install every libguestfs-related package including the language bindings, run the following
command:
# yum install '*guestf*'

16.4 . T he guest fish Shell

g u est f ish is an interactive shell that you can use from the command line or from shell scripts to
access guest virtual machine file systems. All of the functionality of the libguestfs API is available
from the shell.
To begin viewing or editing a virtual machine disk image, run the following command, substituting
the path to your desired disk image:
guestfish --ro -a /path/to/disk/image
- - ro means that the disk image is opened read-only. This mode is always safe but does not allow
write access. Only omit this option when you are cert ain that the guest virtual machine is not
running, or the disk image is not attached to a live guest virtual machine. It is not possible to use
l i bg uest vi rtual machi nefs to edit a live guest virtual machine, and attempting to will result in
irreversible disk corruption.
/p at h /t o /d isk/imag e is the path to the disk. This can be a file, a host physical machine logical
volume (such as /dev/VG/LV), a host physical machine device (/dev/cdrom) or a SAN LUN (/dev/sdf3).

Note
libguestfs and guestfish do not require root privileges. You only need to run them as root if the
disk image being accessed needs root to read and/or write.
When you start guestfish interactively, it will display this prompt:
guestfish --ro -a /path/to/disk/image
Welcome to guestfish, the libguestfs filesystem interactive shell for
editing virtual machine filesystems.
Type: 'help' for help on commands
'man' to read the manual
'quit' to quit the shell
><fs>

24 6

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

At the prompt, type ru n to initiate the library and attach the disk image. This can take up to 30
seconds the first time it is done. Subsequent starts will complete much faster.

Note
libguestfs will use hardware virtualization acceleration such as KVM (if available) to speed up
this process.
Once the ru n command has been entered, other commands can be used, as the following section
demonstrates.

16.4 .1. Viewing File Syst ems wit h guest fish

This section provides information about viewing files with guestfish.

1 6 .4 .1 .1 . Manual List ing and Vie wing

The l i st-fi l esystems command will list file systems found by libguestfs. This output shows a
Red Hat Enterprise Linux 4 disk image:
><fs> run
><fs> list-filesystems
/dev/vda1: ext3
/dev/VolGroup00/LogVol00: ext3
/dev/VolGroup00/LogVol01: swap
This output shows a Windows disk image:
><fs> run
><fs> list-filesystems
/dev/vda1: ntfs
/dev/vda2: ntfs
Other useful commands are l i st-d evi ces, l i st-parti ti o ns, l vs, pvs, vfs-type and fi l e.
You can get more information and help on any command by typing hel p command, as shown in the
following output:
><fs> help vfs-type
NAME
vfs-type - get the Linux VFS type corresponding to a mounted device
SYNOPSIS
vfs-type device
DESCRIPTION
This command gets the filesystem type corresponding to the filesystem
on
"device".
For most filesystems, the result is the name of the Linux VFS module
which would be used to mount this filesystem if you mounted it

24 7

Virt ualiz at ion Administ rat ion G uide

without
specifying the filesystem type. For example a string such as "ext3"
or
"ntfs".
To view the actual contents of a file system, it must first be mounted. This example uses one of the
Windows partitions shown in the previous output (/d ev/vd a2), which in this case is known to
correspond to the C :\ drive:
><fs> mount-ro
><fs> ll /
total 1834753
drwxrwxrwx 1
drwxr-xr-x 21
lrwxrwxrwx 2
drwxrwxrwx 1
drwxrwxrwx 1
drwxrwxrwx 1

/dev/vda2 /

root
root
root
root
root
root

root
root
root
root
root
root

4096
4096
60
4096
4096
16384

Nov
Nov
Jul
Nov
Sep
Sep

1
16
14
15
19
19

11:40
21:45
2009
18:00
10:34
10:34

.
..
Documents and Settings
Program Files
Users
Windows

You can use guestfish commands such as l s, l l , cat, mo re, d o wnl o ad and tar-o ut to view and
download files and directories.

Note
There is no concept of a current working directory in this shell. Unlike ordinary shells, you
cannot for example use the cd command to change directories. All paths must be fully
qualified starting at the top with a forward slash (/) character. Use the Tab key to complete
paths.
To exit from the guestfish shell, type exi t or enter C trl + d .

1 6 .4 .1 .2 . Via gue st fish inspe ct io n

Instead of listing and mounting file systems by hand, it is possible to let guestfish itself inspect the
image and mount the file systems as they would be in the guest virtual machine. To do this, add the - i
option on the command line:
guestfish --ro -a /path/to/disk/image -i
Welcome to guestfish, the libguestfs filesystem interactive shell for
editing virtual machine filesystems.
Type: 'help' for help on commands
'man' to read the manual
'quit' to quit the shell
Operating system: Red Hat Enterprise Linux AS release 4 (Nahant Update
8)
/dev/VolGroup00/LogVol00 mounted on /
/dev/vda1 mounted on /boot
><fs> ll /
total 210

24 8

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

drwxr-xr-x. 24 root root 4096

drwxr-xr-x 21 root root 4096
drwxr-xr-x. 2 root root 4096
drwxr-xr-x. 4 root root 1024
drwxr-xr-x. 4 root root 4096
drwxr-xr-x. 86 root root 12288
[etc]

Oct
Nov
Oct
Oct
Oct
Oct

28
17
27
27
27
28

09:09 .
15:10 ..
22:37 bin
21:52 boot
21:21 dev
09:09 etc

Because guestfish needs to start up the libguestfs back end in order to perform the inspection and
mounting, the run command is not necessary when using the -i option. The -i option works for
many common Linux and Windows guest virtual machines.

1 6 .4 .1 .3. Acce ssing a gue st virt ual m achine by nam e

A guest virtual machine can be accessed from the command line when you specify its name as
known to libvirt (in other words, as it appears in vi rsh l i st --al l ). Use the -d option to access
a guest virtual machine by its name, with or without the -i option:
guestfish --ro -d GuestName -i

16.4 .2. Modifying Files wit h guest fish

To modify files, create directories or make other changes to a guest virtual machine, first heed the
warning at the beginning of this section: your guest virtual machine must be shut down. Editing or
changing a running disk with guestfish will result in disk corruption. This section gives an example
of editing the /bo o t/g rub/g rub. co nf file. When you are sure the guest virtual machine is shut
down you can omit the --ro option in order to get write access via a command such as:
guestfish -d RHEL3 -i
Welcome to guestfish, the libguestfs filesystem interactive shell for
editing virtual machine filesystems.
Type: 'help' for help on commands
'man' to read the manual
'quit' to quit the shell
Operating system: Red Hat Enterprise Linux AS release 3 (Taroon Update
9)
/dev/vda2 mounted on /
/dev/vda1 mounted on /boot
><fs> edit /boot/grub/grub.conf
Commands to edit files include ed i t, vi and emacs. Many commands also exist for creating files
and directories, such as wri te, mkd i r, upl o ad and tar-i n.

16.4 .3. Ot her Act ions wit h guest fish

You can also format file systems, create partitions, create and resize LVM logical volumes and much
more, with commands such as mkfs, part-ad d , l vresi ze, l vcreate, vg create and pvcreate.

16.4 .4 . Shell Script ing wit h guest fish

24 9

Virt ualiz at ion Administ rat ion G uide

Once you are familiar with using guestfish interactively, according to your needs, writing shell scripts
with it may be useful. The following is a simple shell script to add a new MOTD (message of the day)
to a guest:
#!/bin/bash set -e
guestname="$1"
guestfish -d "$guestname" -i <<'EOF'
write /etc/motd "Welcome to Acme Incorporated."
chmod 0644 /etc/motd
EOF

16.4 .5. Augeas and libguest fs Script ing

Combining libguestfs with Augeas can help when writing scripts to manipulate Linux guest virtual
machine configuration. For example, the following script uses Augeas to parse the keyboard
configuration of a guest virtual machine, and to print out the layout. Note that this example only
works with guest virtual machines running Red Hat Enterprise Linux:
#!/bin/bash set -e
guestname="$1"
guestfish -d "$1" -i --ro <<'EOF'
aug-init / 0
aug-get /files/etc/sysconfig/keyboard/LAYOUT
EOF
Augeas can also be used to modify configuration files. You can modify the above script to ch an g e
the keyboard layout:
#!/bin/bash set -e
guestname="$1"
guestfish -d "$1" -i <<'EOF'
aug-init / 0
aug-set /files/etc/sysconfig/keyboard/LAYOUT '"gb"'
aug-save
EOF
Note the three changes between the two scripts:
1. The --ro option has been removed in the second example, giving the ability to write to the
guest virtual machine.
2. The aug -g et command has been changed to aug -set to modify the value instead of
fetching it. The new value will be "g b" (including the quotes).
3. The aug -save command is used here so Augeas will write the changes out to disk.

250

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

Note
More information about Augeas can be found on the website http://augeas.net.
guestfish can do much more than we can cover in this introductory document. For example, creating
disk images from scratch:
guestfish -N fs
Or copying out whole directories from a disk image:
><fs> copy-out /home /tmp/home
For more information see the man page guestfish(1).

16.5. Ot her Commands

This section describes tools that are simpler equivalents to using guestfish to view and edit guest
virtual machine disk images.
vi rt-cat is similar to the guestfish d o wnl o ad command. It downloads and displays a single
file to the guest virtual machine. For example:
# virt-cat RHEL3 /etc/ntp.conf | grep ^server
server
127.127.1.0
# local clock
vi rt-ed i t is similar to the guestfish ed i t command. It can be used to interactively edit a single
file within a guest virtual machine. For example, you may need to edit the g rub. co nf file in a
Linux-based guest virtual machine that will not boot:
# virt-edit LinuxGuest /boot/grub/grub.conf
vi rt-ed i t has another mode where it can be used to make simple non-interactive changes to a
single file. For this, the - e option is used. This command, for example, changes the root password
in a Linux guest virtual machine to having no password:
# virt-edit LinuxGuest /etc/passwd -e 's/^root:.*?:/root::/'
vi rt-l s is similar to the guestfish l s, l l and fi nd commands. It is used to list a directory or
directories (recursively). For example, the following command would recursively list files and
directories under /home in a Linux guest virtual machine:
# virt-ls -R LinuxGuest /home/ | less

16.6. virt -rescue: T he Rescue Shell

This section provides information about using the rescue shell.

16.6.1. Int roduct ion

251

Virt ualiz at ion Administ rat ion G uide

This section describes vi rt-rescue, which can be considered analogous to a rescue CD for virtual
machines. It boots a guest virtual machine into a rescue shell so that maintenance can be performed
to correct errors and the guest virtual machine can be repaired.
There is some overlap between virt-rescue and guestfish. It is important to distinguish their differing
uses. virt-rescue is for making interactive, ad-hoc changes using ordinary Linux file system tools. It is
particularly suited to rescuing a guest virtual machine that has failed . virt-rescue cannot be scripted.
In contrast, guestfish is particularly useful for making scripted, structured changes through a formal
set of commands (the libguestfs API), although it can also be used interactively.

16.6.2. Running virt -rescue

Before you use vi rt-rescue on a guest virtual machine, make sure the guest virtual machine is not
running, otherwise disk corruption will occur. When you are sure the guest virtual machine is not live,
enter:
virt-rescue GuestName
(where GuestName is the guest name as known to libvirt), or:
virt-rescue /path/to/disk/image
(where the path can be any file, any logical volume, LUN, or so on) containing a guest virtual
machine disk.
You will first see output scroll past, as virt-rescue boots the rescue VM. In the end you will see:
Welcome to virt-rescue, the libguestfs rescue shell.
Note: The contents of / are the rescue appliance.
You have to mount the guest virtual machine's partitions under /sysroot
before you can examine them.
bash: cannot set terminal process group (-1): Inappropriate ioctl for
device
bash: no job control in this shell
><rescue>
The shell prompt here is an ordinary bash shell, and a reduced set of ordinary Red Hat Enterprise
Linux commands is available. For example, you can enter:
><rescue> fdisk -l /dev/vda
The previous command will list disk partitions. To mount a file system, it is suggested that you mount
it under /sysro o t, which is an empty directory in the rescue machine for the user to mount anything
you like. Note that the files under / are files from the rescue VM itself:
><rescue> mount /dev/vda1 /sysroot/
EXT4-fs (vda1): mounted filesystem with ordered data mode. Opts: (null)
><rescue> ls -l /sysroot/grub/
total 324
-rw-r--r--. 1 root root
63 Sep 16 18:14 device.map
-rw-r--r--. 1 root root 13200 Sep 16 18:14 e2fs_stage1_5

252

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

-rw-r--r--. 1 root root

-rw-r--r--. 1 root root
-rw-------. 1 root root
[...]

12512 Sep 16 18:14 fat_stage1_5

11744 Sep 16 18:14 ffs_stage1_5
1503 Oct 15 11:19 grub.conf

When you are finished rescuing the guest virtual machine, exit the shell by entering exi t or C trl + d .
vi rt-rescue has many command line options. The options most often used are:
- - ro : Operate in read-only mode on the guest virtual machine. No changes will be saved. You can
use this to experiment with the guest virtual machine. As soon as you exit from the shell, all of your
changes are discarded.
- - n et wo rk: Enable network access from the rescue shell. Use this if you need to, for example,
download RPM or other files into the guest virtual machine.

16.7. virt -df: Monit oring Disk Usage

This section provides information about monitoring disk usage using vi rt-d f.

16.7.1. Int roduct ion

This section describes vi rt-d f, which displays file system usage from a disk image or a guest
virtual machine. It is similar to the Linux d f command, but for virtual machines.

16.7.2. Running virt -df

To display file system usage for all file systems found in a disk image, enter the following:
# virt-df /dev/vg_guests/RHEL6
Filesystem
1K-blocks
RHEL6:/dev/sda1
101086
RHEL6:/dev/VolGroup00/LogVol00 7127864

Used
10233
2272744

Available
85634
4493036

Use%
11%
32%

(Where /d ev/vg _g uests/R HEL6 is a Red Hat Enterprise Linux 6 guest virtual machine disk image.
The path in this case is the host physical machine logical volume where this disk image is located.)
You can also use vi rt-d f on its own to list information about all of your guest virtual machines (ie.
those known to libvirt). The vi rt-d f command recognizes some of the same options as the
standard d f such as -h (human-readable) and -i (show inodes instead of blocks).
vi rt-d f also works on Windows guest virtual machines:
# virt-df -h
Filesystem
Size
F14x64:/dev/sda1
484.2M
F14x64:/dev/vg_f14x64/lv_root
7.4G
RHEL6brewx64:/dev/sda1
484.2M
RHEL6brewx64:/dev/vg_rhel6brewx64/lv_root
13.3G
Win7x32:/dev/sda1
100.0M
Win7x32:/dev/sda2
19.9G
7.4G

Used
66.3M
3.0G
52.6M
3.4G
24.1M
12.5G

Available
392.9M
4.4G
406.6M

Use%
14%
41%
11%

9.2G
75.9M
38%

26%
25%

253

Virt ualiz at ion Administ rat ion G uide

Note
You can use vi rt-d f safely on live guest virtual machines, since it only needs read-only
access. However, you should not expect the numbers to be precisely the same as those from a
d f command running inside the guest virtual machine. This is because what is on disk will be
slightly out of synch with the state of the live guest virtual machine. Nevertheless it should be a
good enough approximation for analysis and monitoring purposes.
virt-df is designed to allow you to integrate the statistics into monitoring tools, databases and so on.
This allows system administrators to generate reports on trends in disk usage, and alerts if a guest
virtual machine is about to run out of disk space. To do this you should use the --csv option to
generate machine-readable Comma-Separated-Values (CSV) output. CSV output is readable by most
databases, spreadsheet software and a variety of other tools and programming languages. The raw
CSV looks like the following:
# virt-df --csv WindowsGuest
Virtual Machine,Filesystem,1K-blocks,Used,Available,Use%
Win7x32,/dev/sda1,102396,24712,77684,24.1%
Win7x32,/dev/sda2,20866940,7786652,13080288,37.3%
For resources and ideas on how to process this output to produce trends and alerts, refer to the
following URL: http://virt-tools.org/learning/advanced-virt-df/.

16.8. virt -resiz e: Resiz ing Guest Virt ual Machines Offline
This section provides information about resizing offline guest virtual machines.

16.8.1. Int roduct ion

This section describes vi rt-resi ze, a tool for expanding or shrinking guest virtual machines. It
only works for guest virtual machines which are offline (shut down). It works by copying the guest
virtual machine image and leaving the original disk image untouched. This is ideal because you can
use the original image as a backup, however there is a trade-off as you need twice the amount of disk
space.

16.8.2. Expanding a Disk Image

This section demonstrates a simple case of expanding a disk image:
1. Locate the disk image to be resized. You can use the command vi rsh d umpxml
G uestName for a libvirt guest virtual machine.
2. D ecide on how you wish to expand the guest virtual machine. Run vi rt-d f -h and vi rtl i st-parti ti o ns -l h on the guest virtual machine disk, as shown in the following
output:
# virt-df -h /dev/vg_guests/RHEL6
Filesystem
Size
RHEL6:/dev/sda1
98.7M
RHEL6:/dev/VolGroup00/LogVol00 6.8G

254

Used
10.0M
2.2G

Available
83.6M
4.3G

Use%
11%
32%

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

# virt-list-partitions -lh /dev/vg_guests/RHEL6

/dev/sda1 ext3 101.9M
/dev/sda2 pv 7.9G
This example will demonstrate how to:
Increase the size of the first (boot) partition, from approximately 100MB to 500MB.
Increase the total disk size from 8GB to 16GB.
Expand the second partition to fill the remaining space.
Expand /d ev/Vo l G ro up0 0 /Lo g Vo l 0 0 to fill the new space in the second partition.
1. Make sure the guest virtual machine is shut down.
2. Rename the original disk as the backup. How you do this depends on the host physical
machine storage environment for the original disk. If it is stored as a file, use the mv
command. For logical volumes (as demonstrated in this example), use l vrename:
# lvrename /dev/vg_guests/RHEL6 /dev/vg_guests/RHEL6.backup
3. Create the new disk. The requirements in this example are to expand the total disk size up to
16GB. Since logical volumes are used here, the following command is used:
# lvcreate -L 16G -n RHEL6 /dev/vg_guests
Logical volume "RHEL6" created
4. The requirements from step 2 are expressed by this command:
# virt-resize \
/dev/vg_guests/RHEL6.backup /dev/vg_guests/RHEL6 \
--resize /dev/sda1=500M \
--expand /dev/sda2 \
--LV-expand /dev/VolGroup00/LogVol00
The first two arguments are the input disk and output disk. --resi ze /d ev/sd a1= 50 0 M
resizes the first partition up to 500MB. --expand /d ev/sd a2 expands the second partition
to fill all remaining space. --LV-expand /d ev/Vo l G ro up0 0 /Lo g Vo l 0 0 expands the
guest virtual machine logical volume to fill the extra space in the second partition.
vi rt-resi ze describes what it is doing in the output:
Summary of changes:
/dev/sda1: partition will be resized from 101.9M to 500.0M
/dev/sda1: content will be expanded using the 'resize2fs' method
/dev/sda2: partition will be resized from 7.9G to 15.5G
/dev/sda2: content will be expanded using the 'pvresize' method
/dev/VolGroup00/LogVol00: LV will be expanded to maximum size
/dev/VolGroup00/LogVol00: content will be expanded using the
'resize2fs' method
Copying /dev/sda1 ...
[#####################################################]
Copying /dev/sda2 ...

255

Virt ualiz at ion Administ rat ion G uide

[#####################################################]
Expanding /dev/sda1 using the 'resize2fs' method
Expanding /dev/sda2 using the 'pvresize' method
Expanding /dev/VolGroup00/LogVol00 using the 'resize2fs' method
5. Try to boot the virtual machine. If it works (and after testing it thoroughly) you can delete the
backup disk. If it fails, shut down the virtual machine, delete the new disk, and rename the
backup disk back to its original name.
6. Use vi rt-d f and/or vi rt-l i st-parti ti o ns to show the new size:
# virt-df -h /dev/vg_pin/RHEL6
Filesystem
Size
RHEL6:/dev/sda1
484.4M
RHEL6:/dev/VolGroup00/LogVol00 14.3G

Used
10.8M
2.2G

Available
448.6M
11.4G

Use%
3%
16%

Resizing guest virtual machines is not an exact science. If vi rt-resi ze fails, there are a number of
tips that you can review and attempt in the virt-resize(1) man page. For some older Red Hat Enterprise
Linux guest virtual machines, you may need to pay particular attention to the tip regarding GRUB.

16.9. virt -inspect or: Inspect ing Guest Virt ual Machines
This section provides information about inspecting guest virtual machines using vi rt-i nspecto r.

16.9.1. Int roduct ion

vi rt-i nspecto r is a tool for inspecting a disk image to find out what operating system it contains.

Note
Red Hat Enterprise Linux 6.2 ships with two variations of this program: vi rt-i nspecto r is
the original program as found in Red Hat Enterprise Linux 6.0 and is now deprecated
upstream. vi rt-i nspecto r2 is the same as the new upstream vi rt-i nspecto r program.

16.9.2. Inst allat ion

To install virt-inspector and the documentation, enter the following command:
# yum install libguestfs-tools libguestfs-devel
To process Windows guest virtual machines you must also install libguestfs-winsupport. Refer to
Section 16.10.2, Installation for details. The documentation, including example XML output and a
Relax-NG schema for the output, will be installed in /usr/share/d o c/l i bg uestfs-d evel -*/
where " *" is replaced by the version number of libguestfs.

16.9.3. Running virt -inspect or

You can run vi rt-i nspecto r against any disk image or libvirt guest virtual machine as shown in
the following example:
virt-inspector --xml disk.img > report.xml

256

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

Or as shown here:
virt-inspector --xml GuestName > report.xml
The result will be an XML report (repo rt. xml ). The main components of the XML file are a top-level
<operatingsytems> element containing usually a single <operatingsystem> element, similar to the
following:
<operatingsystems>
<operatingsystem>
<!-- the type of operating system and Linux distribution -->
<name>linux</name>
<distro>rhel</distro>
<!-- the name, version and architecture -->
<product_name>Red Hat Enterprise Linux Server release 6.4
</product_name>
<major_version>6</major_version>
<minor_version>4</minor_version>
<package_format>rpm</package_format>
<package_management>yum</package_management>
<root>/dev/VolGroup/lv_root</root>
<!-- how the filesystems would be mounted when live -->
<mountpoints>
<mountpoint dev="/dev/VolGroup/lv_root">/</mountpoint>
<mountpoint dev="/dev/sda1">/boot</mountpoint>
<mountpoint dev="/dev/VolGroup/lv_swap">swap</mountpoint>
</mountpoints>
< !-- filesystems-->
<filesystem dev="/dev/VolGroup/lv_root">
<label></label>
<uuid>b24d9161-5613-4ab8-8649-f27a8a8068d3</uuid>
<type>ext4</type>
<content>linux-root</content>
<spec>/dev/mapper/VolGroup-lv_root</spec>
</filesystem>
<filesystem dev="/dev/VolGroup/lv_swap">
<type>swap</type>
<spec>/dev/mapper/VolGroup-lv_swap</spec>
</filesystem>
<!-- packages installed -->
<applications>
<application>
<name>firefox</name>
<version>3.5.5</version>
<release>1.fc12</release>
</application>
</applications>
</operatingsystem>
</operatingsystems>

257

Virt ualiz at ion Administ rat ion G uide

Processing these reports is best done using W3C standard XPath queries. Red Hat Enterprise Linux
6 comes with a command line program (xpath) which can be used for simple instances; however, for
long-term and advanced usage, you should consider using an XPath library along with your favorite
programming language.
As an example, you can list out all file system devices using the following XPath query:
virt-inspector --xml GuestName | xpath //filesystem/@ dev
Found 3 nodes:
-- NODE -dev="/dev/sda1"
-- NODE -dev="/dev/vg_f12x64/lv_root"
-- NODE -dev="/dev/vg_f12x64/lv_swap"
Or list the names of all applications installed by entering:
virt-inspector --xml GuestName | xpath //application/name
[...long list...]

16.10. virt -win-reg: Reading and Edit ing t he Windows Regist ry

16.10.1. Int roduct ion
vi rt-wi n-reg is a tool that manipulates the Registry in Windows guest virtual machines. It can be
used to read out registry keys. You can also use it to make changes to the Registry, but you must
n ever try to do this for live/running guest virtual machines, as it will result in disk corruption.

16.10.2. Inst allat ion

To use vi rt-wi n-reg you must run the following:
# yum install libguestfs-tools libguestfs-winsupport

16.10.3. Using virt -win-reg

To read out Registry keys, specify the name of the guest virtual machine (or its disk image) and the
name of the Registry key. You must use single quotes to surround the name of the desired key:
# virt-win-reg WindowsGuest \
'HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Uninstall'
\
| less
The output is in the standard text-based format used by . R EG files on Windows.

258

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

Note
Hex-quoting is used for strings because the format does not properly define a portable
encoding method for strings. This is the only way to ensure fidelity when transporting . R EG
files from one machine to another.
You can make hex-quoted strings printable by piping the output of vi rt-wi n-reg through
this simple Perl script:
perl -MEncode -pe's?hex\((\d+)\):(\S+)?
$t=$1;$_=$2;s,\,,,g;"str($t):\"".decode(utf16le=>pack("H*",$_))."\""
?eg'

To merge changes into the Windows Registry of an offline guest virtual machine, you must first
prepare a . R EG file. There is a great deal of documentation about doing this available here. When
you have prepared a . R EG file, enter the following:
# virt-win-reg --merge WindowsGuest input.reg
This will update the registry in the guest virtual machine.

16.11. Using t he API from Programming Languages

The libguestfs API can be used directly from the following languages in Red Hat Enterprise Linux 6.2:
C, C++, Perl, Python, Java, Ruby and OCaml.
To install C and C++ bindings, enter the following command:
# yum install libguestfs-devel
To install Perl bindings:
# yum install 'perl(Sys::Guestfs)'
To install Python bindings:
# yum install python-libguestfs
To install Java bindings:
# yum install libguestfs-java libguestfs-java-devel libguestfs-javadoc
To install Ruby bindings:
# yum install ruby-libguestfs
To install OCaml bindings:
# yum install ocaml-libguestfs ocaml-libguestfs-devel

259

Virt ualiz at ion Administ rat ion G uide

The binding for each language is essentially the same, but with minor syntactic changes. A C
statement:
guestfs_launch (g);
Would appear like the following in Perl:
$g->launch ()
Or like the following in OCaml:
g#launch ()
Only the API from C is detailed in this section.
In the C and C++ bindings, you must manually check for errors. In the other bindings, errors are
converted into exceptions; the additional error checks shown in the examples below are not
necessary for other languages, but conversely you may wish to add code to catch exceptions. Refer
to the following list for some points of interest regarding the architecture of the libguestfs API:
The libguestfs API is synchronous. Each call blocks until it has completed. If you want to make
calls asynchronously, you have to create a thread.
The libguestfs API is not thread safe: each handle should be used only from a single thread, or if
you want to share a handle between threads you should implement your own mutex to ensure that
two threads cannot execute commands on one handle at the same time.
You should not open multiple handles on the same disk image. It is permissible if all the handles
are read-only, but still not recommended.
You should not add a disk image for writing if anything else could be using that disk image (eg. a
live VM). D oing this will cause disk corruption.
Opening a read-only handle on a disk image which is currently in use (eg. by a live VM) is
possible; however, the results may be unpredictable or inconsistent particularly if the disk image
is being heavily written to at the time you are reading it.

16.11.1. Int eract ion wit h t he API via a C Program

Your C program should start by including the <guestfs.h> header file, and creating a handle:
#include <stdio.h>
#include <stdlib.h>
#include <guestfs.h>
int
main (int argc, char *argv[])
{
guestfs_h *g;
g = guestfs_create ();
if (g == NULL) {
perror ("failed to create libguestfs handle");
exit (EXIT_FAILURE);
}

260

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

/* ... */
guestfs_close (g);
exit (EXIT_SUCCESS);
}
Save this program to a file (test. c). Compile this program and run it with the following two
commands:
gcc -Wall test.c -o test -lguestfs
./test
At this stage it should print no output. The rest of this section demonstrates an example showing how
to extend this program to create a new disk image, partition it, format it with an ext4 file system, and
create some files in the file system. The disk image will be called d i sk. i mg and be created in the
current directory.
The outline of the program is:
Create the handle.
Add disk(s) to the handle.
Launch the libguestfs back end.
Create the partition, file system and files.
Close the handle and exit.
Here is the modified program:
#include
#include
#include
#include
#include
#include

<stdio.h>
<stdlib.h>
<string.h>
<fcntl.h>
<unistd.h>
<guestfs.h>

int
main (int argc, char *argv[])
{
guestfs_h *g;
size_t i;
g = guestfs_create ();
if (g == NULL) {
perror ("failed to create libguestfs handle");
exit (EXIT_FAILURE);
}
/* Create a raw-format sparse disk image, 512 MB in size. */
int fd = open ("disk.img", O_CREAT|O_WRONLY|O_TRUNC|O_NOCTTY, 0666);
if (fd == -1) {
perror ("disk.img");
exit (EXIT_FAILURE);
}

261

Virt ualiz at ion Administ rat ion G uide

if (ftruncate (fd, 512 * 1024 * 1024) == -1) {

perror ("disk.img: truncate");
exit (EXIT_FAILURE);
}
if (close (fd) == -1) {
perror ("disk.img: close");
exit (EXIT_FAILURE);
}
/* Set the trace flag so that we can see each libguestfs call. */
guestfs_set_trace (g, 1);
/* Set the autosync flag so that the disk will be synchronized
* automatically when the libguestfs handle is closed.
*/
guestfs_set_autosync (g, 1);
/* Add the disk image to libguestfs. */
if (guestfs_add_drive_opts (g, "disk.img",
GUESTFS_ADD_DRIVE_OPTS_FORMAT, "raw", /* raw format */
GUESTFS_ADD_DRIVE_OPTS_READONLY, 0,
/* for write */
-1 /* this marks end of optional arguments */ )
== -1)
exit (EXIT_FAILURE);
/* Run the libguestfs back-end. */
if (guestfs_launch (g) == -1)
exit (EXIT_FAILURE);
/* Get the list of devices. Because we only added one drive
* above, we expect that this list should contain a single
* element.
*/
char **devices = guestfs_list_devices (g);
if (devices == NULL)
exit (EXIT_FAILURE);
if (devices[0] == NULL || devices[1] != NULL) {
fprintf (stderr,
"error: expected a single device from list-devices\n");
exit (EXIT_FAILURE);
}
/* Partition the disk as one single MBR partition. */
if (guestfs_part_disk (g, devices[0], "mbr") == -1)
exit (EXIT_FAILURE);
/* Get the list of partitions. We expect a single element, which
* is the partition we have just created.
*/
char **partitions = guestfs_list_partitions (g);
if (partitions == NULL)
exit (EXIT_FAILURE);
if (partitions[0] == NULL || partitions[1] != NULL) {
fprintf (stderr,
"error: expected a single partition from listpartitions\n");

262

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

exit (EXIT_FAILURE);
}
/* Create an ext4 filesystem on the partition. */
if (guestfs_mkfs (g, "ext4", partitions[0]) == -1)
exit (EXIT_FAILURE);
/* Now mount the filesystem so that we can add files. */
if (guestfs_mount_options (g, "", partitions[0], "/") == -1)
exit (EXIT_FAILURE);
/* Create some files and directories. */
if (guestfs_touch (g, "/empty") == -1)
exit (EXIT_FAILURE);
const char *message = "Hello, world\n";
if (guestfs_write (g, "/hello", message, strlen (message)) == -1)
exit (EXIT_FAILURE);
if (guestfs_mkdir (g, "/foo") == -1)
exit (EXIT_FAILURE);
/* This uploads the local file /etc/resolv.conf into the disk image.
*/
if (guestfs_upload (g, "/etc/resolv.conf", "/foo/resolv.conf") == -1)
exit (EXIT_FAILURE);
/* Because 'autosync' was set (above) we can just close the handle
* and the disk contents will be synchronized. You can also do
* this manually by calling guestfs_umount_all and guestfs_sync.
*/
guestfs_close (g);
/* Free up the lists. */
for (i = 0; devices[i] != NULL; ++i)
free (devices[i]);
free (devices);
for (i = 0; partitions[i] != NULL; ++i)
free (partitions[i]);
free (partitions);
exit (EXIT_SUCCESS);
}
Compile and run this program with the following two commands:
gcc -Wall test.c -o test -lguestfs
./test
If the program runs to completion successfully then you should be left with a disk image called
d i sk. i mg , which you can examine with guestfish:
guestfish --ro -a disk.img -m /dev/sda1
><fs> ll /
><fs> cat /foo/resolv.conf

263

Virt ualiz at ion Administ rat ion G uide

By default (for C and C++ bindings only), libguestfs prints errors to stderr. You can change this
behavior by setting an error handler. The guestfs(3) man page discusses this in detail.

16.12. virt -sysprep: Reset t ing Virt ual Machine Set t ings
The virt-sysprep command line tool can be used to reset or unconfigure a guest virtual machine so
that clones can be made from it. This process involves removing SSH host keys, persistent network
MAC configuration, and user accounts. virt-sysprep can also customize a virtual machine, for
instance by adding SSH keys, users or logos. Each step can be enabled or disabled as required.
The term " sysprep" is derived from the System Preparation tool (sysprep.exe) which is used with the
Microsoft Windows systems. D espite this, the tool does not currently work on Windows guests.

Note
libguestfs and guestfish do not require root privileges. You only need to run them as root if the
disk image being accessed needs root access to read and/or write.
The virt-sysprep tool is part of the libguestfs-tools-c package, which is installed with the following
command:
$ yum i nstal l libguestfs-tools-c
Alternatively, just the virt-sysprep tool can be installed with the following command:
$ yum i nstal l /usr/bi n/vi rt-sysprep

Important
virt-sysprep modifies the guest or disk image in place. To use virt-sysprep, the guest virtual
machine must be offline, so you must shut it down before running the commands. To preserve
the existing contents of the guest virtual machine, you must snapshot, copy or clone the disk
first. Refer to libguestfs.org for more information on copying and cloning disks.
The following commands are available to use with virt-sysprep:
T ab le 16 .1. virt-sysprep co mman d s
C o mman d

D escrip t io n

Examp le

--help

D isplays a brief help entry

about a particular command or
about the whole package. For
additional help, see the virtsysprep man page.

$ vi rt-sysprep --hel p

264

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

C o mman d

D escrip t io n

Examp le

-a [file] or --add [file]

Adds the specified file, which

should be a disk image from a
guest virtual machine. The
format of the disk image is
auto-detected. To override this
and force a particular format,
use the --fo rmat option.
Connects to the given URI, if
using libvirt. If omitted, it will
connect via the KVM
hypervisor. If you specify guest
block devices directly (vi rtsysprep -a), then libvirt is not
used at all.
Adds all the disks from the
specified guest virtual machine.
D omain UUID s can be used
instead of domain names.
Performs a read-only " dry run"
sysprep operation on the guest
virtual machine. This runs the
sysprep operation, but throws
away any changes to the disk
at the end.
Enables the specified
operations. To list the possible
operations, use the --list
command.
The default for the -a option is
to auto-detect the format of the
disk image. Using this forces
the disk format for -a options
which follow on the command
line. Using --format auto
switches back to auto-detection
for subsequent -a options (see
the -a command above).

$ vi rt-sysprep --ad d
/d ev/vms/d i sk. i mg

-c [URI] or --connect [URI]

-d [guest] or --domain [guest]

-n or --dry-run or --dryrun

--enable [operations]

--format [raw|q co w2|auto ]

$ vi rt-sysprep -c
q emu: ///system

$ vi rt-sysprep --d o mai n

90df2f3f-8857-5ba92714-7d95907b1c9e
$ vi rt-sysprep -n

$ vi rt-sysprep --enabl e
ssh-ho tkeys,ud evpersi stent-net
$ vi rt-sysprep --fo rmat
raw -a disk.img forces raw
format (no auto-detection) for
disk.img, but vi rt-sysprep -fo rmat raw -a disk.img
--fo rmat auto -a
ano ther. i mg forces raw
format (no auto-detection) for
d i sk. i mg and reverts to autodetection for ano ther. i mg . If
you have untrusted raw-format
guest disk images, you should
use this option to specify the
disk format. This avoids a
possible security problem with
malicious guests.

265

Virt ualiz at ion Administ rat ion G uide

C o mman d

D escrip t io n

Examp le

--list-operations

Lists the operations supported

by the virt-sysprep program.
These are listed one per line,
with one or more single-spaceseparated fields. The first field
in the output is the operation
name, which can be supplied to
the --enabl e flag. The second
field is a * character if the
operation is enabled by default,
or is blank if not. Additional
fields on the same line include
a description of the operation.
Sets the mount options for each
mount point in the guest virtual
machine. Use a semicolonseparated list of
mountpoint:options pairs. You
may need to place quotes
around this list to protect it from
the shell.
virt-sysprep does not always
schedule a SELinux relabelling
at the first boot of the guest. In
some cases, a relabel is
performed (for example, when
virt-sysprep has modified files).
However, when all operations
only remove files (for example,
when using --enabl e
d el ete --d el ete
/so me/fi l e) no relabelling is
scheduled. Using the -sel i nux-rel abel option
always forces SELinux
relabelling, while with --no sel i nux-rel abel set,
relabelling is never scheduled.
It is recommended to use -sel i nux-rel abel to ensure
that files have the correct
SELinux labels.
Prevents the printing of log
messages.
Enables verbose messages for
debugging purposes.
D isplays the virt-sysprep
version number and exits.

$ vi rt-sysprep --l i sto perati o ns

--mount-options

--selinux-relabel and --noselinux-relabel

-q or --quiet
-v or --verbose
-V or --version

266

$ vi rt-sysprep --mo unto pti o ns "/: no ti me" will

mount the root directory with
the no ti me operation.

$ vi rt-sysprep -sel i nux-rel abel

$ vi rt-sysprep -q
$ vi rt-sysprep -v
$ vi rt-sysprep -V

Chapt er 1 6 . G uest Virt ual Machine Disk Access wit h O ffline T ools

C o mman d

D escrip t io n

Examp le

--root-password

Sets the root password. Can

either be used to specify the
new password explicitly, or to
use the string from the first line
of a selected file, which is more
secure.

$ vi rt-sysprep --ro o tpasswo rd passwo rd : 123456

-a guest.img
or
$ vi rt-sysprep --ro o tpasswo rd
fi l e: SOURCE_FILE_PATH -a
guest.img

For more information, refer to the libguestfs documentation.

16.13. T roubleshoot ing

A test tool is available to check that libguestfs is working. Run the following command after installing
libguestfs (root access not required) to test for normal operation:
$ libguestfs-test-tool
This tool prints a large amount of text to test the operation of libguestfs. If the test is successful, the
following text will appear near the end of the output:
===== TEST FINISHED OK =====

16.14 . Where t o Find Furt her Document at ion

The primary source for documentation for libguestfs and the tools are the Unix man pages. The API is
documented in guestfs(3). guestfish is documented in guestfish(1). The virt tools are documented in
their own man pages (eg. virt-df(1)).

267

Virt ualiz at ion Administ rat ion G uide

Chapter 17. Using Simple Tools for Guest Virtual Machine

Management
In addition to virt-manager, there are other smaller more minimal tools that can allow you to have
access to your guest virtual machine's console. The sections that follow describe and explain these
tools.

17.1. Using virt -viewer

virt-viewer is a minimal tool for displaying the graphical console of a guest virtual machine. The
console is accessed using the VNC or SPICE protocol. The guest can be referred to based on its
name, ID , or UUID . If the guest is not already running, then the viewer can be told to wait until is starts
before attempting to connect to the console. The viewer can connect to remote hosts to lookup the
console information and then also connect to the remote console using the same network transport.
To install the virt-viewer tool, run:
# sud o yum i nstal l vi rt-vi ewer
The basic virt viewer commands are as follows:
# vi rt-vi ewer [O P T IO NS] D O MAIN-NAME| ID | UUID
The following options may be used with virt-viewer
-h, or --hel p - D isplays the command line help summary.
-V, or --versi o n - D isplays the virt-viewer version number.
-v, or --verbo se - D isplays information about the connection to the guest virtual machine
-c [UR I], or --co nnect= UR I - Specifies the hypervisor connection URI
-w, or --wai t - Causes the domain to start up before attempting to connect to the console.
-r, or --reco nnect - Automatically reconnects to the domain if it shuts down and restarts
-z P C T , or --zo o m= P C T - Adjusts the zoom level of the display window in the specified
percentage. Accepted range 10-200% .
-a, or --attach - Uses libvirt to directly attach to a local display, instead of making a TCP/UNIX
socket connection. This avoids the need to authenticate with the remote display, if authentication
with libvirt is already allowed. This option does not work with remote displays.
-f, or --ful l -screen - Starts with the command window maximized to its fullscreen size.
-h HO T KEY S, or --ho tkeys HO T KEY S - Overrides the default hotkey settings with the new
specified hotkey. Refer to Example 17.4, Setting hot keys .
--d ebug - Prints debugging information

Examp le 17.1. C o n n ect in g t o a g u est virt u al mach in e

If using a XEN hypervisor:

268

Chapt er 1 7 . Using Simple T ools for G uest Virt ual Machine Management

# vi rt-vi ewer g uest-name

If using a KVM-QEMU hypervisor:
# vi rt-vi ewer --co nnect q emu: ///system 7

Examp le 17.2. T o wait f o r a sp ef ic g u est t o st art b ef o re co n n ect in g

Enter the following command:
# vi rt-vi ewer --reco nnect --wai t 6 6 ab33c0 -6 9 19 -a3f7-e6 59 -16 c82d 24 8521

Examp le 17.3. T o co n n ect t o a remo t e co n so le u sin g T LS

Enter the following command:
#vi rt-vi ewer --co nnect xen: //exampl e. o rg / d emo
To connect to a remote host using SSH, lookup the guest configuration and then make a direct
non-tunnelled connection to the console

Examp le 17.4 . Set t in g h o t keys

To create a customized hotkey, run the following command:
# virt-viewer --hotkeys=([action1]=[key-combination1]), ([action2]=[keycombination2])
The following actions can be assigned to a hotkey:
toggle-fullscreen
release-cursor
smartcard-insert
smartcard-remove
Key name combinations are case insensitive. Each hot key setting should have a unique key
combination.
For example, to create a hotkey to change to full screen mode:
#vi rt-vi ewer --ho tkeys= to g g l e-ful l screen= shi ft+ f11 q emu: ///system 7

17.2. remot e-viewer

The remo t e- viewer is a simple remote desktop display client that supports SPICE and VNC.
To install the virt-viewer tool, run:

269

Virt ualiz at ion Administ rat ion G uide

# sud o yum i nstal l remo te-vi ewer

The basic remote viewer commands are as follows:remo te-vi ewer [O P T IO NS] D O MAINNAME| ID | UUID
The following options may be used with remote-viewer
-h, or --hel p - D isplays the command line help summary.
-V, or --versi o n - D isplays the remote-viewer version number.
-v, or --verbo se - D isplays information about the connection to the guest virtual machine
-z P C T , or --zo o m= P C T - Adjusts the zoom level of the display window in the specified
percentage. Accepted range 10-200% .
-f, or --ful l -screen - Starts with the command window maximized to its fullscreen size.
-t T IT LE, or --ti tl e T IT LE - Sets the window title to the string given
--spi ce-co ntro l l er - Uses the SPICE controller to initialize the connection with the SPICE
server. This option is used by the SPICE browser addons to allow web page to start a client.
--d ebug - Prints debugging information
For more information see the man page for the remote-viewer.

270

Chapt er 1 8 . Virt ual Net working

Chapter 18. Virtual Networking

This chapter introduces the concepts needed to create, start, stop, remove, and modify virtual
networks with libvirt.
Additional information can be found in the libvirt reference chapter

18.1. Virt ual Net work Swit ches

Libvirt virtual networking uses the concept of a virtual network switch. A virtual network switch is a
software construct that operates on a host physical machine server, to which virtual machines
(guests) connect. The network traffic for a guest is directed through this switch:

Fig u re 18.1. Virt u al n et wo rk swit ch wit h t wo g u est s

Linux host physical machine servers represent a virtual network switch as a network interface. When
the libvirtd daemon (l i bvi rtd ) is first installed and started, the default network interface
representing the virtual network switch is vi rbr0 .

271

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.2. Lin u x h o st p h ysical mach in e wit h an in t erf ace t o a virt u al n et wo rk swit ch
This vi rbr0 interface can be viewed with the i fco nfi g and i p commands like any other interface:
$ ifconfig virbr0
virbr0
Link encap:Ethernet HWaddr 1B:C4:94:CF:FD:17
inet addr:192.168.122.1 Bcast:192.168.122.255
Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:11 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:0 (0.0 b) TX bytes:3097 (3.0 KiB)
$ ip addr show virbr0
3: virbr0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
state UNKNOWN
link/ether 1b:c4:94:cf:fd:17 brd ff:ff:ff:ff:ff:ff
inet 192.168.122.1/24 brd 192.168.122.255 scope global virbr0

18.2. Bridged Mode

When using Bridged mode, all of the guest virtual machines appear within the same subnet as the host
physical machine. All other physical machines on the same physical network are aware of the virtual
machines, and can access the virtual machines. Bridging operates on Layer 2 of the OSI networking
model.
It is possible to use multiple physical interfaces on the hypervisor by joining them together with a
bond. The bond is then added to a bridge and then guest virtual machines are added onto the
bridge as well. However, the bonding driver has several modes of operation, and only a few of these
modes work with a bridge where virtual guest machines are in use.

272

Chapt er 1 8 . Virt ual Net working

Fig u re 18.3. Virt u al n et wo rk swit ch in b rid g ed mo d e

Warning
The only bonding modes that should be used with a guest virtual machine are Mode 1, Mode
2, and Mode 4. Under no circumstances should Modes 0, 3, 5, or 6 be used. It should also be
noted that mii-monitoring should be used to monitor bonding modes as arp-monitoring does
not work.
For more information on bonding modes, refer to the knowledgebase article on bonding modes, or
The Red Hat Enterprise Linux 6 D eployment Guide.
For a detailed explanation of bridge_opts parameters, see the Red Hat Enterprise Virtualization
Administration Guide.

18.3. Net work Address T ranslat ion Mode

By default, virtual network switches operate in NAT mode. They use IP masquerading rather than
SNAT (Source-NAT) or D NAT (D estination-NAT). IP masquerading enables connected guests to use
the host physical machine IP address for communication to any external network. By default,
computers that are placed externally to the host physical machine cannot communicate to the guests
inside when the virtual network switch is operating in NAT mode, as shown in the following diagram:

273

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.4 . Virt u al n et wo rk swit ch u sin g N AT wit h t wo g u est s

Warning
Virtual network switches use NAT configured by iptables rules. Editing these rules while the
switch is running is not recommended, as incorrect rules may result in the switch being unable
to communicate.
If the switch is not running, you can set th public IP range for forward mode NAT in order to create a
port masquerading range by running:
# i ptabl es -j SNAT --to -so urce [start]-[end ]

18.3.1. DNS and DHCP

IP information can be assigned to guests via D HCP. A pool of addresses can be assigned to a
virtual network switch for this purpose. Libvirt uses the d nsmasq program for this. An instance of
dnsmasq is automatically configured and started by libvirt for each virtual network switch that needs
it.

274

Chapt er 1 8 . Virt ual Net working

Fig u re 18.5. Virt u al n et wo rk swit ch ru n n in g d n smasq

18.4 . Rout ed Mode

When using Routed mode, the virtual switch connects to the physical LAN connected to the host
physical machine, passing traffic back and forth without the use of NAT. The virtual switch can
examine all traffic and use the information contained within the network packets to make routing
decisions. When using this mode, all of the virtual machines are in their own subnet, routed through
a virtual switch. This situation is not always ideal as no other host physical machines on the
physical network are aware of the virtual machines without manual physical router configuration,
and cannot access the virtual machines. Routed mode operates at Layer 3 of the OSI networking
model.

275

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.6 . Virt u al n et wo rk swit ch in ro u t ed mo d e

18.5. Isolat ed Mode

When using Isolated mode, guests connected to the virtual switch can communicate with each other,
and with the host physical machine, but their traffic will not pass outside of the host physical
machine, nor can they receive traffic from outside the host physical machine. Using dnsmasq in this
mode is required for basic functionality such as D HCP. However, even if this network is isolated from
any physical network, D NS names are still resolved. Therefore a situation can arise when D NS
names resolve but ICMP echo request (ping) commands fail.

276

Chapt er 1 8 . Virt ual Net working

Fig u re 18.7. Virt u al n et wo rk swit ch in iso lat ed mo d e

18.6. T he Default Configurat ion

When the libvirtd daemon (l i bvi rtd ) is first installed, it contains an initial virtual network switch
configuration in NAT mode. This configuration is used so that installed guests can communicate to
the external network, through the host physical machine. The following image demonstrates this
default configuration for l i bvi rtd :

277

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.8. D ef au lt lib virt n et wo rk co n f ig u rat io n

Note
A virtual network can be restricted to a specific physical interface. This may be useful on a
physical system that has several interfaces (for example, eth0 , eth1 and eth2). This is only
useful in routed and NAT modes, and can be defined in the d ev= <i nterface> option, or in
vi rt-manag er when creating a new virtual network.

18.7. Examples of Common Scenarios

This section demonstrates different virtual networking modes and provides some example scenarios.

18.7.1. Bridged Mode

Bridged mode operates on Layer 2 of the OSI model. When used, all of the guest virtual machines will
appear on the same subnet as the host physical machine. The most common use cases for bridged
mode include:
D eploying guest virtual machines in an existing network alongside host physical machines
making the difference between virtual and physical machines transparent to the end user.
D eploying guest virtual machines without making any changes to existing physical network
configuration settings.

278

Chapt er 1 8 . Virt ual Net working

D eploying guest virtual machines which must be easily accessible to an existing physical
network. Placing guest virtual machines on a physical network where they must access services
within an existing broadcast domain, such as D HCP.
Connecting guest virtual machines to an exsting network where VLANs are used.

18.7.2. Rout ed Mode

This section provides information about routed mode.
D MZ
Consider a network where one or more nodes are placed in a controlled subnetwork for security
reasons. The deployment of a special subnetwork such as this is a common practice, and the
subnetwork is known as a D MZ . Refer to the following diagram for more details on this layout:

Fig u re 18.9 . Samp le D MZ co n f ig u rat io n

Host physical machines in a D MZ typically provide services to WAN (external) host physical
machines as well as LAN (internal) host physical machines. As this requires them to be accessible
from multiple locations, and considering that these locations are controlled and operated in different
ways based on their security and trust level, routed mode is the best configuration for this
environment.
Virt u al Server H o st in g
Consider a virtual server hosting company that has several host physical machines, each with two

279

Virt ualiz at ion Administ rat ion G uide

physical network connections. One interface is used for management and accounting, the other is for
the virtual machines to connect through. Each guest has its own public IP address, but the host
physical machines use private IP address as management of the guests can only be performed by
internal administrators. Refer to the following diagram to understand this scenario:

Fig u re 18.10. Virt u al server h o st in g samp le co n f ig u rat io n

When the host physical machine has a public IP address and the virtual machines have static public
IP addresses, bridged networking cannot be used, as the provider only accepts packets from the
MAC address of the public host physical machine. The following diagram demonstrates this:

Fig u re 18.11. Virt u al server u sin g st at ic IP ad d resses

18.7.3. NAT Mode

280

Chapt er 1 8 . Virt ual Net working

NAT (Network Address Translation) mode is the default mode. It can be used for testing when there is
no need for direct network visibility.

18.7.4 . Isolat ed Mode

Isolated mode allows virtual machines to communicate with each other only. They are unable to
interact with the physical network.

18.8. Managing a Virt ual Net work

To configure a virtual network on your system:
1. From the Ed i t menu, select C o nnecti o n D etai l s.

Fig u re 18.12. Select in g a h o st p h ysical mach in e' s d et ails

2. This will open the C o n n ect io n D et ails menu. Click the Vi rtual Netwo rks tab.

281

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.13. Virt u al n et wo rk co n f ig u rat io n

3. All available virtual networks are listed on the left-hand box of the menu. You can edit the
configuration of a virtual network by selecting it from this box and editing as you see fit.

18.9. Creat ing a Virt ual Net work

To create a virtual network on your system:
1. Open the Vi rtual Netwo rks tab from within the C o nnecti o n D etai l s menu. Click the
Ad d Netwo rk button, identified by a plus sign (+) icon. For more information, refer to
Section 18.8, Managing a Virtual Network .

282

Chapt er 1 8 . Virt ual Net working

Fig u re 18.14 . Virt u al n et wo rk co n f ig u rat io n

This will open the C reat e a n ew virt u al n et wo rk window. Click Fo rward to continue.

283

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.15. C reat in g a n ew virt u al n et wo rk

2. Enter an appropriate name for your virtual network and click Fo rward .

284

Chapt er 1 8 . Virt ual Net working

Fig u re 18.16 . N amin g yo u r virt u al n et wo rk

3. Enter an IPv4 address space for your virtual network and click Fo rward .

285

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.17. C h o o sin g an IPv4 ad d ress sp ace

4. D efine the D HCP range for your virtual network by specifying a Start and End range of IP
addresses. Click Fo rward to continue.

286

Chapt er 1 8 . Virt ual Net working

Fig u re 18.18. Select in g t h e D H C P ran g e

5. Select how the virtual network should connect to the physical network.

287

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.19 . C o n n ect in g t o p h ysical n et wo rk

If you select Fo rward i ng to physi cal netwo rk, choose whether the D esti nati o n
should be Any physi cal d evi ce or a specific physical device. Also select whether the
Mo d e should be NAT or R o uted .
Click Fo rward to continue.
6. You are now ready to create the network. Check the configuration of your network and click
Fi ni sh.

288

Chapt er 1 8 . Virt ual Net working

Fig u re 18.20. R ead y t o creat e n et wo rk

7. The new virtual network is now available in the Vi rtual Netwo rks tab of the C o nnecti o n
D etai l s window.

18.10. At t aching a Virt ual Net work t o a Guest

To attach a virtual network to a guest:
1. In the Vi rtual Machi ne Manag er window, highlight the guest that will have the network
assigned.

289

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.21. Select in g a virt u al mach in e t o d isp lay

2. From the Virtual Machine Manager Ed i t menu, select Vi rtual Machi ne D etai l s.

Fig u re 18.22. D isp layin g t h e virt u al mach in e d et ails

3. Click the Ad d Hard ware button on the Virtual Machine D etails window.

290

Chapt er 1 8 . Virt ual Net working

Fig u re 18.23. T h e Virt u al Mach in e D et ails win d o w

4. In the Ad d new vi rtual hard ware window, select Netwo rk from the left pane, and select
your network name (network1 in this example) from the Ho st d evi ce menu and click
Fi ni sh.

291

Virt ualiz at ion Administ rat ion G uide

Fig u re 18.24 . Select yo u r n et wo rk f ro m t h e Ad d n ew virt u al h ard ware win d o w

5. The new network is now displayed as a virtual network interface that will be presented to the
guest upon launch.

292

Chapt er 1 8 . Virt ual Net working

Fig u re 18.25. N ew n et wo rk sh o wn in g u est h ard ware list

18.11. At t aching Direct ly t o a Physical Int erface

The instructions provided in this chapter will assist in the direct attachment of the virtual machine's
NIC to the given physical interface of the host physical machine. This setup requires the Linux
macvtap driver to be available. There are four modes that you can choose for the operation mode of
the macvtap device, with 'vepa' being the default mode. Their behavior is as follows:
Ph ysical in t erf ace d elivery mo d es
vep a
All VMs' packets are sent to the external bridge. Packets whose destination is a VM on the
same host physical machine as where the packet originates from are sent back to the host
physical machine by the VEPA capable bridge (today's bridges are typically not VEPA
capable).
b rid g e
Packets whose destination is on the same host physical machine as where they originate
from are directly delivered to the target macvtap device. Both origin and destination devices
need to be in bridged mode for direct delivery. If either one of them is in vepa mode, a VEPA
capable bridge is required.

293

Virt ualiz at ion Administ rat ion G uide

p rivat e
All packets are sent to the external bridge and will only be delivered to a target VM on the
same host physical machine if they are sent through an external router or gateway and that
device sends them back to the host physical machine. This procedure is followed if either
the source or destination device is in private mode.
p asst h ro u g h
This feature attaches a virtual function of a SRIOV capable NIC directly to a VM without
losing the migration capability. All packets are sent to the VF/IF of the configured network
device. D epending on the capabilities of the device additional prerequisites or limitations
may apply; for example, on Linux this requires kernel 2.6.38 or newer.
Each of the four modes is configured by changing the domain xml file. Once this file is opened,
change the mode setting as shown:
<devices>
...
<interface type='direct'>
<source dev='eth0' mode='vepa'/>
</interface>
</devices>
The network access of direct attached guest virtual machines can be managed by the hardware
switch to which the physical interface of the host physical machine is connected to.
The interface can have additional parameters as shown below, if the switch is conforming to the IEEE
802.1Qbg standard. The parameters of the virtualport element are documented in more detail in the
IEEE 802.1Qbg standard. The values are network specific and should be provided by the network
administrator. In 802.1Qbg terms, the Virtual Station Interface (VSI) represents the virtual interface of
a virtual machine.
Note that IEEE 802.1Qbg requires a non-zero value for the VLAN ID . Also if the switch is conforming
to the IEEE 802.1Qbh standard, the values are network specific and should be provided by the
network administrator.
Virt u al St at io n In t erf ace t yp es
man ag erid
The VSI Manager ID identifies the database containing the VSI type and instance
definitions. This is an integer value and the value 0 is reserved.
t yp eid
The VSI Type ID identifies a VSI type characterizing the network access. VSI types are
typically managed by network administrator. This is an integer value.
t yp eid versio n
The VSI Type Version allows multiple versions of a VSI Type. This is an integer value.
in st an ceid
The VSI Instance ID Identifier is generated when a VSI instance (i.e. a virtual interface of a
virtual machine) is created. This is a globally unique identifier.
p ro f ileid

294

Chapt er 1 8 . Virt ual Net working

The profile ID contains the name of the port profile that is to be applied onto this interface.
This name is resolved by the port profile database into the network parameters from the port
profile, and those network parameters will be applied to this interface.
Each of the four types is configured by changing the domain xml file. Once this file is opened,
change the mode setting as shown:
<devices>
...
<interface type='direct'>
<source dev='eth0.2' mode='vepa'/>
<virtualport type="802.1Qbg">
<parameters managerid="11" typeid="1193047" typeidversion="2"
instanceid="09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f"/>
</virtualport>
</interface>
</devices>
The profile ID is shown here:
<devices>
...
<interface type='direct'>
<source dev='eth0' mode='private'/>
<virtualport type='802.1Qbh'>
<parameters profileid='finance'/>
</virtualport>
</interface>
</devices>
...

18.12. Applying Net work Filt ering

This section provides an introduction to libvirt's network filters, their goals, concepts and XML format.

18.12.1. Int roduct ion

The goal of the network filtering, is to enable administrators of a virtualized system to configure and
enforce network traffic filtering rules on virtual machines and manage the parameters of network
traffic that virtual machines are allowed to send or receive. The network traffic filtering rules are
applied on the host physical machine when a virtual machine is started. Since the filtering rules
cannot be circumvented from within the virtual machine, it makes them mandatory from the point of
view of a virtual machine user.
From the point of view of the guest virtual machine, the network filtering system allows each virtual
machine's network traffic filtering rules to be configured individually on a per interface basis. These
rules are applied on the host physical machine when the virtual machine is started and can be
modified while the virtual machine is running. The latter can be achieved by modifying the XML
description of a network filter.
Multiple virtual machines can make use of the same generic network filter. When such a filter is
modified, the network traffic filtering rules of all running virtual machines that reference this filter are
updated. The machines that are not running will update on start.

295

Virt ualiz at ion Administ rat ion G uide

As previously mentioned, applying network traffic filtering rules can be done on individual network
interfaces that are configured for certain types of network configurations. Supported network types
include:
network
ethernet -- must be used in bridging mode
bridge

Examp le 18.1. An examp le o f n et wo rk f ilt erin g

The interface XML is used to reference a top-level filter. In the following example, the interface
description references the filter clean-traffic.
<devices>
<interface type='bridge'>
<mac address='00:16:3e:5d:c7:9e'/>
<filterref filter='clean-traffic'/>
</interface>
</devices>
Network filters are written in XML and may either contain: references to other filters, rules for traffic
filtering, or hold a combination of both. The above referenced filter clean-traffic is a filter that only
contains references to other filters and no actual filtering rules. Since references to other filters can
be used, a tree of filters can be built. The clean-traffic filter can be viewed using the command: #
vi rsh nwfi l ter-d umpxml cl ean-traffi c.
As previously mentioned, a single network filter can be referenced by multiple virtual machines.
Since interfaces will typically have individual parameters associated with their respective traffic
filtering rules, the rules described in a filter's XML can be generalized using variables. In this case,
the variable name is used in the filter XML and the name and value are provided at the place where
the filter is referenced.

Examp le 18.2. D escrip t io n ext en d ed

In the following example, the interface description has been extended with the parameter IP and a
dotted IP address as a value.
<devices>
<interface type='bridge'>
<mac address='00:16:3e:5d:c7:9e'/>
<filterref filter='clean-traffic'>
<parameter name='IP' value='10.0.0.1'/>
</filterref>
</interface>
</devices>
In this particular example, the clean-traffic network traffic filter will be represented with the IP
address parameter 10.0.0.1 and as per the rule dictates that all traffic from this interface will always
be using 10.0.0.1 as the source IP address, which is one of the purpose of this particular filter.

18.12.2. Filt ering Chains

296

Chapt er 1 8 . Virt ual Net working

Filtering rules are organized in filter chains. These chains can be thought of as having a tree
structure with packet filtering rules as entries in individual chains (branches).
Packets start their filter evaluation in the root chain and can then continue their evaluation in other
chains, return from those chains back into the root chain or be dropped or accepted by a filtering
rule in one of the traversed chains.
Libvirt's network filtering system automatically creates individual root chains for every virtual
machine's network interface on which the user chooses to activate traffic filtering. The user may write
filtering rules that are either directly instantiated in the root chain or may create protocol-specific
filtering chains for efficient evaluation of protocol-specific rules.
The following chains exist:
root
mac
stp (spanning tree protocol)
vlan
arp and rarp
ipv4
ipv6
Multiple chains evaluating the mac, stp, vlan, arp, rarp, ipv4, or ipv6 protocol can be created using
the protocol name only as a prefix in the chain's name.

Examp le 18.3. AR P t raf f ic f ilt erin g

This example allows chains with names arp-xyz or arp-test to be specified and have their ARP
protocol packets evaluated in those chains.
The following filter XML shows an example of filtering ARP traffic in the arp chain.
<filter name='no-arp-spoofing' chain='arp' priority='-500'>
<uuid>f88f1932-debf-4aa1-9fbe-f10d3aa4bc95</uuid>
<rule action='drop' direction='out' priority='300'>
<mac match='no' srcmacaddr='$MAC'/>
</rule>
<rule action='drop' direction='out' priority='350'>
<arp match='no' arpsrcmacaddr='$MAC'/>
</rule>
<rule action='drop' direction='out' priority='400'>
<arp match='no' arpsrcipaddr='$IP'/>
</rule>
<rule action='drop' direction='in' priority='450'>
<arp opcode='Reply'/>
<arp match='no' arpdstmacaddr='$MAC'/>
</rule>
<rule action='drop' direction='in' priority='500'>
<arp match='no' arpdstipaddr='$IP'/>
</rule>
<rule action='accept' direction='inout' priority='600'>
<arp opcode='Request'/>

297

Virt ualiz at ion Administ rat ion G uide

</rule>
<rule action='accept' direction='inout' priority='650'>
<arp opcode='Reply'/>
</rule>
<rule action='drop' direction='inout' priority='1000'/>
</filter>
The consequence of putting ARP-specific rules in the arp chain, rather than for example in the root
chain, is that packets protocols other than ARP do not need to be evaluated by ARP protocolspecific rules. This improves the efficiency of the traffic filtering. However, one must then pay
attention to only putting filtering rules for the given protocol into the chain since other rules will not
be evaluated. For example, an IPv4 rule will not be evaluated in the ARP chain since IPv4 protocol
packets will not traverse the ARP chain.

18.12.3. Filt ering Chain Priorit ies

As previously mentioned, when creating a filtering rule, all chains are connected to the root chain.
The order in which those chains are accessed is influenced by the priority of the chain. The following
table shows the chains that can be assigned a priority and their default priorities.
T ab le 18.1. Filt erin g ch ain d ef au lt p rio rit ies valu es
C h ain ( p ref ix)

D ef au lt p rio rit y

stp
mac
vlan
ipv4
ipv6
arp
rarp

-810
-800
-750
-700
-600
-500
-400

Note
A chain with a lower priority value is accessed before one with a higher value.
The chains listed in Table 18.1, Filtering chain default priorities values can be also be
assigned custom priorities by writing a value in the range [-1000 to 1000] into the priority
(XML) attribute in the filter node. Section 18.12.2, Filtering Chains filter shows the default
priority of -500 for arp chains, for example.

18.12.4 . Usage of Variables in Filt ers

There are two variables that have been reserved for usage by the network traffic filtering subsystem:
MAC and IP.
MAC is designated for the MAC address of the network interface. A filtering rule that references this
variable will automatically be replaced with the MAC address of the interface. This works without the
user having to explicitly provide the MAC parameter. Even though it is possible to specify the MAC
parameter similar to the IP parameter above, it is discouraged since libvirt knows what MAC address
an interface will be using.

298

Chapt er 1 8 . Virt ual Net working

The parameter IP represents the IP address that the operating system inside the virtual machine is
expected to use on the given interface. The IP parameter is special in so far as the libvirt daemon will
try to determine the IP address (and thus the IP parameter's value) that is being used on an interface
if the parameter is not explicitly provided but referenced. For current limitations on IP address
detection, consult the section on limitations Section 18.12.12, Limitations on how to use this feature
and what to expect when using it. The XML file shown in Section 18.12.2, Filtering Chains contains
the filter no-arp-spoofing, which is an example of using a network filter XML to reference the MAC
and IP variables.
Note that referenced variables are always prefixed with the character $. The format of the value of a
variable must be of the type expected by the filter attribute identified in the XML. In the above example,
the IP parameter must hold a legal IP address in standard format. Failure to provide the correct
structure will result in the filter variable not being replaced with a value and will prevent a virtual
machine from starting or will prevent an interface from attaching when hotplugging is being used.
Some of the types that are expected for each XML attribute are shown in the example Example 18.4,
Sample variable types .

Examp le 18.4 . Samp le variab le t yp es

As variables can contain lists of elements, (the variable IP can contain multiple IP addresses that
are valid on a particular interface, for example), the notation for providing multiple elements for the
IP variable is:
<devices>
<interface type='bridge'>
<mac address='00:16:3e:5d:c7:9e'/>
<filterref filter='clean-traffic'>
<parameter name='IP' value='10.0.0.1'/>
<parameter name='IP' value='10.0.0.2'/>
<parameter name='IP' value='10.0.0.3'/>
</filterref>
</interface>
</devices>
This XML file creates filters to enable multiple IP addresses per interface. Each of the IP addresses
will result in a seperate filtering rule. Therefore using the XML above and the the following rule,
three individual filtering rules (one for each IP address) will be created:
<rule action='accept' direction='in' priority='500'>
<tcp srpipaddr='$IP'/>
</rule>
As it is possible to access individual elements of a variable holding a list of elements, a filtering
rule like the following accesses the 2nd element of the variable DSTPORTS.
<rule action='accept' direction='in' priority='500'>
<udp dstportstart='$DSTPORTS[1]'/>
</rule>

Examp le 18.5. U sin g a variet y o f variab les

As it is possible to create filtering rules that represent all possible combinations of rules from
different lists using the notation $VAR IABLE[@ <i terato r i d = "x">]. The following rule allows

299

Virt ualiz at ion Administ rat ion G uide

a virtual machine to receive traffic on a set of ports, which are specified in DSTPORTS, from the set
of source IP address specified in SRCIPADDRESSES. The rule generates all combinations of
elements of the variable DSTPORTS with those of SRCIPADDRESSES by using two independent
iterators to access their elements.
<rule action='accept' direction='in' priority='500'>
<ip srcipaddr='$SRCIPADDRESSES[@ 1]' dstportstart='$DSTPORTS[@ 2]'/>
</rule>
Assign concrete values to SRCIPADDRESSES and DSTPORTS as shown:
SRCIPADDRESSES = [ 10.0.0.1, 11.1.2.3 ]
DSTPORTS = [ 80, 8080 ]
Assigning values to the variables using $SR C IP AD D R ESSES[@ 1] and $D ST P O R T S[@ 2] would
then result in all combinations of addresses and ports being created as shown:
10.0.0.1, 80
10.0.0.1, 8080
11.1.2.3, 80
11.1.2.3, 8080
Accessing the same variables using a single iterator, for example by using the notation
$SR C IP AD D R ESSES[@ 1] and $D ST P O R T S[@ 1], would result in parallel access to both lists
and result in the following combination:
10.0.0.1, 80
11.1.2.3, 8080

Note
$VAR IABLE is short-hand for $VAR IABLE[@ 0 ]. The former notation always assumes the role
of iterator with i terato r i d = "0 " added as shown in the opening paragraph at the top of
this section.

18.12.5. Aut omat ic IP Address Det ect ion and DHCP Snooping
This section provides information about automatic IP address detection and D HCP snooping.

1 8 .1 2 .5 .1 . Int ro duct io n
The detection of IP addresses used on a virtual machine's interface is automatically activated if the
variable IP is referenced but no value has been assigned to it. The variable CTRL_IP_LEARNING
can be used to specify the IP address learning method to use. Valid values include: any, dhcp, or
none.
The value any instructs libvirt to use any packet to determine the address in use by a virtual machine,
which is the default setting if the variable TRL_IP_LEARNING is not set. This method will only detect
a single IP address per interface. Once a guest virtual machine's IP address has been detected, its IP

300

Chapt er 1 8 . Virt ual Net working

network traffic will be locked to that address, if for example, IP address spoofing is prevented by one
of its filters. In that case, the user of the VM will not be able to change the IP address on the interface
inside the guest virtual machine, which would be considered IP address spoofing. When a guest
virtual machine is migrated to another host physical machine or resumed after a suspend operation,
the first packet sent by the guest virtual machine will again determine the IP address that the guest
virtual machine can use on a particular interface.
The value of dhcp instructs libvirt to only honor D HCP server-assigned addresses with valid leases.
This method supports the detection and usage of multiple IP address per interface. When a guest
virtual machine resumes after a suspend operation, any valid IP address leases are applied to its
filters. Otherwise the guest virtual machine is expected to use D HCP to obtain a new IP addresses.
When a guest virtual machine migrates to another physical host physical machine, the guest virtual
machine is required to re-run the D HCP protocol.
If CTRL_IP_LEARNING is set to none, libvirt does not do IP address learning and referencing IP
without assigning it an explicit value is an error.

1 8 .1 2 .5 .2 . DHCP Sno o ping

C T R L_IP _LEAR NING = dhcp (D HCP snooping) provides additional anti-spoofing security,
especially when combined with a filter allowing only trusted D HCP servers to assign IP addresses.
To enable this, set the variable DHCPSERVER to the IP address of a valid D HCP server and provide
filters that use this variable to filter incoming D HCP responses.
When D HCP snooping is enabled and the D HCP lease expires, the guest virtual machine will no
longer be able to use the IP address until it acquires a new, valid lease from a D HCP server. If the
guest virtual machine is migrated, it must get a new valid D HCP lease to use an IP address (e.g., by
bringing the VM interface down and up again).

Note
Automatic D HCP detection listens to the D HCP traffic the guest virtual machine exchanges with
the D HCP server of the infrastructure. To avoid denial-of-service attacks on libvirt, the
evaluation of those packets is rate-limited, meaning that a guest virtual machine sending an
excessive number of D HCP packets per second on an interface will not have all of those
packets evaluated and thus filters may not get adapted. Normal D HCP client behavior is
assumed to send a low number of D HCP packets per second. Further, it is important to setup
appropriate filters on all guest virtual machines in the infrastructure to avoid them being able
to send D HCP packets. Therefore guest virtual machines must either be prevented from
sending UD P and TCP traffic from port 67 to port 68 or the D HCPSERVER variable should be
used on all guest virtual machines to restrict D HCP server messages to only be allowed to
originate from trusted D HCP servers. At the same time anti-spoofing prevention must be
enabled on all guest virtual machines in the subnet.

Examp le 18.6 . Act ivat in g IPs f o r D H C P sn o o p in g

The following XML provides an example for the activation of IP address learning using the D HCP
snooping method:
<interface type='bridge'>
<source bridge='virbr0'/>
<filterref filter='clean-traffic'>
<parameter name='CTRL_IP_LEARNING' value='dhcp'/>

301

Virt ualiz at ion Administ rat ion G uide

</filterref>
</interface>

18.12.6. Reserved Variables

Table 18.2, Reserved variables shows the variables that are considered reserved and are used by
libvirt:
T ab le 18.2. R eserved variab les
Variab le N ame

D ef in it io n

MAC
IP
IPV6

The MAC address of the interface

The list of IP addresses in use by an interface
Not currently implemented: the list of IPV6
addresses in use by an interface
The list of IP addresses of trusted D HCP servers
Not currently implemented: The list of IPv6
addresses of trusted D HCP servers
The choice of the IP address detection mode

D HCPSERVER
D HCPSERVERV6
CTRL_IP_LEARNING

18.12.7. Element and At t ribut e Overview

The root element required for all network filters is named <fi l ter> with two possible attributes. The
name attribute provides a unique name of the given filter. The chai n attribute is optional but allows
certain filters to be better organized for more efficient processing by the firewall subsystem of the
underlying host physical machine. Currently the system only supports the following chains: ro o t,
i pv4 , i pv6 , arp and rarp.

18.12.8. References t o Ot her Filt ers

Any filter may hold references to other filters. Individual filters may be referenced multiple times in a
filter tree but references between filters must not introduce loops.

Examp le 18.7. An Examp le o f a clean t raf f ic f ilt er

The following shows the XML of the clean-traffic network filter referencing several other filters.
<filter name='clean-traffic'>
<uuid>6ef53069-ba34-94a0-d33d-17751b9b8cb1</uuid>
<filterref filter='no-mac-spoofing'/>
<filterref filter='no-ip-spoofing'/>
<filterref filter='allow-incoming-ipv4'/>
<filterref filter='no-arp-spoofing'/>
<filterref filter='no-other-l2-traffic'/>
<filterref filter='qemu-announce-self'/>
</filter>
To reference another filter, the XML node filterref needs to be provided inside a filter node. This
node must have the attribute filter whose value contains the name of the filter to be referenced.
New network filters can be defined at any time and may contain references to network filters that are

302

Chapt er 1 8 . Virt ual Net working

not known to libvirt, yet. However, once a virtual machine is started or a network interface referencing
a filter is to be hotplugged, all network filters in the filter tree must be available. Otherwise the virtual
machine will not start or the network interface cannot be attached.

18.12.9. Filt er Rules

The following XML shows a simple example of a network traffic filter implementing a rule to drop traffic
if the IP address (provided through the value of the variable IP) in an outgoing IP packet is not the
expected one, thus preventing IP address spoofing by the VM.

Examp le 18.8. Examp le o f n et wo rk t raf f ic f ilt erin g

<filter name='no-ip-spoofing' chain='ipv4'>
<uuid>fce8ae33-e69e-83bf-262e-30786c1f8072</uuid>
<rule action='drop' direction='out' priority='500'>
<ip match='no' srcipaddr='$IP'/>
</rule>
</filter>

The traffic filtering rule starts with the rule node. This node may contain up to three of the following
attributes:
action is mandatory can have the following values:
drop (matching the rule silently discards the packet with no further analysis)
reject (matching the rule generates an ICMP reject message with no further analysis)
accept (matching the rule accepts the packet with no further analysis)
return (matching the rule passes this filter, but returns control to the calling filter for further
analysis)
continue (matching the rule goes on to the next rule for further analysis)
direction is mandatory can have the following values:
in for incomming traffic
out for outgoing traffic
inout for incoming and outgoing traffic
priority is optional. The priority of the rule controls the order in which the rule will be instantiated
relative to other rules. Rules with lower values will be instantiated before rules with higher values.
Valid values are in the range of -1000 to 1000. If this attribute is not provided, priority 500 will be
assigned by default. Note that filtering rules in the root chain are sorted with filters connected to
the root chain following their priorities. This allows to interleave filtering rules with access to filter
chains. Refer to Section 18.12.3, Filtering Chain Priorities for more information.
statematch is optional. Possible values are '0' or 'false' to turn the underlying connection state
matching off. The default setting is 'true' or 1
For more information see Section 18.12.11, Advanced Filter Configuration Topics .

303

Virt ualiz at ion Administ rat ion G uide

The above example Example 18.7, An Example of a clean traffic filter indicates that the traffic of type
ip will be associated with the chain ipv4 and the rule will have pri o ri ty= 500. If for example another
filter is referenced whose traffic of type ip is also associated with the chain ipv4 then that filter's rules
will be ordered relative to the pri o ri ty= 500 of the shown rule.
A rule may contain a single rule for filtering of traffic. The above example shows that traffic of type ip
is to be filtered.

18.12.10. Support ed Prot ocols

The following sections list and give some details about the protocols that are supported by the
network filtering subsystem. This type of traffic rule is provided in the rule node as a nested node.
D epending on the traffic type a rule is filtering, the attributes are different. The above example showed
the single attribute srcipaddr that is valid inside the ip traffic filtering node. The following sections
show what attributes are valid and what type of data they are expecting. The following datatypes are
available:
UINT8 : 8 bit integer; range 0-255
UINT16: 16 bit integer; range 0-65535
MAC_AD D R: MAC address in dotted decimal format, i.e., 00:11:22:33:44:55
MAC_MASK: MAC address mask in MAC address format, i.e., FF:FF:FF:FC:00:00
IP_AD D R: IP address in dotted decimal format, i.e., 10.1.2.3
IP_MASK: IP address mask in either dotted decimal format (255.255.248.0) or CID R mask (0-32)
IPV6_AD D R: IPv6 address in numbers format, i.e., FFFF::1
IPV6_MASK: IPv6 mask in numbers format (FFFF:FFFF:FC00::) or CID R mask (0-128)
STRING: A string
BOOLEAN: 'true', 'yes', '1' or 'false', 'no', '0'
IPSETFLAGS: The source and destination flags of the ipset described by up to 6 'src' or 'dst'
elements selecting features from either the source or destination part of the packet header;
example: src,src,dst. The number of 'selectors' to provide here depends on the type of ipset that is
referenced
Every attribute except for those of type IP_MASK or IPV6_MASK can be negated using the match
attribute with value no. Multiple negated attributes may be grouped together. The following XML
fragment shows such an example using abstract attributes.
[...]
<rule action='drop' direction='in'>
<protocol match='no' attribute1='value1' attribute2='value2'/>
<protocol attribute3='value3'/>
</rule>
[...]
Rules behave evaluate the rule as well as look at it logically within the boundaries of the given
protocol attributes. Thus, if a single attribute's value does not match the one given in the rule, the
whole rule will be skipped during the evaluation process. Therefore, in the above example incoming
traffic will only be dropped if: the protocol property attribute1 does not match both value1 and

304

Chapt er 1 8 . Virt ual Net working

the protocol property attribute2 does not match value2 and the protocol property attribute3
matches value3.

1 8 .1 2 .1 0 .1 . MAC (Et he rne t )

Protocol ID : mac
Rules of this type should go into the root chain.
T ab le 18.3. MAC p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

dstmacaddr
dstmacmask

MAC_AD D R
MAC_MASK

protocolid

UINT16 (0x600-0xffff), STRING

comment

STRING

MAC address of sender

Mask applied to MAC address
of sender
MAC address of destination
Mask applied to MAC address
of destination
Layer 3 protocol ID . Valid
strings include [arp, rarp, ipv4,
ipv6]
text string up to 256 characters

The filter can be written as such:

[...]
<mac match='no' srcmacaddr='$MAC'/>
[...]

1 8 .1 2 .1 0 .2 . VLAN (8 0 2 .1 Q)
Protocol ID : vlan
Rules of this type should go either into the root or vlan chain.
T ab le 18.4 . VLAN p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

dstmacaddr
dstmacmask

MAC_AD D R
MAC_MASK

vlan-id
encap-protocol

UINT16 (0x0-0xfff, 0 - 4095)

UINT16 (0x03c-0xfff), String

comment

STRING

MAC address of sender

Mask applied to MAC address
of sender
MAC address of destination
Mask applied to MAC address
of destination
VLAN ID
Encapsulated layer 3 protocol
ID , valid strings are arp, ipv4,
ipv6
text string up to 256 characters

1 8 .1 2 .1 0 .3. ST P (Spanning T re e Pro t o co l)

Protocol ID : stp

305

Virt ualiz at ion Administ rat ion G uide

Rules of this type should go either into the root or stp chain.
T ab le 18.5. ST P p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

type

UINT8

flags
root-priority
root-priority-hi
root-address
root-address-mask
roor-cost
root-cost-hi
sender-priority-hi
sender-address
sender-address-mask

UINT8
UINT16
UINT16 (0x0-0xfff, 0 - 4095)
MAC _AD D RESS
MAC _MASK
UINT32
UINT32
UINT16
MAC_AD D RESS
MAC_MASK

port
port_hi
msg-age
msg-age-hi
max-age-hi
hello-time
hello-time-hi
forward-delay
forward-delay-hi
comment

UINT16
UINT16
UINT16
UINT16
UINT16
UINT16
UINT16
UINT16
UINT16
STRING

MAC address of sender

Mask applied to MAC address
of sender
Bridge Protocol D ata Unit
(BPD U) type
BPD U flagdstmacmask
Root priority range start
Root priority range end
root MAC Address
root MAC Address mask
Root path cost (range start)
Root path cost range end
Sender prioriry range end
BPD U sender MAC address
BPD U sender MAC address
mask
Port identifier (range start)
Port identifier range end
Message age timer (range start)
Message age timer range end
Maximum age time range end
Hello time timer (range start)
Hello time timer range end
Forward delay (range start)
Forward delay range end
text string up to 256 characters

1 8 .1 2 .1 0 .4 . ARP/RARP
Protocol ID : arp or rarp
Rules of this type should either go into the root or arp/rarp chain.
T ab le 18.6 . AR P an d R AR P p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

dstmacaddr
dstmacmask

MAC_AD D R
MAC_MASK

hwtype
protocoltype

UINT16
UINT16

MAC address of sender

Mask applied to MAC address
of sender
MAC address of destination
Mask applied to MAC address
of destination
Hardware type
Protocol type

306

Chapt er 1 8 . Virt ual Net working

At t rib u t e N ame

D at at yp e

D ef in it io n

opcode

UINT16, STRING

arpsrcmacaddr

MAC_AD D R

arpdstmacaddr

MAC _AD D R

arpsrcipaddr

IP_AD D R

arpdstipaddr

IP_AD D R

gratututous

BOOLEAN

comment

STRING

Opcode valid strings are:

Request, Reply,
Request_Reverse,
Reply_Reverse,
D RARP_Request,
D RARP_Reply, D RARP_Error,
InARP_Request, ARP_NAK
Source MAC address in
ARP/RARP packet
D estination MAC address in
ARP/RARP packet
Source IP address in
ARP/RARP packet
D estination IP address in
ARP/RARP packet
Boolean indiating whether to
check for a gratuitous ARP
packet
text string up to 256 characters

1 8 .1 2 .1 0 .5 . IPv4
Protocol ID : ip
Rules of this type should either go into the root or ipv4 chain.
T ab le 18.7. IPv4 p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

dstmacaddr
dstmacmask

MAC_AD D R
MAC_MASK

srcipaddr
srcipmask

IP_AD D R
IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

protocol

UINT8, STRING

srcportstart

UINT16

srcportend

UINT16

dstportstart

UNIT16

MAC address of sender

Mask applied to MAC address
of sender
MAC address of destination
Mask applied to MAC address
of destination
Source IP address
Mask applied to source IP
address
D estination IP address
Mask applied to destination IP
address
Layer 4 protocol identifier. Valid
strings for protocol are: tcp,
udp, udplite, esp, ah, icmp,
igmp, sctp
Start of range of valid source
ports; requires protocol
End of range of valid source
ports; requires protocol
Start of range of valid
destination ports; requires
protocol

307

Virt ualiz at ion Administ rat ion G uide

At t rib u t e N ame

D at at yp e

D ef in it io n

dstportend

UNIT16

comment

STRING

End of range of valid

destination ports; requires
protocol
text string up to 256 characters

1 8 .1 2 .1 0 .6 . IPv6
Protocol ID : ipv6
Rules of this type should either go into the root or ipv6 chain.
T ab le 18.8. IPv6 p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

dstmacaddr
dstmacmask

MAC_AD D R
MAC_MASK

srcipaddr
srcipmask

IP_AD D R
IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

protocol

UINT8, STRING

scrportstart

UNIT16

srcportend

UINT16

dstportstart

UNIT16

dstportend

UNIT16

comment

STRING

MAC address of sender

Mask applied to MAC address
of sender
MAC address of destination
Mask applied to MAC address
of destination
Source IP address
Mask applied to source IP
address
D estination IP address
Mask applied to destination IP
address
Layer 4 protocol identifier. Valid
strings for protocol are: tcp,
udp, udplite, esp, ah, icmpv6,
sctp
Start of range of valid source
ports; requires protocol
End of range of valid source
ports; requires protocol
Start of range of valid
destination ports; requires
protocol
End of range of valid
destination ports; requires
protocol
text string up to 256 characters

1 8 .1 2 .1 0 .7 . T CP/UDP/SCT P
Protocol ID : tcp, udp, sctp
The chain parameter is ignored for this type of traffic and should either be omitted or set to root. .
T ab le 18.9 . T C P/U D P/SC T P p ro t o co l t yp es

308

Chapt er 1 8 . Virt ual Net working

At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcipaddr
srcipmask

MAC_AD D R
IP_AD D R
IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

scripto

IP_AD D R

srcipfrom

IP_AD D R

dstipfrom

IP_AD D R

dstipto

IP_AD D R

scrportstart

UNIT16

srcportend

UINT16

dstportstart

UNIT16

dstportend

UNIT16

comment
state

STRING
STRING

flags

STRING

ipset

STRING

ipsetflags

IPSETFLAGS

MAC address of sender

Source IP address
Mask applied to source IP
address
D estination IP address
Mask applied to destination IP
address
Start of range of source IP
address
End of range of source IP
address
Start of range of destination IP
address
End of range of destination IP
address
Start of range of valid source
ports; requires protocol
End of range of valid source
ports; requires protocol
Start of range of valid
destination ports; requires
protocol
End of range of valid
destination ports; requires
protocol
text string up to 256 characters
comma separated list of
NEW,ESTABLISHED ,RELATED ,I
NVALID or NONE
TCP-only: format of mask/flags
with mask and flags each being
a comma separated list of
SYN,ACK,URG,PSH,FIN,RST or
NONE or ALL
The name of an IPSet managed
outside of libvirt
flags for the IPSet; requires
ipset attribute

1 8 .1 2 .1 0 .8 . ICMP
Protocol ID : icmp
Note: The chain parameter is ignored for this type of traffic and should either be omitted or set to root.
T ab le 18.10. IC MP p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

MAC address of sender

Mask applied to the MAC
address of the sender

309

Virt ualiz at ion Administ rat ion G uide

At t rib u t e N ame

D at at yp e

D ef in it io n

dstmacaddr
dstmacmask

MAD _AD D R
MAC_MASK

srcipaddr
srcipmask

IP_AD D R
IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

srcipfrom

IP_AD D R

scripto

IP_AD D R

dstipfrom

IP_AD D R

dstipto

IP_AD D R

type
code
comment
state

UNIT16
UNIT16
STRING
STRING

ipset

STRING

ipsetflags

IPSETFLAGS

MAC address of the destination

Mask applied to the MAC
address of the destination
Source IP address
Mask applied to source IP
address
D estination IP address
Mask applied to destination IP
address
start of range of source IP
address
end of range of source IP
address
Start of range of destination IP
address
End of range of destination IP
address
ICMP type
ICMP code
text string up to 256 characters
comma separated list of
NEW,ESTABLISHED ,RELATED ,I
NVALID or NONE
The name of an IPSet managed
outside of libvirt
flags for the IPSet; requires
ipset attribute

1 8 .1 2 .1 0 .9 . IGMP, ESP, AH, UDPLIT E, 'ALL'

Protocol ID : igmp, esp, ah, udplite, all
The chain parameter is ignored for this type of traffic and should either be omitted or set to root.
T ab le 18.11. IG MP, ESP, AH , U D PLIT E, ' ALL'
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcmacmask

MAC_AD D R
MAC_MASK

dstmacaddr
dstmacmask

MAD _AD D R
MAC_MASK

srcipaddr
srcipmask

IP_AD D R
IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

srcipfrom

IP_AD D R

MAC address of sender

Mask applied to the MAC
address of the sender
MAC address of the destination
Mask applied to the MAC
address of the destination
Source IP address
Mask applied to source IP
address
D estination IP address
Mask applied to destination IP
address
start of range of source IP
address

310

Chapt er 1 8 . Virt ual Net working

At t rib u t e N ame

D at at yp e

D ef in it io n

scripto

IP_AD D R

dstipfrom

IP_AD D R

dstipto

IP_AD D R

comment
state

STRING
STRING

ipset

STRING

ipsetflags

IPSETFLAGS

end of range of source IP

address
Start of range of destination IP
address
End of range of destination IP
address
text string up to 256 characters
comma separated list of
NEW,ESTABLISHED ,RELATED ,I
NVALID or NONE
The name of an IPSet managed
outside of libvirt
flags for the IPSet; requires
ipset attribute

1 8 .1 2 .1 0 .1 0 . T CP/UDP/SCT P o ve r IPV6
Protocol ID : tcp-ipv6, udp-ipv6, sctp-ipv6
The chain parameter is ignored for this type of traffic and should either be omitted or set to root.
T ab le 18.12. T C P, U D P, SC T P o ver IPv6 p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcipaddr
srcipmask

MAC_AD D R
IP_AD D R
IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

srcipfrom

IP_AD D R

scripto

IP_AD D R

dstipfrom

IP_AD D R

dstipto

IP_AD D R

srcportstart

UINT16

srcportend

UINT16

dstportstart

UINT16

dstportend

UINT16

comment
state

STRING
STRING

MAC address of sender

Source IP address
Mask applied to source IP
address
D estination IP address
Mask applied to destination IP
address
start of range of source IP
address
end of range of source IP
address
Start of range of destination IP
address
End of range of destination IP
address
Start of range of valid source
ports
End of range of valid source
ports
Start of range of valid
destination ports
End of range of valid
destination ports
text string up to 256 characters
comma separated list of
NEW,ESTABLISHED ,RELATED ,I
NVALID or NONE

311

Virt ualiz at ion Administ rat ion G uide

At t rib u t e N ame

D at at yp e

D ef in it io n

ipset

STRING

ipsetflags

IPSETFLAGS

The name of an IPSet managed

outside of libvirt
flags for the IPSet; requires
ipset attribute

1 8 .1 2 .1 0 .1 1 . ICMPv6
Protocol ID : icmpv6
The chain parameter is ignored for this type of traffic and should either be omitted or set to root.
T ab le 18.13. IC MPv6 p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcipaddr
srcipmask

MAC_AD D R
IP_AD D R
IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

srcipfrom

IP_AD D R

scripto

IP_AD D R

dstipfrom

IP_AD D R

dstipto

IP_AD D R

type
code
comment
state

UINT16
UINT16
STRING
STRING

ipset

STRING

ipsetflags

IPSETFLAGS

MAC address of sender

Source IP address
Mask applied to source IP
address
D estination IP address
Mask applied to destination IP
address
start of range of source IP
address
end of range of source IP
address
Start of range of destination IP
address
End of range of destination IP
address
ICMPv6 type
ICMPv6 code
text string up to 256 characters
comma separated list of
NEW,ESTABLISHED ,RELATED ,I
NVALID or NONE
The name of an IPSet managed
outside of libvirt
flags for the IPSet; requires
ipset attribute

1 8 .1 2 .1 0 .1 2 . IGMP, ESP, AH, UDPLIT E, 'ALL' o ve r IPv6

Protocol ID : igmp-ipv6, esp-ipv6, ah-ipv6, udplite-ipv6, all-ipv6
The chain parameter is ignored for this type of traffic and should either be omitted or set to root.
T ab le 18.14 . IG MP, ESP, AH , U D PLIT E, ' ALL' o ver IPv p ro t o co l t yp es
At t rib u t e N ame

D at at yp e

D ef in it io n

srcmacaddr
srcipaddr

MAC_AD D R
IP_AD D R

MAC address of sender

Source IP address

312

Chapt er 1 8 . Virt ual Net working

At t rib u t e N ame

D at at yp e

D ef in it io n

srcipmask

IP_MASK

dstipaddr
dstipmask

IP_AD D R
IP_MASK

srcipfrom

IP_AD D R

scripto

IP_AD D R

dstipfrom

IP_AD D R

dstipto

IP_AD D R

comment
state

STRING
STRING

ipset

STRING

ipsetflags

IPSETFLAGS

Mask applied to source IP

address
D estination IP address
Mask applied to destination IP
address
start of range of source IP
address
end of range of source IP
address
Start of range of destination IP
address
End of range of destination IP
address
text string up to 256 characters
comma separated list of
NEW,ESTABLISHED ,RELATED ,I
NVALID or NONE
The name of an IPSet managed
outside of libvirt
flags for the IPSet; requires
ipset attribute

18.12.11. Advanced Filt er Configurat ion T opics

The following sections discuss advanced filter configuration topics.

1 8 .1 2 .1 1 .1 . Co nne ct io n t racking
The network filtering subsystem (on Linux) makes use of the connection tracking support of IP tables.
This helps in enforcing the directionality of network traffic (state match) as well as counting and
limiting the number of simultaneous connections towards a guest virtual machine. As an example, if a
guest virtual machine has TCP port 8080 open as a server, clients may connect to the guest virtual
machine on port 8080. Connection tracking and enforcement of directionality then prevents the guest
virtual machine from initiating a connection from (TCP client) port 8080 to the host physical machine
back to a remote host physical machine. More importantly, tracking helps to prevent remote attackers
from establishing a connection back to a guest virtual machine. For example, if the user inside the
guest virtual machine established a connection to port 80 on an attacker site, then the attacker will
not be able to initiate a connection from TCP port 80 back towards the guest virtual machine. By
default the connection state match that enables connection tracking and then enforcement of
directionality of traffic is turned on.

Examp le 18.9 . XML examp le f o r t u rn in g o f f co n n ect io n s t o t h e T C P p o rt

The following shows an example XML fragment where this feature has been turned off for incoming
connections to TCP port 12345.
[...]
<rule direction='in' action='accept' statematch='false'>
<cp dstportstart='12345'/>
</rule>
[...]

313

Virt ualiz at ion Administ rat ion G uide

This now allows incoming traffic to TCP port 12345, but would also enable the initiation from
(client) TCP port 12345 within the VM, which may or may not be desirable.

1 8 .1 2 .1 1 .2 . Lim it ing Num be r o f Co nne ct io ns

To limit the number of connections a guest virtual machine may establish, a rule must be provided
that sets a limit of connections for a given type of traffic. If for example a VM is supposed to be
allowed to only ping one other IP address at a time and is supposed to have only one active
incoming ssh connection at a time.

Examp le 18.10. XML samp le f ile t h at set s limit s t o co n n ect io n s

The following XML fragment can be used to limit connections
[...]
<rule action='drop' direction='in' priority='400'>
<tcp connlimit-above='1'/>
</rule>
<rule action='accept' direction='in' priority='500'>
<tcp dstportstart='22'/>
</rule>
<rule action='drop' direction='out' priority='400'>
<icmp connlimit-above='1'/>
</rule>
<rule action='accept' direction='out' priority='500'>
<icmp/>
</rule>
<rule action='accept' direction='out' priority='500'>
<udp dstportstart='53'/>
</rule>
<rule action='drop' direction='inout' priority='1000'>
<all/>
</rule>
[...]

314

Chapt er 1 8 . Virt ual Net working

Note
Limitation rules must be listed in the XML prior to the rules for accepting traffic. According to
the XML file in Example 18.10, XML sample file that sets limits to connections , an additional
rule for allowing D NS traffic sent to port 22 go out the guest virtual machine, has been added
to avoid ssh sessions not getting established for reasons related to D NS lookup failures by
the ssh daemon. Leaving this rule out may result in the ssh client hanging unexpectedly as it
tries to connect. Additional caution should be used in regards to handling timeouts related to
tracking of traffic. An ICMP ping that the user may have terminated inside the guest virtual
machine may have a long timeout in the host physical machine's connection tracking system
and will therefore not allow another ICMP ping to go through.
The best solution is to tune the timeout in the host physical machine's sysfs with the
following command:# echo 3 >
/pro c/sys/net/netfi l ter/nf_co nntrack_i cmp_ti meo ut. This command sets the
ICMP connection tracking timeout to 3 seconds. The effect of this is that once one ping is
terminated, another one can start after 3 seconds.
If for any reason the guest virtual machine has not properly closed its TCP connection, the
connection to be held open for a longer period of time, especially if the TCP timeout value was
set for a large amount of time on the host physical machine. In addition, any idle connection
may result in a time out in the connection tracking system which can be re-activated once
packets are exchanged.
However, if the limit is set too low, newly initiated connections may force an idle connection
into TCP backoff. Therefore, the limit of connections should be set rather high so that
fluctuations in new TCP connections don't cause odd traffic behavior in relation to idle
connections.

1 8 .1 2 .1 1 .3. Co m m and line t o o ls

virsh has been extended with life-cycle support for network filters. All commands related to the
network filtering subsystem start with the prefix nwfilter. The following commands are available:
nwfi l ter-l i st : lists UUID s and names of all network filters
nwfi l ter-d efi ne : defines a new network filter or updates an existing one (must supply a
name)
nwfi l ter-und efi ne : deletes a specified network filter (must supply a name). D o not delete a
network filter currently in use.
nwfi l ter-d umpxml : displays a specified network filter (must supply a name)
nwfi l ter-ed i t : edits a specified network filter (must supply a name)

1 8 .1 2 .1 1 .4 . Pre -e xist ing ne t wo rk filt e rs

The following is a list of example network filters that are automatically installed with libvirt:
T ab le 18.15. IC MPv6 p ro t o co l t yp es
C o mman d N ame

D escrip t io n

315

Virt ualiz at ion Administ rat ion G uide

C o mman d N ame

D escrip t io n

no-arp-spoofing

Prevents a guest virtual machine from spoofing

ARP traffic; this filter only allows ARP request
and reply messages and enforces that those
packets contain the MAC and IP addresses of
the guest virtual machine.
Allows a guest virtual machine to request an IP
address via D HCP (from any D HCP server)
Allows a guest virtual machine to request an IP
address from a specified D HCP server. The
dotted decimal IP address of the D HCP server
must be provided in a reference to this filter. The
name of the variable must be DHCPSERVER.
Prevents a guest virtual machine from sending
IP packets with a source IP address different
from the one inside the packet.
Prevents a guest virtual machine from sending
IP multicast packets.
Prevents MAC, IP and ARP spoofing. This filter
references several other filters as building
blocks.

allow-dhcp
allow-dhcp-server

no-ip-spoofing

no-ip-multicast
clean-traffic

These filters are only building blocks and require a combination with other filters to provide useful
network traffic filtering. The most used one in the above list is the clean-traffic filter. This filter itself can
for example be combined with the no-ip-multicast filter to prevent virtual machines from sending IP
multicast traffic on top of the prevention of packet spoofing.

1 8 .1 2 .1 1 .5 . Writ ing yo ur o wn filt e rs

Since libvirt only provides a couple of example networking filters, you may consider writing your own.
When planning on doing so there are a couple of things you may need to know regarding the
network filtering subsystem and how it works internally. Certainly you also have to know and
understand the protocols very well that you want to be filtering on so that no further traffic than what
you want can pass and that in fact the traffic you want to allow does pass.
The network filtering subsystem is currently only available on Linux host physical machines and only
works for Qemu and KVM type of virtual machines. On Linux, it builds upon the support for ebtables,
iptables and ip6tables and makes use of their features. Considering the list found in
Section 18.12.10, Supported Protocols the following protocols can be implemented using ebtables:
mac
stp (spanning tree protocol)
vlan (802.1Q)
arp, rarp
ipv4
ipv6
Any protocol that runs over IPv4 is supported using iptables, those over IPv6 are implemented using
ip6tables.

316

Chapt er 1 8 . Virt ual Net working

Using a Linux host physical machine, all traffic filtering rules created by libvirt's network filtering
subsystem first passes through the filtering support implemented by ebtables and only afterwards
through iptables or ip6tables filters. If a filter tree has rules with the protocols including: mac, stp,
vlan arp, rarp, ipv4, or ipv6; the ebtable rules and values listed will automatically be used first.
Multiple chains for the same protocol can be created. The name of the chain must have a prefix of
one of the previously enumerated protocols. To create an additional chain for handling of ARP
traffic, a chain with name arp-test, can for example be specified.
As an example, it is possible to filter on UD P traffic by source and destination ports using the IP
protocol filter and specifying attributes for the protocol, source and destination IP addresses and
ports of UD P packets that are to be accepted. This allows early filtering of UD P traffic with ebtables.
However, once an IP or IPv6 packet, such as a UD P packet, has passed the ebtables layer and there
is at least one rule in a filter tree that instantiates iptables or ip6tables rules, a rule to let the UD P
packet pass will also be necessary to be provided for those filtering layers. This can be achieved with
a rule containing an appropriate udp or udp-ipv6 traffic filtering node.

Examp le 18.11. C reat in g a cu st o m f ilt er

Suppose a filter is needed to fulfill the following list of requirements:
prevents a VM's interface from MAC, IP and ARP spoofing
opens only TCP ports 22 and 80 of a VM's interface
allows the VM to send ping traffic from an interface but not let the VM be pinged on the interface
allows the VM to do D NS lookups (UD P towards port 53)
The requirement to prevent spoofing is fulfilled by the existing clean-traffic network filter, thus
the way to do this is to reference it from a custom filter.
To enable traffic for TCP ports 22 and 80, two rules are added to enable this type of traffic. To
allow the guest virtual machine to send ping traffic a rule is added for ICMP traffic. For simplicity
reasons, general ICMP traffic will be allowed to be initiated from the guest virtual machine, and will
not be specified to ICMP echo request and response messages. All other traffic will be prevented to
reach or be initiated by the guest virtual machine. To do this a rule will be added that drops all
other traffic. Assuming the guest virtual machine is called test and the interface to associate our
filter with is called eth0 , a filter is created named test-eth0 .
The result of these considerations is the following network filter XML:
<filter name='test-eth0'>
<!- - This rule references the clean traffic filter to prevent MAC,
IP and ARP spoofing. By not providing an IP address parameter, libvirt
will detect the IP address the guest virtual machine is using. - ->
<filterref filter='clean-traffic'/>
<!- - This rule enables TCP ports 22 (ssh) and 80 (http) to be
reachable - ->
<rule action='accept' direction='in'>
<tcp dstportstart='22'/>
</rule>
<rule action='accept' direction='in'>
<tcp dstportstart='80'/>
</rule>

317

Virt ualiz at ion Administ rat ion G uide

<!- - This rule enables general ICMP traffic to be initiated by the

guest virtual machine including ping traffic - ->
<rule action='accept' direction='out'>
<icmp/>
</rule>>
<!- - This rule enables outgoing DNS lookups using UDP - ->
<rule action='accept' direction='out'>
<udp dstportstart='53'/>
</rule>
<!- - This rule drops all other traffic - ->
<rule action='drop' direction='inout'>
<all/>
</rule>
</filter>

1 8 .1 2 .1 1 .6 . Sam ple cust o m filt e r

Although one of the rules in the above XML contains the IP address of the guest virtual machine as
either a source or a destination address, the filtering of the traffic works correctly. The reason is that
whereas the rule's evaluation occurs internally on a per-interface basis, the rules are additionally
evaluated based on which (tap) interface has sent or will receive the packet, rather than what their
source or destination IP address may be.

Examp le 18.12. Samp le XML f o r n et wo rk in t erf ace d escrip t io n s

An XML fragment for a possible network interface description inside the domain XML of the test
guest virtual machine could then look like this:
[...]
<interface type='bridge'>
<source bridge='mybridge'/>
<filterref filter='test-eth0'/>
</interface>
[...]
To more strictly control the ICMP traffic and enforce that only ICMP echo requests can be sent from
the guest virtual machine and only ICMP echo responses be received by the guest virtual machine,
the above ICMP rule can be replaced with the following two rules:
<!- - enable outgoing ICMP echo requests- ->
<rule action='accept' direction='out'>
<icmp type='8'/>
</rule>
<!- - enable incoming ICMP echo replies- ->
<rule action='accept' direction='in'>
<icmp type='0'/>
</rule>

318

Chapt er 1 8 . Virt ual Net working

Examp le 18.13. Seco n d examp le cu st o m f ilt er

This example demonstrates how to build a similar filter as in the example above, but extends the
list of requirements with an ftp server located inside the guest virtual machine. The requirements for
this filter are:
prevents a guest virtual machine's interface from MAC, IP, and ARP spoofing
opens only TCP ports 22 and 80 in a guest virtual machine's interface
allows the guest virtual machine to send ping traffic from an interface but does not allow the
guest virtual machine to be pinged on the interface
allows the guest virtual machine to do D NS lookups (UD P towards port 53)
enables the ftp server (in active mode) so it can run inside the guest virtual machine
The additional requirement of allowing an FTP server to be run inside the guest virtual machine
maps into the requirement of allowing port 21 to be reachable for FTP control traffic as well as
enabling the guest virtual machine to establish an outgoing TCP connection originating from the
guest virtual machine's TCP port 20 back to the FTP client (FTP active mode). There are several
ways of how this filter can be written and two possible solutions are included in this example.
The first solution makes use of the state attribute of the TCP protocol that provides a hook into the
connection tracking framework of the Linux host physical machine. For the guest virtual machineinitiated FTP data connection (FTP active mode) the RELATED state is used to enable detection
that the guest virtual machine-initiated FTP data connection is a consequence of ( or 'has a
relationship with' ) an existing FTP control connection, thereby allowing it to pass packets
through the firewall. The RELATED state, however, is only valid for the very first packet of the
outgoing TCP connection for the FTP data path. Afterwards, the state is ESTABLISHED , which
then applies equally to the incoming and outgoing direction. All this is related to the FTP data
traffic originating from TCP port 20 of the guest virtual machine. This then leads to the following
solution:
<filter name='test-eth0'>
<!- - This filter (eth0) references the cl ean traffi c filter to
prevent MAC, IP, and ARP spoofing. By not providing an IP address
parameter, libvirt will detect the IP address the guest virtual
machine is using. - ->
<filterref filter='clean-traffic'/>
<!- - This rule enables TCP port 21 (FTP-control) to be reachable ->
<rule action='accept' direction='in'>
<tcp dstportstart='21'/>
</rule>
<!- - This rule enables TCP port 20 for guest virtual machineinitiated FTP data connection related to an existing FTP control
connection - ->
<rule action='accept' direction='out'>
<tcp srcportstart='20' state='RELATED,ESTABLISHED'/>
</rule>
<!- - This rule accepts all packets from a client on the FTP data
connection - ->

319

Virt ualiz at ion Administ rat ion G uide

<rule action='accept' direction='in'>

<tcp dstportstart='20' state='ESTABLISHED'/>
</rule>
<!- - This rule enables TCP port 22 (SSH) to be reachable - ->
<rule action='accept' direction='in'>
<tcp dstportstart='22'/>
</rule>
<!- -This rule enables TCP port 80 (HTTP) to be reachable - ->
<rule action='accept' direction='in'>
<tcp dstportstart='80'/>
</rule>
<!- - This rule enables general ICMP traffic to be initiated by the
guest virtual machine, including ping traffic - ->
<rule action='accept' direction='out'>
<icmp/>
</rule>
<!- - This rule enables outgoing DNS lookups using UDP - ->
<rule action='accept' direction='out'>
<udp dstportstart='53'/>
</rule>
<!- - This rule drops all other traffic - ->
<rule action='drop' direction='inout'>
<all/>
</rule>
</filter>
Before trying out a filter using the RELATED state, you have to make sure that the appropriate
connection tracking module has been loaded into the host physical machine's kernel. D epending
on the version of the kernel, you must run either one of the following two commands before the FTP
connection with the guest virtual machine is established:
#mo d pro be nf_co nntrack_ftp - where available OR
#mo d pro be i p_co nntrack_ftp if above is not available
If protocols other than FTP are used in conjunction with the RELATED state, their corresponding
module must be loaded. Modules are available for the protocols: ftp, tftp, irc, sip, sctp, and
amanda.
The second solution makes use of the state flags of connections more than the previous solution
did. This solution takes advantage of the fact that the NEW state of a connection is valid when the
very first packet of a traffic flow is detected. Subsequently, if the very first packet of a flow is
accepted, the flow becomes a connection and thus enters into the ESTABLISHED state. Therefore
a general rule can be written for allowing packets of ESTABLISHED connections to reach the guest
virtual machine or be sent by the guest virtual machine. This is done writing specific rules for the
very first packets identified by the NEW state and dictates the ports that the data is acceptable. All
packets meant for ports that are not explicitly accepted are dropped, thus not reaching an
ESTABLISHED state. Any subsequent packets sent from that port are dropped as well.
<filter name='test-eth0'>
<!- - This filter references the cl ean traffi c filter to prevent MAC,

320

Chapt er 1 8 . Virt ual Net working

IP and ARP spoofing. By not providing and IP address parameter, libvirt

will detect the IP address the VM is using. - ->
<filterref filter='clean-traffic'/>
<!- - This rule allows the packets of all previously accepted
connections to reach the guest virtual machine - ->
<rule action='accept' direction='in'>
<all state='ESTABLISHED'/>
</rule>
<!- - This rule allows the packets of all previously accepted and
related connections be sent from the guest virtual machine - ->
<rule action='accept' direction='out'>
<all state='ESTABLISHED,RELATED'/>
</rule>
<!- - This rule enables traffic towards port 21 (FTP) and port 22
(SSH)- ->
<rule action='accept' direction='in'>
<tcp dstportstart='21' dstportend='22' state='NEW'/>
</rule>
<!- - This rule enables traffic towards port 80 (HTTP) - ->
<rule action='accept' direction='in'>
<tcp dstportstart='80' state='NEW'/>
</rule>
<!- - This rule enables general ICMP traffic to be initiated by the
guest virtual machine, including ping traffic - ->
<rule action='accept' direction='out'>
<icmp state='NEW'/>
</rule>
<!- - This rule enables outgoing DNS lookups using UDP - ->
<rule action='accept' direction='out'>
<udp dstportstart='53' state='NEW'/>
</rule>
<!- - This rule drops all other traffic - ->
<rule action='drop' direction='inout'>
<all/>
</rule>
</filter>

18.12.12. Limit at ions

The following is a list of the currently known limitations of the network filtering subsystem.
VM migration is only supported if the whole filter tree that is referenced by a guest virtual
machine's top level filter is also available on the target host physical machine. The network filter
cl ean-traffi c for example should be available on all libvirt installations and thus enable
migration of guest virtual machines that reference this filter. To assure version compatibility is not
a problem make sure you are using the most current version of libvirt by updating the package
regularly.

321

Virt ualiz at ion Administ rat ion G uide

Migration must occur between libvirt installations of version 0.8.1 or later in order not to lose the
network traffic filters associated with an interface.
VLAN (802.1Q) packets, if sent by a guest virtual machine, cannot be filtered with rules for
protocol ID s arp, rarp, ipv4 and ipv6. They can only be filtered with protocol ID s, MAC and VLAN.
Therefore, the example filter clean-traffic Example 18.1, An example of network filtering will not
work as expected.

18.13. Creat ing T unnels

This section will demonstrate how to implement different tunneling scenarios.

18.13.1. Creat ing Mult icast T unnels

A multicast group is set up to represent a virtual network. Any guest virtual machines whose network
devices are in the same multicast group can talk to each other even across host physical machines.
This mode is also available to unprivileged users. There is no default D NS or D HCP support and no
outgoing network access. To provide outgoing network access, one of the guest virtual machines
should have a second NIC which is connected to one of the first four network types, thus providing
appropriate routing. The multicast protocol is compatible with the guest virtual machine user mode.
Note that the source address that you provide must be from the address used for the multicast
address block.
To create a multicast tunnel, specify the following XML details into the <d evi ces> element:

...
<devices>
<interface type='mcast'>
<mac address='52:54:00:6d:90:01'>
<source address='230.0.0.1' port='5558'/>
</interface>
</devices>
...

Fig u re 18.26 . Mu lt icast t u n n el XML examp le

18.13.2. Creat ing T CP T unnels

A TCP client/server architecture provides a virtual network. In this configuration, one guest virtual
machine provides the server end of the network while all other guest virtual machines are configured
as clients. All network traffic is routed between the guest virtual machine clients via the guest virtual
machine server. This mode is also available for unprivileged users. Note that this mode does not
provide default D NS or D HCP support nor does it provide outgoing network access. To provide
outgoing network access, one of the guest virtual machines should have a second NIC which is
connected to one of the first four network types thus providing appropriate routing.
To create a TCP tunnel place the following XML details into the <d evi ces> element:

322

Chapt er 1 8 . Virt ual Net working

...
<devices>
<interface type='server'>
<mac address='52:54:00:22:c9:42'>
<source address='192.168.0.1' port='5558'/>
</interface>
...
<interface type='client'>
<mac address='52:54:00:8b:c9:51'>
<source address='192.168.0.1' port='5558'/>
</interface>
</devices>
...

Fig u re 18.27. T C P t u n n el d o main XMl examp le

18.14 . Set t ing vLAN T ags

virtual local area network (vLAN) tags are added using the vi rsh net-ed i t command. This tag can
also be used with PCI device assignment with SR-IOV devices. For more information, refer to
Section 9.1.7, Configuring PCI Assignment (Passthrough) with SR-IOV D evices .

<network>
<name>ovs-net</name>
<forward mode='bridge'/>
<bridge name='ovsbr0'/>
<virtualport type='openvswitch'>
<parameters interfaceid='09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f'/>
</virtualport>
<vlan trunk='yes'>
<tag id='42' nativeMode='untagged'/>
<tag id='47'/>
</vlan>
<portgroup name='dontpanic'>
<vlan>
<tag id='42'/>
</vlan>
</portgroup>
</network>
Fig u re 18.28. vSet t in g VLAN t ag ( o n su p p o rt ed n et wo rk t yp es o n ly)
If (and only if) the network type supports vlan tagging transparent to the guest, an optional <vl an>
element can specify one or more vlan tags to apply to the traffic of all guests using this network.
(openvswitch and type='hostdev' SR-IOV networks do support transparent VLAN tagging of guest
traffic; everything else, including standard linux bridges and libvirt's own virtual networks, do not
support it. 802.1Qbh (vn-link) and 802.1Qbg (VEPA) switches provide their own way (outside of

323

Virt ualiz at ion Administ rat ion G uide

libvirt) to tag guest traffic onto specific vlans.) As expected, the tag attribute specifies which vlan tag
to use. If a network has more than one <vl an> element defined, it is assumed that the user wants to
do VLAN trunking using all the specified tags. In the case that VLAN trunking with a single tag is
desired, the optional attribute trunk='yes' can be added to the VLAN element.
For network connections using openvswitch it is possible to configure the 'native-tagged' and
'native-untagged' VLAN modes. This uses the optional nativeMode attribute on the <tag > element:
nativeMode may be set to 'tagged' or 'untagged'. The id attribute of the element sets the native vlan.
<vl an> elements can also be specified in a <po rtg ro up> element, as well as directly in a domain's
<i nterface> element. In the case that a vlan tag is specified in multiple locations, the setting in
<i nterface> takes precedence, followed by the setting in the <po rtg ro up> selected by the
interface config. The <vl an> in <netwo rk> will be selected only if none is given in <po rtg ro up> or
<i nterface>.

18.15. Applying QoS t o Your Virt ual Net work

Quality of Service (QoS) refers to the resource control systems that guarantees an optimal experience
for all users on a network, making sure that there is no delay, jitter, or packet loss. QoS can be
application specific or user / group specific. Refer to Section 20.16.9.14, Quality of service for more
information.

324

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

Chapter 19. qemu-kvm Commands, Flags, and Arguments

19.1. Int roduct ion
Note
The primary objective of this chapter is to provide a list of the q emu-kvm utility commands,
flags, and arguments that are used as an emulator and a hypervisor in Red Hat Enterprise
Linux 6. This is a comprehensive summary of the options that are known to work but are to be
used at your own risk. Red Hat Enterprise Linux 6 uses KVM as an underlying virtualization
technology. The machine emulator and hypervisor used is a modified version of QEMU called
qemu-kvm. This version does not support all configuration options of the original QEMU and it
also adds some additional options.
Options n o t listed here should not be performed.

Whit e list Fo rm at
<n ame> - When used in a syntax description, this string should be replaced by user-defined
value.
[ a|b |c] - When used in a syntax description, only one of the strings separated by | is used.
When no comment is present, an option is supported with all possible values.

19.2. Basic Opt ions

This section provides information about the basic options.

Em ulat e d Machine
- M <machine-type>
- mach in e <machine-type>[,<pro perty>[= <val ue>][,. . ]]

Pro ce sso r T ype

- cp u <model>[,<FEATURE>][...]
Additional models are visible by running -cpu ? command.
O p t ero n _G 5 - AMD Opteron 63xx class CPU
O p t ero n _G 4 - AMD Opteron 62xx class CPU
O p t ero n _G 3 - AMD Opteron 23xx (AMD Opteron Gen 3)
O p t ero n _G 2 - AMD Opteron 22xx (AMD Opteron Gen 2)
O p t ero n _G 1 - AMD Opteron 240 (AMD Opteron Gen 1)
West mere - Westmere E56xx/L56xx/X56xx (Nehalem-C)

325

Virt ualiz at ion Administ rat ion G uide

H aswell - Intel Core Processor (Haswell)

San d yB rid g e - Intel Xeon E312xx (Sandy Bridge)
N eh alem - Intel Core i7 9xx (Nehalem Class Core i7)
Pen ryn - Intel Core 2 D uo P9xxx (Penryn Class Core 2)
C o n ro e - Intel Celeron_4x0 (Conroe/Merom Class Core 2)
cp u 6 4 - rh el5 - Red Hat Enterprise Linux 5 supported QEMU Virtual CPU version
cp u 6 4 - rh el6 - Red Hat Enterprise Linux 6 supported QEMU Virtual CPU version
d ef au lt - special option use default option from above.

Pro ce sso r T o po lo gy
- smp <n>[,cores=<ncores>][,threads=<nthreads>][,sockets=<nsocks>][,maxcpus=<maxcpus>]
Hypervisor and guest operating system limits on processor topology apply.

NUMA Syst e m
- n u ma <nodes>[,mem=<size>][,cpus=<cpu[-cpu>]][,nodeid=<node>]
Hypervisor and guest operating system limits on processor topology apply.

Me m o ry Size
- m <megs>
Supported values are limited by guest minimal and maximal values and hypervisor limits.

Ke ybo ard Layo ut

- k <language>

Gue st Nam e
- n ame <name>

Gue st UUID
- u u id <uuid>

19.3. Disk Opt ions

This section provides information about disk options.

Ge ne ric Drive
- d rive <option>[,<option>[,<option>[,...]]]
Supported with the following options:

326

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

read o n ly[on|off]
werro r[enospc|report|stop|ignore]
rerro r[report|stop|ignore]
id =<id>
Id of the drive has the following limitation for if=none:
ID E disk has to have <id> in following format: drive-ide0-<BUS>-<UNIT>
Example of correct format:
-drive if=none,id=drive-ide0-<BUS>-<UNIT>,... -device ide-drive,drive=drive-ide0-<BUS><UNIT>,bus=ide.<BUS>,unit=<UNIT>
f ile=<file>
Value of <file> is parsed with the following rules:
Passing floppy device as <file> is not supported.
Passing cd-rom device as <file> is supported only with cdrom media type (media=cdrom) and
only as ID E drive (either if=ide or if=none + -device ide-drive).
If <file> is neither block nor character device, it must not contain ':'.
if =<interface>
The following interfaces are supported: none, ide, virtio, floppy.
in d ex=<index>
med ia=<media>
cach e=<cache>
Supported values: none, writeback or writethrough.
co p y- o n - read =[on|off]
sn ap sh o t =[yes|no]
serial=<serial>
aio =<aio>
f o rmat =<format>
This option is not required and can be omitted. However, this is not recommended for raw images
because it represents security risk. Supported formats are:
q co w2
raw

Bo o t Opt io n
- b o o t [order=<drives>][,menu=[on|off]]

327

Virt ualiz at ion Administ rat ion G uide

Snapsho t Mo de
- sn ap sh o t

19.4 . Display Opt ions

This section provides information about display options.

Disable Graphics
- n o g rap h ic

VGA Card Em ulat io n

- vg a <type>
Supported types:
cirru s - Cirrus Logic GD 5446 Video card.
st d - Standard VGA card with Bochs VBE extensions.
q xl - Spice paravirtual card.
n o n e - D isable VGA card.

VNC Display
- vn c <display>[,<option>[,<option>[,...]]]
Supported display value:
[<host>]:<port>
unix:<path>
sh are[allow-exclusive|force-shared|ignore]
none - Supported with no other options specified.
Supported options are:
t o =<port>
reverse
p asswo rd
t ls
x509 =</path/to/certificate/dir> - Supported when t ls specified.
x509 verif y=</path/to/certificate/dir> - Supported when t ls specified.
sasl
acl

328

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

Spice De skt o p
- sp ice option[,option[,...]]
Supported options are:
p o rt =<number>
ad d r=<addr>
ip v4
ip v6
p asswo rd =<secret>
d isab le- t icket in g
d isab le- co p y- p ast e
t ls- p o rt =<number>
x509 - d ir=</path/to/certificate/dir>
x509 - key- f ile=<file>
x509 - key- p asswo rd =<file>
x509 - cert - f ile=<file>
x509 - cacert - f ile=<file>
x509 - d h - key- f ile=<file>
t ls- cip h er=<list>
t ls- ch an n el[main|display|cursor|inputs|record|playback]
p lain t ext - ch an n el[main|display|cursor|inputs|record|playback]
imag e- co mp ressio n =<compress>
jp eg - wan - co mp ressio n =<value>
z lib - g lz - wan - co mp ressio n =<value>
st reamin g - vid eo =[off|all|filter]
ag en t - mo u se=[on|off]
p layb ack- co mp ressio n =[on|off]
seamless- mig rat io =[on|off]

19.5. Net work Opt ions

This section provides information about network options.

T AP ne t wo rk
- n et d ev t ap ,id=<id>][,<options>...]

329

Virt ualiz at ion Administ rat ion G uide

The following options are supported (all use name=value format):

ifname
fd
script
downscript
sndbuf
vnet_hdr
vhost
vhostfd
vhostforce

19.6. Device Opt ions

This section provides information about device options.

Ge ne ral De vice
- d evice <driver>[,<prop>[=<value>][,...]]
All drivers support following properties
id
bus
Following drivers are supported (with available properties):
p ci- assig n
host
bootindex
configfd
addr
rombar
romfile
multifunction
If the device has multiple functions, all of them need to be assigned to the same guest.
rt l8139
mac
netdev

330

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

bootindex
addr
e1000
mac
netdev
bootindex
addr
virt io - n et - p ci
ioeventfd
vectors
indirect
event_idx
csum
guest_csum
gso
guest_tso4
guest_tso6
guest_ecn
guest_ufo
host_tso4
host_tso6
host_ecn
host_ufo
mrg_rxbuf
status
ctrl_vq
ctrl_rx
ctrl_vlan
ctrl_rx_extra
mac
netdev
bootindex

331

Virt ualiz at ion Administ rat ion G uide

x-txtimer
x-txburst
tx
addr
q xl
ram_size
vram_size
revision
cmdlog
addr
id e- d rive
unit
drive
physical_block_size
bootindex
ver
wwn
virt io - b lk- p ci
class
drive
logical_block_size
physical_block_size
min_io_size
opt_io_size
bootindex
ioeventfd
vectors
indirect_desc
event_idx
scsi
addr
virt io - scsi- p ci - tech-preview in 6.3, supported since 6.4.

332

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

For Windows guests, Windows Server 2003, which was tech-preview, is no longer supported
since 6.5. However, Windows Server 2008 and 2012, and Windows desktop 7 and 8 are fully
supported since 6.5.
vectors
indirect_desc
event_idx
num_queues
addr
isa- d eb u g co n
isa- serial
index
iobase
irq
chardev
virt serialp o rt
nr
chardev
name
virt co n so le
nr
chardev
name
virt io - serial- p ci
vectors
class
indirect_desc
event_idx
max_ports
flow_control
addr
ES1370
addr
AC 9 7

333

Virt ualiz at ion Administ rat ion G uide

addr
in t el- h d a
addr
h d a- d u p lex
cad
h d a- micro
cad
h d a- o u t p u t
cad
i6 300esb
addr
ib 700 - no properties
sg a - no properties
virt io - b allo o n - p ci
indirect_desc
event_idx
addr
u sb - t ab let
migrate
port
u sb - kb d
migrate
port
u sb - mo u se
migrate
port
u sb - ccid - supported since 6.2
port
slot
u sb - h o st - tech preview since 6.2
hostbus
hostaddr

334

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

hostport
vendorid
productid
isobufs
port
u sb - h u b - supported since 6.2
port
u sb - eh ci - tech preview since 6.2
freq
maxframes
port
u sb - st o rag e - tech preview since 6.2
drive
bootindex
serial
removable
port
u sb - red ir - tech preview for 6.3, supported since 6.4
chardev
filter
scsi- cd - tech preview for 6.3, supported since 6.4
drive
logical_block_size
physical_block_size
min_io_size
opt_io_size
bootindex
ver
serial
scsi-id
lun
channel-scsi

335

Virt ualiz at ion Administ rat ion G uide

wwn
scsi- h d -tech preview for 6.3, supported since 6.4
drive
logical_block_size
physical_block_size
min_io_size
opt_io_size
bootindex
ver
serial
scsi-id
lun
channel-scsi
wwn
scsi- b lo ck -tech preview for 6.3, supported since 6.4
drive
bootindex
scsi- d isk -tech preview for 6.3
drive=drive
logical_block_size
physical_block_size
min_io_size
opt_io_size
bootindex
ver
serial
scsi-id
lun
channel-scsi
wwn
p iix3- u sb - u h ci
p iix4 - u sb - u h ci

336

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

ccid - card - p asst h ru

Glo bal De vice Se t t ing

- g lo b al <device>.<property>=<value>
Supported devices and properties as in " General device" section with these additional devices:
isa- f d c
driveA
driveB
bootindexA
bootindexB
q xl- vg a
ram_size
vram_size
revision
cmdlog
addr

Charact e r De vice
- ch ard ev backend,id=<id>[,<options>]
Supported backends are:
n u ll,id=<id> - null device
so cket ,id=<id>,port=<port>[,host=<host>][,to=<to>][,ipv4][,ipv6][,nodelay][,server][,nowait][,telnet]
- tcp socket
so cket ,id=<id>,path=<path>[,server][,nowait][,telnet] - unix socket
f ile,id=<id>,path=<path> - trafit to file.
st d io ,id=<id> - standard i/o
sp icevmc,id=<id>,name=<name> - spice channel

Enable USB
- u sb

19.7. Linux/Mult iboot Boot

This section provides information about Linux and multiboot booting.

Ke rne l File

337

Virt ualiz at ion Administ rat ion G uide

- kern el <bzImage>
Note: multiboot images are not supported

Ram Disk
- in it rd <file>

Co m m and Line Param e t e r

- ap p en d <cmdline>

19.8. Expert Opt ions

This section provides information about expert options.

KVM Virt ualizat io n

- en ab le- kvm
QEMU-KVM supports only KVM virtualization and it is used by default if available. If -enable-kvm is
used and KVM is not available, qemu-kvm fails. However, if -enable-kvm is not used and KVM is not
available, q emu - kvm runs in TCG mode, which is not supported.

Disable Ke rne l Mo de PIT Re inje ct io n

- n o - kvm- p it - rein ject io n

No Shut do wn
- n o - sh u t d o wn

No Re bo o t
- n o - reb o o t

Se rial Po rt , Mo nit o r, QMP

- serial <dev>
- mo n it o r <dev>
- q mp <dev>
Supported devices are:
st d io - standard input/output
n u ll - null device
f ile:<filename> - output to file.
t cp :[<host>]:<port>[,server][,nowait][,nodelay] - TCP Net console.
u n ix:<path>[,server][,nowait] - Unix domain socket.

338

Chapt er 1 9 . qemu- kvm Commands, Flags, and Argument s

mo n :<dev_string> - Any device above, used to multiplex monitor too.

n o n e - disable, valid only for -serial.
ch ard ev:<id> - character device created with -chardev.

Mo nit o r Re dire ct
- mo n <chardev_id>[,mode=[readline|control]][,default=[on|off]]

Manual CPU St art

-S

RT C
- rt c [base=utc|localtime|date][,clock=host|vm][,driftfix=none|slew]

Wat chdo g
- wat ch d o g model

Wat chdo g Re act io n

- wat ch d o g - act io n <action>

Gue st Me m o ry Backing
- mem- p reallo c - mem- p at h /dev/hugepages

SMBIOS Ent ry
- smb io s type=0[,vendor=<str>][,<version=str>][,date=<str>][,release=% d.% d]
- smb io s type=1[,manufacturer=<str>][,product=<str>][,version=<str>][,serial=<str>][,uuid=<uuid>]
[,sku=<str>][,family=<str>]

19.9. Help and Informat ion Opt ions

This section provides information about help and information options.

He lp
-h
- h elp

Ve rsio n
- versio n

Audio He lp
- au d io - h elp

339

Virt ualiz at ion Administ rat ion G uide

19.10. Miscellaneous Opt ions

This section provides information about miscellaneous options.

Migrat io n
- in co min g

No De fault Co nfigurat io n
- n o d ef co n f ig
- n o d ef au lt s
Running without -nodefaults is not supported

De vice Co nfigurat io n File

- read co n f ig <file>
- writ eco n f ig <file>

Lo ade d Save d St at e
- lo ad vm <file>

34 0

Chapt er 2 0 . Manipulat ing t he Domain XML

Chapter 20. Manipulating the Domain XML

This section describes the XML format used to represent domains. Here the term domain refers to the
root <d o mai n> element required for all guest virtual machine. The domain XML has two attributes:
type specifies the hypervisor used for running the domain. The allowed values are driver specific,
but include KVM and others. i d is a unique integer identifier for the running guest virtual machine.
Inactive machines have no id value. The sections in this chapter will address the components of the
domain XML. Additional chapters in this manual may refer to this chapter when manipulation of the
domain XML is required.

20.1. General Informat ion and Met adat a

This information is in this part of the domain XML:

<domain type='xen' id='3'>

<name>fv0</name>
<uuid>4dea22b31d52d8f32516782e98ab3fa0</uuid>
<title>A short description - title - of the domain</title>
<description>Some human readable description</description>
<metadata>
<app1:foo xmlns:app1="http://app1.org/app1/">..</app1:foo>
<app2:bar xmlns:app2="http://app1.org/app2/">..</app2:bar>
</metadata>
...
</domain>
Fig u re 20.1. D o main XML met ad at a
The components of this section of the domain XML are as follows:
T ab le 20.1. G en eral met ad at a elemen t s
Elemen t

D escrip t io n

<name>

Assigns a name for the virtual machine. This

name should consist only of alpha-numeric
characters and is required to be unique within
the scope of a single host physical machine. It is
often used to form the filename for storing the
persistent configuration files.
assigns a globally unique identifier for the
virtual machine. The format must be RFC 4122
compliant, eg 3e3fce4 5-4 f53-4 fa7-bb3211f34 16 8b82b. If omitted when
defining/creating a new machine, a random
UUID is generated. It is also possible to provide
the UUID via a sysinfo specification.
ti tl e Creates space for a short description of
the domain. The title should not contain any
newlines.

<uui d >

<ti tl e>

34 1

Virt ualiz at ion Administ rat ion G uide

Elemen t

D escrip t io n

<d escri pti o n>

D ifferent from the title, This data is not used by

libvirt in any way, it can contain any information
the user wants to display.
Can be used by applications to store custom
metadata in the form of XML nodes/trees.
Applications must use custom namespaces on
their XML nodes/trees, with only one top-level
element per namespace (if the application needs
structure, they should have sub-elements to their
namespace element)

<metad ata>

20.2. Operat ing Syst em Boot ing

There are a number of different ways to boot virtual machines each with their own pros and cons.
Each one is described in the sub-sections that follow and include: BIOS bootloader, host physical
machine bootloader, and direct kernel boot.

20.2.1. BIOS Boot loader

Booting via the BIOS is available for hypervisors supporting full virtualization. In this case the BIOS
has a boot order priority (floppy, harddisk, cdrom, network) determining where to obtain/find the boot
image. The OS section of the domain XML contains the information as follows:

...
<os>
<type>hvm</type>
<loader>/usr/lib/xen/boot/hvmloader</loader>
<boot dev='hd'/>
<boot dev='cdrom'/>
<bootmenu enable='yes'/>
<smbios mode='sysinfo'/>
<bios useserial='yes' rebootTimeout='0'/>
</os>
...
Fig u re 20.2. B IO S b o o t lo ad er d o main XML
The components of this section of the domain XML are as follows:
T ab le 20.2. B IO S b o o t lo ad er elemen t s
Elemen t

34 2

D escrip t io n

Chapt er 2 0 . Manipulat ing t he Domain XML

Elemen t

D escrip t io n

<type>

Specifies the type of operating system to be

booted on the guest virtual machine. hvm
indicates that the OS is one designed to run on
bare metal, so requires full virtualization. l i nux
refers to an OS that supports the Xen 3
hypervisor guest ABI. There are also two
optional attributes, arch specifying the CPU
architecture to virtualization, and machi ne
referring to the machine type. Refer to Driver
Capabilities for more information.
refers to a piece of firmware that is used to assist
the domain creation process. It is only needed
for using Xen fully virtualized domains.
takes one of the values:fd , hd , cd ro m or
netwo rk and is used to specify the next boot
device to consider. The boot element can be
repeated multiple times to setup a priority list of
boot devices to try in turn. Multiple devices of the
same type are sorted according to their targets
while preserving the order of buses. After
defining the domain, its XML configuration
returned by libvirt (through
virD omainGetXMLD esc) lists devices in the
sorted order. Once sorted, the first device is
marked as bootable. For more information see
BIOS bootloader.
determines whether or not to enable an
interactive boot menu prompt on guest virtual
machine startup. The enabl e attribute can be
either yes or no . If not specified, the hypervisor
default is used
determines how SMBIOS information is made
visible in the guest virtual machine. The mo d e
attribute must be specified, as either emul ate
(lets the hypervisor generate all values),
ho st(copies all of Block 0 and Block 1, except
for the UUID , from the host physical machine's
SMBIOS values; the virConnectGetSysinfo call
can be used to see what values are copied), or
sysi nfo (uses the values in the sysinfo
element). If not specified, the hypervisor default
setting is used.
This element has attribute useseri al with
possible values yes or no . The attribute
enables or disables Serial Graphics Adapter
which allows users to see BIOS messages on a
serial port. Therefore, one needs to have serial
port defined. Note there is another attribute,
rebo o tT i meo ut that controls whether and
after how long the guest virtual machine should
start booting again in case the boot fails
(according to BIOS). The value is in
milliseconds with maximum of 6 5535 and
special value -1 disables the reboot.

<l o ad er>

<bo o t>

<bo o tmenu>

<smbi o s>

<bi o s>

34 3

Virt ualiz at ion Administ rat ion G uide

20.2.2. Host Physical Machine Boot loader

Hypervisors employing paravirtualization do not usually emulate a BIOS, but instead the host
physical machine is responsible for the operating system boot. This may use a pseudo-bootloader in
the host physical machine to provide an interface to choose a kernel for the guest virtual machine. An
example is pygrub with Xen.

...
<bootloader>/usr/bin/pygrub</bootloader>
<bootloader_args>--append single</bootloader_args>
...
Fig u re 20.3. H o st p h ysical mach in e b o o t lo ad er d o main XML
The components of this section of the domain XML are as follows:
T ab le 20.3. B IO S b o o t lo ad er elemen t s
Elemen t

D escrip t io n

<bo o tl o ad er>

provides a fully qualified path to the bootloader

executable in the host physical machine OS.
This bootloader will choose which kernel to
boot. The required output of the bootloader is
dependent on the hypervisor in use.
allows command line arguments to be passed to
the bootloader (optional command)

<bo o tl o ad er_arg s>

20.2.3. Direct kernel boot

When installing a new guest virtual machine OS, it is often useful to boot directly from a kernel and
initrd stored in the host physical machine OS, allowing command line arguments to be passed
directly to the installer. This capability is usually available for both para and full virtualized guest
virtual machines.

...
<os>
<type>hvm</type>
<loader>/usr/lib/xen/boot/hvmloader</loader>
<kernel>/root/f8-i386-vmlinuz</kernel>
<initrd>/root/f8-i386-initrd</initrd>
<cmdline>console=ttyS0 ks=http://example.com/f8-i386/os/</cmdline>
<dtb>/root/ppc.dtb</dtb>
</os>
...

Fig u re 20.4 . D irect K ern el B o o t

34 4

Chapt er 2 0 . Manipulat ing t he Domain XML

The components of this section of the domain XML are as follows:

T ab le 20.4 . D irect kern el b o o t elemen t s
Elemen t

D escrip t io n

<type>
<l o ad er>
<kernel >

same as described in the BIOS boot section

same as described in the BIOS boot section
specifies the fully-qualified path to the kernel
image in the host physical machine OS
specifies the fully-qualified path to the (optional)
ramdisk image in the host physical machine OS.
specifies arguments to be passed to the kernel
(or installer) at boot time. This is often used to
specify an alternate primary console (eg serial
port), or the installation media source / kickstart
file

<i ni trd >

<cmd l i ne>

20.3. SMBIOS Syst em Informat ion

Some hypervisors allow control over what system information is presented to the guest virtual
machine (for example, SMBIOS fields can be populated by a hypervisor and inspected via the
dmidecode command in the guest virtual machine). The optional sysinfo element covers all such
categories of information.

...
<os>
<smbios mode='sysinfo'/>
...
</os>
<sysinfo type='smbios'>
<bios>
<entry name='vendor'>LENOVO</entry>
</bios>
<system>
<entry name='manufacturer'>Fedora</entry>
<entry name='vendor'>Virt-Manager</entry>
</system>
</sysinfo>
...
Fig u re 20.5. SMB IO S syst em in f o rmat io n
The <sysi nfo > element has a mandatory attribute type that determines the layout of sub-elements,
and may be defined as follows:
smbi o s - Sub-elements call out specific SMBIOS values, which will affect the guest virtual
machine if used in conjunction with the smbios sub-element of the <o s> element. Each subelement of sysinfo names a SMBIOS block, and within those elements can be a list of entry
elements that describe a field within the block. The following blocks and entries are recognized:

34 5

Virt ualiz at ion Administ rat ion G uide

bi o s - This is block 0 of SMBIOS, with entry names drawn from vend o r, versi o n, d ate, and
rel ease.
<system> - This is block 1 of SMBIOS, with entry names drawn from manufacturer,
pro d uct, versi o n, seri al , uui d , sku, and fami l y. If a uui d entry is provided alongside
a top-level uuid element, the two values must match.

20.4 . CPU Allocat ion

<domain>
...
<vcpu placement='static' cpuset="1-4,^3,6" current="1">2</vcpu>
...
</domain>
Fig u re 20.6 . C PU allo cat io n
The <cpu> element defines the maximum number of virtual CPUs (vCPUs) allocated for the guest
virtual machine operating system, which must be between 1 and the maximum supported by the
hypervisor. This element can contain an optional cpuset attribute, which is a comma-separated list
of physical CPU numbers that domain processes and virtual CPUs can be pinned to by default.
Note that the pinning policy of domain processes and virtual CPUs can be specified separately by
using the cputune attribute. If the emul ato rpi n attribute is specified in <cputune>, the cpuset
value specified by <vcpu> will be ignored.
Similarly, virtual CPUs that have set a value for vcpupi n cause cpuset settings to be ignored.
Virtual CPUs where vcpupi n is not specified will be pinned to the physical CPUs specified by
cpuset. Each element in the cpuset list is either a single CPU number, a range of CPU numbers, or
a caret (^) followed by a CPU number to be excluded from a previous range. The attribute current
can be used to specify whether fewer than the maximum number of virtual CPUs should be enabled.
The optional attribute pl acement can be used to specify the CPU placement mode for the domain
process. pl acement can be set as either stati c or auto . If you set <vcpu pl acement= ' auto ' >,
the system will query numad and use the settings specified in the <numatune> tag, and ignore any
other settings in <vcpu> . If you set <vcpu pl acement= ' stati c' >, the system will use the settings
specified in the <vcpu pl acement> tag instead of the settings in <numatune>.

20.5. CPU T uning

<domain>
...
<cputune>
<vcpupin vcpu="0" cpuset="1-4,^2"/>
<vcpupin vcpu="1" cpuset="0,1"/>
<vcpupin vcpu="2" cpuset="2,3"/>
<vcpupin vcpu="3" cpuset="0,4"/>
<emulatorpin cpuset="1-3"/>

34 6

Chapt er 2 0 . Manipulat ing t he Domain XML

<shares>2048</shares>
<period>1000000</period>
<quota>-1</quota>
<emulator_period>1000000</emulator_period>
<emulator_quota>-1</emulator_quota>
</cputune>
...
</domain>
Fig u re 20.7. C PU t u n in g
Although all are optional, the components of this section of the domain XML are as follows:
T ab le 20.5. C PU t u n in g elemen t s
Elemen t

D escrip t io n

<cputune>

Provides details regarding the CPU tunable

parameters for the domain. This is optional.
Specifies which of host physical machine's
physical CPUs the domain VCPU will be pinned
to. If this is omitted, and attribute cpuset of
element <vcpu> is not specified, the vCPU is
pinned to all the physical CPUs by default. It
contains two required attributes, the attribute
vcpu specifies i d , and the attribute cpuset is
same as attribute cpuset of element <vcpu>.
Specifies which of the host physical machine
CPUs, the " emulator" , a subset of a domains not
including vcpu, will be pinned to. If this is
omitted, and attribute cpuset of element
<vcpu> is not specified, the " emulator" is
pinned to all the physical CPUs by default. It
contains one required attribute cpuset
specifying which physical CPUs to pin to.
emul ato rpi n is not allowed if attribute
pl acement of element <vcpu> is auto .
Specifies the proportional weighted share for the
domain. If this is omitted, it defaults to the default
value inherent in the operating system. If there is
no unit for the value, it is calculated relative to
the setting of other guest virtual machines. For
example, if a guest virtual machine is configured
with value of 2048, it will get twice as much
processing time as a guest virtual machine
configured with value of 1024.
Specifies the enforcement interval in
microseconds. By using peri o d , each of the
domain's vcpu will not be allowed to consume
more than its allotted quota worth of run time.
This value should be within the following range:
10 0 0 -10 0 0 0 0 0 . A peri o d > with a value of 0
means no value.

<vcpupi n>

<emul ato rpi n>

<shares>

<peri o d >

34 7

Virt ualiz at ion Administ rat ion G uide

Elemen t

D escrip t io n

<q uo ta>

Specifies the maximum allowed bandwidth in

microseconds. A domain with q uo ta as any
negative value indicates that the domain has
infinite bandwidth, which means that it is not
bandwidth controlled. The value should be
within the following range:10 0 0 184 4 6 74 4 0 7370 9 551 or less than 0 . A q uo ta
with value of 0 means no value. You can use
this feature to ensure that all vcpus run at the
same speed.
Specifies the enforcement interval in
microseconds. Within an
<emul ato r_peri o d >, emulator threads (those
excluding vcpus) of the domain will not be
allowed to consume more than the
<emul ato r_q uo ta> worth of run time. The
<emul ato r_peri o d > value should be in the
following range: 10 0 0 - 10 0 0 0 0 0 . An
<emul ato r_peri o d > with value of 0 , means
no value.
Specifies the maximum allowed bandwidth in
microseconds for the domain's emulator threads
(those excluding vcpus). A domain with an
<emul ato r_q uo ta> as a negative value
indicates that the domain has infinite bandwidth
for emulator threads (those excluding vcpus),
which means that it is not bandwidth controlled.
The value should be in the following range:
10 0 0 - 184 4 6 74 4 0 7370 9 551, or less than
0 . An <emul ato r_q uo ta> with value 0 means
no value.

<emul ato r_peri o d >

<emul ato r_q uo ta>

20.6. Memory Backing

Memory backing allows the hypervisor to properly manage large pages within the guest virtual
machine.
The optional <memo ryBacki ng > element may have an <hug epag es> element set within it. This tells
the hypervisor that the guest virtual machine should have its memory allocated using hugepages
instead of the normal native page size.

<domain>
...
<memoryBacking>
<hugepages/>
</memoryBacking>
...
</domain>
Fig u re 20.8. Memo ry b ackin g

34 8

Chapt er 2 0 . Manipulat ing t he Domain XML

20.7. Memory t uning

<domain>
...
<memtune>
<hard_limit unit='G'>1</hard_limit>
<soft_limit unit='M'>128</soft_limit>
<swap_hard_limit unit='G'>2</swap_hard_limit>
<min_guarantee unit='bytes'>67108864</min_guarantee>
</memtune>
...
</domain>
Fig u re 20.9 . Memo ry T u n in g
Although all are optional, the components of this section of the domain XML are as follows:
T ab le 20.6 . Memo ry t u n in g elemen t s
Elemen t

D escrip t io n

<memtune>

Provides details regarding the memory tunable

parameters for the domain. If this is omitted, it
defaults to the OS provided defaults. The
parameters are applied to the process as a
whole therefore when setting limits, one needs to
add up guest virtual machine RAM, guest virtual
machine video RAM, and allow for some memory
overhead. The last piece is hard to determine so
one use trial and error. For each tunable, it is
possible to designate which unit the number is
in on input, using the same values as for
<memo ry>. For backwards compatibility, output
is always in KiB.
This is the maximum memory the guest virtual
machine can use. The uni t for this value is
expressed in ki bi bytes (i.e. blocks of 1024
bytes)
This is the memory limit to enforce during
memory contention. The uni t for this value is
expressed in kibibytes (i.e. blocks of 1024 bytes)
This is the maximum memory plus swap the
guest virtual machine can use. The uni t for this
value is expressed in kibibytes (i.e. blocks of
1024 bytes). This has to be more than
<hard _l i mi t> value provided
This is the guaranteed minimum memory
allocation for the guest virtual machine. The
units for this value is expressed in kibibytes (i.e.
blocks of 1024 bytes)

<hard _l i mi t>

<so ft_l i mi t>

<swap_hard _l i mi t>

<mi n_g uarantee>

20.8. NUMA Node T uning

34 9

Virt ualiz at ion Administ rat ion G uide

20.8. NUMA Node T uning

Once NUMA node tuning is done using conventional management tools the following domain XML
parameters are effected:

>
<domain>
...
<numatune>
<memory mode="strict" nodeset="1-4,^3"/>
</numatune>
...
</domain>
Fig u re 20.10. N U MA n o d e t u n in g
Although all are optional, the components of this section of the domain XML are as follows:
T ab le 20.7. N U MA n o d e t u n in g elemen t s
Elemen t

D escrip t io n

<numatune>

Provides details of how to tune the performance

of a NUMA host physical machine via
controlling NUMA policy for domain process.
Specifies how to allocate memory for the domain
process on a NUMA host physical machine. It
contains several optional attributes. Attribute
mo d e is either i nterl eave, stri ct, or
preferred . If no value is given it defaults to
stri ct. Attribute no d eset specifies the NUMA
nodes, using the same syntax as attribute
cpuset of element <vcpu>. Attribute
pl acement can be used to indicate the memory
placement mode for the domain process. Its
value can be either stati c or auto . If attribute
<no d eset> is specified it defaults to the
<pl acement> of <vcpu>, or stati c. auto
indicates the domain process will only allocate
memory from the advisory nodeset returned from
querying numad and the value of attribute
nodeset will be ignored if it's specified. If
attribute pl acement of vcpu is auto , and
attribute <numatune> is not specified, a default
numatune with <pl acement> auto and mode
stri ct will be added implicitly.

<memo ry>

20.9. Block I/O t uning

350

Chapt er 2 0 . Manipulat ing t he Domain XML

<domain>
...
<blkiotune>
<weight>800</weight>
<device>
<path>/dev/sda</path>
<weight>1000</weight>
</device>
<device>
<path>/dev/sdb</path>
<weight>500</weight>
</device>
</blkiotune>
...
</domain>
Fig u re 20.11. B lo ck I/O T u n in g
Although all are optional, the components of this section of the domain XML are as follows:
T ab le 20.8. B lo ck I/O t u n in g elemen t s
Elemen t

D escrip t io n

<bl ki o tune>

This optional element provides the ability to tune

Blkio cgroup tunable parameters for the domain.
If this is omitted, it defaults to the OS provided
defaults.
This optional weight element is the overall I/O
weight of the guest virtual machine. The value
should be within the range 100 - 1000.
The domain may have multiple <d evi ce>
elements that further tune the weights for each
host physical machine block device in use by
the domain. Note that multiple guest virtual
machine disks can share a single host physical
machine block device. In addition, as they are
backed by files within the same host physical
machine file system, this tuning parameter is at
the global domain level, rather than being
associated with each guest virtual machine disk
device (contrast this to the <i o tune> element
which can be applied to a single <d i sk>). Each
device element has two mandatory subelements, <path> describing the absolute path
of the device, and <wei g ht> giving the relative
weight of that device, which has an acceptable
range of 100 - 1000.

<wei g ht>

<d evi ce>

20.10. Resource Part it ioning

Hypervisors may allow for virtual machines to be placed into resource partitions, potentially with
nesting of said partitions. The <reso urce> element groups together configuration related to
resource partitioning. It currently supports a child element partition whose content defines the path of

351

Virt ualiz at ion Administ rat ion G uide

the resource partition in which to place the domain. If no partition is listed, then the domain will be
placed in a default partition. It is the responsibility of the app/admin to ensure that the partition exists
prior to starting the guest virtual machine. Only the (hypervisor specific) default partition can be
assumed to exist by default.

<resource>
<partition>/virtualmachines/production</partition>
</resource>
Fig u re 20.12. R eso u rce p art it io n in g
Resource partitions are currently supported by the QEMU and LXC drivers, which map partition paths
to cgroups directories in all mounted controllers.

20.11. CPU Model and T opology

This section covers the requirements for CPU model. Note that every hypervisor has its own policy for
which CPU features guest will see by default. The set of CPU features presented to the guest by
QEMU/KVM depends on the CPU model chosen in the guest virtual machine configuration. q emu32
and q emu6 4 are basic CPU models but there are other models (with additional features) available.
Each model and its topology is specified using the following elements from the domain XML:

<cpu match='exact'>
<model fallback='allow'>core2duo</model>
<vendor>Intel</vendor>
<topology sockets='1' cores='2' threads='1'/>
<feature policy='disable' name='lahf_lm'/>
</cpu>
Fig u re 20.13. C PU mo d el an d t o p o lo g y examp le 1

<cpu mode='host-model'>
<model fallback='forbid'/>
<topology sockets='1' cores='2' threads='1'/>
</cpu>
Fig u re 20.14 . C PU mo d el an d t o p o lo g y examp le 2

<cpu mode='host-passthrough'/>
Fig u re 20.15. C PU mo d el an d t o p o lo g y examp le 3

352

Chapt er 2 0 . Manipulat ing t he Domain XML

In cases where no restrictions are to be put on either the CPU model nor its features, a simpler cpu
element such as the following may be used.

<cpu>
<topology sockets='1' cores='2' threads='1'/>
</cpu>
Fig u re 20.16 . C PU mo d el an d t o p o lo g y examp le 4
The components of this section of the domain XML are as follows:
T ab le 20.9 . C PU mo d el an d t o p o lo g y elemen t s
Elemen t

D escrip t io n

<cpu>

This element contains all parameters for the

vCPU feature set.
Specifies how closely the features indicated in
the <cpu> element must match the vCPUs that
are available. The match attribute can be
omitted if <to po l o g y> is the only element
nested in the <cpu> element. Possible values for
the match attribute are:

<match>

mi ni mum - The features listed are the

minimum requirement. There may be more
features available in the vCPU then are
indicated, but this is the minimum that will be
accepted. This value will fail if the minimum
requirements are not met.
exact - the virtual CPU provided to the guest
virtual machine must exactly match the
features specified. If no match is found, an
error will result.
stri ct - the guest virtual machine will not be
created unless the host physical machine
CPU exactly matches the specification.
If the match attribute is omitted from the <cpu>
element, the default setting match= ' exact' is
used.

353

Virt ualiz at ion Administ rat ion G uide

Elemen t

D escrip t io n

<mo d e>

This optional attribute may be used to make it

easier to configure a guest virtual machine CPU
to be as close to the host physical machine CPU
as possible. Possible values for the mode
attribute are:
custo m - describes how the CPU is
presented to the guest virtual machine. This
is the default setting when the mo d e attribute
is not specified. This mode makes it so that a
persistent guest virtual machine will see the
same hardware no matter what host physical
machine the guest virtual machine is booted
on.
ho st-mo d el - this is essentially a shortcut
to copying host physical machine CPU
definition from the capabilities XML into the
domain XML. As the CPU definition is copied
just before starting a domain, the same XML
can be used on different host physical
machines while still providing the best guest
virtual machine CPU each host physical
machine supports. Neither the match
attribute nor any feature elements can be
used in this mode. For more information see
libvirt domain XML CPU models
ho st-passthro ug h With this mode, the
CPU visible to the guest virtual machine is
exactly the same as the host physical
machine CPU including elements that cause
errors within libvirt. The obvious the
downside of this mode is that the guest
virtual machine environment cannot be
reproduced on different hardware and
therefore this mode is recommended with
great caution. Neither mo d el nor feature
elements are allowed in this mode.
Note that in both ho st-mo d el and ho stpassthro ug h mode, the real (approximate
in host-passthrough mode) CPU definition
which would be used on current host
physical machine can be determined by
specifying VIR_D OMAIN_XML_UPD ATE_CPU
flag when calling virD omainGetXMLD esc
API. When running a guest virtual machine
that might be prone to operating system
reactivation when presented with different
hardware, and which will be migrated
between host physical machines with
different capabilities, you can use this output
to rewrite XML to the custom mode for more
robust migration.

354

Chapt er 2 0 . Manipulat ing t he Domain XML

Elemen t
<mo d el >

<vend o r>

<to po l o g y>

<feature>

D escrip t io n
Specifies CPU model requested by the guest
virtual machine. The list of available CPU
models and their definition can be found in
cpu_map. xml file installed in libvirt's data
directory. If a hypervisor is not able to use the
exact CPU model, libvirt automatically falls back
to a closest model supported by the hypervisor
while maintaining the list of CPU features. An
optional fal l back attribute can be used to
forbid this behavior, in which case an attempt to
start a domain requesting an unsupported CPU
model will fail. Supported values for fallback
attribute are: al l o w (this is the default), and
fo rbi d . The optional vend o r_i d attribute can
be used to set the vendor id seen by the guest
virtual machine. It must be exactly 12 characters
long. If not set, the vendor id of the host physical
machine is used. Typical possible values are
Authenti cAMD and G enui neIntel .
Specifies CPU vendor requested by the guest
virtual machine. If this element is missing, the
guest virtual machine runs on a CPU matching
given features regardless of its vendor. The list
of supported vendors can be found in
cpu_map. xml .
Specifies requested topology of virtual CPU
provided to the guest virtual machine. Three
non-zero values have to be given for sockets,
cores, and threads: total number of CPU
sockets, number of cores per socket, and
number of threads per core, respectively.
Can contain zero or more elements used to finetune features provided by the selected CPU
model. The list of known feature names can be
found in the same file as CPU models. The
meaning of each feature element depends on its
policy attribute, which has to be set to one of the
following values:
fo rce - forces the virtual to be supported
regardless of whether it is actually supported
by host physical machine CPU.
req ui re - dictates that guest virtual
machine creation will fail unless the feature is
supported by host physical machine CPU.
This is the default setting
o pti o nal - this feature is supported by
virtual CPU but and only if it is supported by
host physical machine CPU.
d i sabl e - this is not supported by virtual
CPU.
fo rbi d - guest virtual machine creation will
fail if the feature is supported by host
physical machine CPU.

355

Virt ualiz at ion Administ rat ion G uide

20.11.1. Guest virt ual machine NUMA t opology

Guest virtual machine NUMA topology can be specified using the <numa> element and the following
from the domain XML:

<cpu>
<numa>
<cell cpus='0-3' memory='512000'/>
<cell cpus='4-7' memory='512000'/>
</numa>
</cpu>
...
Fig u re 20.17. G u est Virt u al Mach in e N U MA T o p o lo g y
Each cell element specifies a NUMA cell or a NUMA node. cpus specifies the CPU or range of CPUs
that are part of the node. memo ry specifies the node memory in kibibytes (i.e. blocks of 1024 bytes).
Each cell or node is assigned cel l i d or no d ei d in increasing order starting from 0.

20.12. Event s Configurat ion

Using the following sections of domain XML it is possible to override the default actions taken on
various events.

<on_poweroff>destroy</on_poweroff>
<on_reboot>restart</on_reboot>
<on_crash>restart</on_crash>
<on_lockfailure>poweroff</on_lockfailure>
Fig u re 20.18. Even t s co n f ig u rat io n
The following collections of elements allow the actions to be specified when a guest virtual machine
OS triggers a life cycle operation. A common use case is to force a reboot to be treated as a poweroff
when doing the initial OS installation. This allows the VM to be re-configured for the first post-install
bootup.
The components of this section of the domain XML are as follows:
T ab le 20.10. Even t co n f ig u rat io n elemen t s
St at e

356

D escrip t io n

Chapt er 2 0 . Manipulat ing t he Domain XML

St at e

D escrip t io n

<o n_po wero ff>

Specifies the action that is to be executed when

the guest virtual machine requests a poweroff.
Four possible arguements are possible:
d estro y - this action terminates the domain
completely and releases all resources
restart - this action terminates the domain
completely and restarts it with the same
configuration
preserve - this action terminates the domain
completely but and its resources are
preserved to allow for future analysis.
rename-restart - this action terminates the
domain completely and then restarts it with a
new name

<o n_rebo o t>

Specifies the action that is to be executed when

the guest virtual machine requests a reboot.Four
possible arguements are possible:
d estro y - this action terminates the domain
completely and releases all resources
restart - this action terminates the domain
completely and restarts it with the same
configuration
preserve - this action terminates the domain
completely but and its resources are
preserved to allow for future analysis.
rename-restart - this action terminates the
domain completely and then restarts it with a
new name

357

Virt ualiz at ion Administ rat ion G uide

St at e

D escrip t io n

<o n_crash>

Specifies the action that is to be executed when

the guest virtual machine crashes. In addition, it
supports these additional actions:
co red ump-d estro y - the crashed domain's
core is dumped, domain is terminated
completely, and all resources are released.
co red ump-restart - the crashed domain's
core is dumped, and the domain is restarted
with the same configuration settings
Four possible arguements are possible:
d estro y - this action terminates the domain
completely and releases all resources
restart - this action terminates the domain
completely and restarts it with the same
configuration
preserve - this action terminates the domain
completely but and its resources are
preserved to allow for future analysis.
rename-restart - this action terminates the
domain completely and then restarts it with a
new name

<o n_l o ckfai l ure>

Specifies what action should be taken when a

lock manager loses resource locks. The
following actions are recognized by libvirt,
although not all of them need to be supported
by individual lock managers. When no action is
specified, each lock manager will take its default
action. The following arguments are possible:
po wero ff - forcefully powers off the domain
restart - restarts the domain to reacquire its
locks.
pause - pauses the domain so that it can be
manually resumed when lock issues are
solved.
i g no re - keeps the domain running as if
nothing happened.

20.13. Power Management

It is possible to forcibly enable or disable BIOS advertisements to the guest virtual machine OS using
conventional management tools which effects the following section of the domain XML:

...
<pm>
<suspend-to-disk enabled='no'/>

358

Chapt er 2 0 . Manipulat ing t he Domain XML

<suspend-to-mem enabled='yes'/>
</pm>
...
Fig u re 20.19 . Po wer man ag emen t
The <pm> element can be enabled using the arguement yes or disabled using the argument no .
BIOS support can be implemented for S3 using the argument suspend -to -d i sk and S4 using the
argument suspend -to -mem ACPI sleep states. If nothing is specified, the hypervisor will be left with
its default value.

20.14 . Hypervisor Feat ures

Hypervisors may allow certain CPU / machine features to be enabled (state= ' o n' ) or disabled
(state= ' o ff' ).

...
<features>
<pae/>
<acpi/>
<apic/>
<hap/>
<privnet/>
<hyperv>
<relaxed state='on'/>
</hyperv>
</features>
...

Fig u re 20.20. H yp erviso r f eat u res

All features are listed within the <features> element, if a <state> is not specified it is disabled. The
available features can be found by calling the capabi l i ti es XML, but a common set for fully
virtualized domains are:
T ab le 20.11. H yp erviso r f eat u res elemen t s
St at e

D escrip t io n

<pae>

Physical address extension mode allows 32-bit

guest virtual machines to address more than 4
GB of memory.
Useful for power management, for example, with
KVM guest virtual machines it is required for
graceful shutdown to work.

<acpi >

359

Virt ualiz at ion Administ rat ion G uide

St at e

D escrip t io n

<api c>

Allows the use of programmable IRQ

management. For this element, there is an
optional attribute eo i with values o n and o ff
which sets the availability of EOI (End of
Interrupt) for the guest virtual machine.
Enables the use of Hardware Assisted Paging if
it is available in the hardware.
Enables various features to improve the
behavior of guest virtual machines running
Microsoft Windows. Using the optional attribute
rel axed with values o n or o ff enables or
disables the relax constraints on timers

<hap>
hyperv

20.15. T imekeeping
The guest virtual machine clock is typically initialized from the host physical machine clock. Most
operating systems expect the hardware clock to be kept in UTC, which is the default setting. Note that
for Windows guest virtual machines the guest virtual machine must be set in l o cal ti me.

...
<clock offset='localtime'>
<timer name='rtc' tickpolicy='catchup' track='guest'>
<catchup threshold='123' slew='120' limit='10000'/>
</timer>
<timer name='pit' tickpolicy='delay'/>
</clock>
...
Fig u re 20.21. T imekeep in g
The components of this section of the domain XML are as follows:
T ab le 20.12. T ime keep in g elemen t s
St at e

360

D escrip t io n

Chapt er 2 0 . Manipulat ing t he Domain XML

St at e

D escrip t io n

<cl o ck>

The o ffset attribute takes four possible values,

allowing for fine grained control over how the
guest virtual machine clock is synchronized to
the host physical machine. Note that
hypervisors are not required to support all
policies across all time sources
utc - Synchronizes the clock to UTC when
booted. utc mode can be converted to
vari abl e mode, which can be controlled by
using the adjustment attribute. If the value is
reset, the conversion is not done. A numeric
value forces the conversion to vari abl e
mode using the value as the initial
adjustment. The default adjustment is
hypervisor specific.
l o cal ti me - Synchronizes the guest virtual
machine clock with the host physical
machine's configured timezone when booted.
The adjustment attribute behaves the same
as in 'utc' mode.
ti mezo ne - Synchronizes the guest virtual
machine clock to the requested timezone
using the timezone attribute.
vari abl e - Gives the guest virtual machine
clock an arbitrary offset applied relative to
UTC or localtime, depending on the basis
attribute. The delta relative to UTC (or
localtime) is specified in seconds, using the
ad justment attribute. The guest virtual
machine is free to adjust the RTC over time
and expect that it will be honored at next
reboot. This is in contrast to utc and
l o cal ti me mode (with the optional attribute
ad justment= ' reset' ), where the RTC
adjustments are lost at each reboot. In
addition the basi s attribute can be either
utc (default) or l o cal ti me. The cl o ck
element may have zero or more <ti mer>
elements.

<ti mer>
<freq uency>
<mo d e>

<present>

See Note
This is an unsigned integer specifying the
frequency at which name="tsc" runs.
The mo d e attribute controls how the
name= "tsc" <ti mer> is managed, and can be
set to: auto , nati ve, emul ate, paravi rt, or
smpsafe. Other timers are always emulated.
Specifies whether a particular timer is available
to the guest virtual machine. Can be set to yes
or no

361

Virt ualiz at ion Administ rat ion G uide

Note
Each <ti mer> element must contain a name attribute, and may have the following attributes
depending on the name specified.
<name> - selects which ti mer is being modified. The following values are
acceptable:kvmcl o ck (QEMU-KVM), pi t(QEMU-KVM), or rtc(QEMU-KVM), or tsc(libxl
only). Note that pl atfo rm is currently unsupported.
track - specifies the timer track. The following values are acceptable: bo o t, g uest, or
wal l . track is only valid for name= "rtc".
ti ckpo l i cy - determines what happens whens the deadline for injecting a tick to the
guest virtual machine is missed. The following values can be assigned:
d el ay -will continue to deliver ticks at the normal rate. The guest virtual machine time
will be delayed due to the late tick
catchup - delivers ticks at a higher rate in order to catch up with the missed tick. The
guest virtual machine time is not displayed once catchup is complete. In addition, there
can be three optional attributes, each a positive integer, as follows: threshold, slew, and
limit.
merg e - merges the missed tick(s) into one tick and injects them. The guest virtual
machine time may be delayed, depending on how the merge is done.
d i scard - throws away the missed tick(s) and continues with future injection at its
default interval setting. The guest virtual machine time may be delayed, unless there is
an explicit statement for handling lost ticks

20.16. Devices
This set of XML elements are all used to describe devices provided to the guest virtual machine
domain. All of the devices below are indicated as children of the main devices element.
The following virtual devices are supported:
virtio-scsi-pci - PCI bus storage device
virtio-9p-pci - PCI bus storage device
virtio-blk-pci - PCI bus storage device
virtio-net-pci - PCI bus network device also known as virtio-net
virtio-serial-pci - PCI bus input device
virtio-balloon-pci - PCI bus memory balloon device
virtio-rng-pci - PCI bus virtual random number generator device

362

Chapt er 2 0 . Manipulat ing t he Domain XML

Important
If a virtio device is created where the number of vectors is set to a value higher than 32, the
device behaves as if it was set to a zero value on Red Hat Enterprise Linux 6, but not on
Enterprise Linux 7. The resulting vector setting mismatch causes a migration error if the
number of vectors on any virtio device on either platform is set to 33 or higher. It is therefore
not reccomended to set the vecto r value to be greater than 32. All virtio devices with the
exception of virtio-balloon-pci and virtio-rng-pci will accept a vecto r argument.

...
<devices>
<emulator>/usr/lib/xen/bin/qemu-dm</emulator>
</devices>
...
Fig u re 20.22. D evices - ch ild elemen t s
The contents of the <emul ato r> element specify the fully qualified path to the device model emulator
binary. The capabilities XML specifies the recommended default emulator to use for each particular
domain type or architecture combination.

20.16.1. Hard Drives, Floppy Disks, CDROMs

This section of the domain XML specifies any device that looks like a disk, be it a floppy, harddisk,
cdrom, or paravirtualized driver is specified via the disk element.

...
<devices>
<disk type='file' snapshot='external'>
<driver name="tap" type="aio" cache="default"/>
<source file='/var/lib/xen/images/fv0' startupPolicy='optional'>
<seclabel relabel='no'/>
</source>
<target dev='hda' bus='ide'/>
<iotune>
<total_bytes_sec>10000000</total_bytes_sec>
<read_iops_sec>400000</read_iops_sec>
<write_iops_sec>100000</write_iops_sec>
</iotune>
<boot order='2'/>
<encryption type='...'>
...
</encryption>
<shareable/>
<serial>
...

363

Virt ualiz at ion Administ rat ion G uide

</serial>
</disk>
...
<disk type='network'>
<driver name="qemu" type="raw" io="threads" ioeventfd="on"
event_idx="off"/>
<source protocol="sheepdog" name="image_name">
<host name="hostname" port="7000"/>
</source>
<target dev="hdb" bus="ide"/>
<boot order='1'/>
<transient/>
<address type='drive' controller='0' bus='1' unit='0'/>
</disk>
<disk type='network'>
<driver name="qemu" type="raw"/>
<source protocol="rbd" name="image_name2">
<host name="hostname" port="7000"/>
</source>
<target dev="hdd" bus="ide"/>
<auth username='myuser'>
<secret type='ceph' usage='mypassid'/>
</auth>
</disk>
<disk type='block' device='cdrom'>
<driver name='qemu' type='raw'/>
<target dev='hdc' bus='ide' tray='open'/>
<readonly/>
</disk>
<disk type='block' device='lun'>
<driver name='qemu' type='raw'/>
<source dev='/dev/sda'/>
<target dev='sda' bus='scsi'/>
<address type='drive' controller='0' bus='0' target='3' unit='0'/>
</disk>
<disk type='block' device='disk'>
<driver name='qemu' type='raw'/>
<source dev='/dev/sda'/>
<geometry cyls='16383' heads='16' secs='63' trans='lba'/>
<blockio logical_block_size='512' physical_block_size='4096'/>
<target dev='hda' bus='ide'/>
</disk>
<disk type='volume' device='disk'>
<driver name='qemu' type='raw'/>
<source pool='blk-pool0' volume='blk-pool0-vol0'/>
<target dev='hda' bus='ide'/>
</disk>
</devices>
...
Fig u re 20.23. D evices - H ard d rives, f lo p p y d isks, C D R O Ms

2 0 .1 6 .1 .1 . Disk e le m e nt

364

Chapt er 2 0 . Manipulat ing t he Domain XML

The <d i sk> element is the main container for describing disks. The attribute type can be used with
the <d i sk> element. The following types are allowed:
fi l e
bl o ck
dir
netwo rk
For more information, see D isk Elements

2 0 .1 6 .1 .2 . So urce e le m e nt
If the <d i sk type= ' fi l e' ' >, then the fi l e attribute specifies the fully-qualified path to the file
holding the disk. If the <d i sk type= ' bl o ck' >, then the d ev attribute specifies the path to the host
physical machine device to serve as the disk. With both fi l e and bl o ck, one or more optional subelements secl abel , described below, can be used to override the domain security labeling policy
for just that source file. If the disk type is d i r, then the d i r attribute specifies the fully-qualified path
to the directory to use as the disk. If the disk type is netwo rk, then the protocol attribute specifies the
protocol to access to the requested image; possible values are nbd , rbd , sheepd o g or g l uster.
If the protocol attribute is rbd , sheepd o g or g l uster, an additional attribute name is mandatory to
specify which volume and or image will be used. When the disk type is netwo rk, the so urce may
have zero or more ho st sub-elements used to specify the host physical machines to connect,
including: type= ' d i r' and type= ' netwo rk' . For a fi l e disk type which represents a cdrom or
floppy (the device attribute), it is possible to define policy what to do with the disk if the source file is
not accessible. This is done by manipulating the startupP o l i cy attribute, with the following
values:
mand ato ry causes a failure if missing for any reason. This is the default setting.
req ui si te causes a failure if missing on boot up, drops if missing on migrate/restore/revert
o pti o nal drops if missing at any start attempt

2 0 .1 6 .1 .3. Mirro r e le m e nt
This element is present if the hypervisor has started a Bl o ckC o py operation, where the <mi rro r>
location in the attribute file will eventually have the same contents as the source, and with the file
format in attribute format (which might differ from the format of the source). If an attribute ready is
present, then it is known the disk is ready to pivot; otherwise, the disk is probably still copying. For
now, this element only valid in output; it is ignored on input.

2 0 .1 6 .1 .4 . T arge t e le m e nt
The <targ et> element controls the bus / device under which the disk is exposed to the guest virtual
machine OS. The dev attribute indicates the logical device name. The actual device name specified is
not guaranteed to map to the device name in the guest virtual machine OS. The optional bus attribute
specifies the type of disk device to emulate; possible values are driver specific, with typical values
being i d e, scsi , vi rti o , xen, usb or sata. If omitted, the bus type is inferred from the style of the
device name. eg, a device named ' sd a' will typically be exported using a SCSI bus. The optional
attribute tray indicates the tray status of the removable disks (i.e. CD ROM or Floppy disk), the value
can be either o pen or cl o sed . The default setting is cl o sed . For more information, see target
Elements

365

Virt ualiz at ion Administ rat ion G uide

2 0 .1 6 .1 .5 . io t une
The optional <i o tune> element provides the ability to provide additional per-device I/O tuning, with
values that can vary for each device (contrast this to the bl ki o tune element, which applies globally
to the domain). This element has the following optional sub-elements. Note that any sub-element not
specified or at all or specified with a value of 0 implies no limit.
<to tal _bytes_sec> - the total throughput limit in bytes per second. This element cannot be
used with <read _bytes_sec> or <wri te_bytes_sec>.
<read _bytes_sec> - the read throughput limit in bytes per second.
<wri te_bytes_sec> - the write throughput limit in bytes per second.
<to tal _i o ps_sec> - the total I/O operations per second. This element cannot be used with
<read _i o ps_sec> or <wri te_i o ps_sec>.
<read _i o ps_sec> - the read I/O operations per second.
<wri te_i o ps_sec> - the write I/O operations per second.

2 0 .1 6 .1 .6 . drive r
The optional <d ri ver> element allows specifying further details related to the hypervisor driver that
is used to provide the disk. The following options may be used:
If the hypervisor supports multiple backend drivers, then the name attribute selects the primary
backend driver name, while the optional type attribute provides the sub-type. For a list of possible
types refer to D river Elements
The optional cache attribute controls the cache mechanism, possible values are: d efaul t,
no ne, wri tethro ug h, wri teback, d i rectsync (similar to wri tethro ug h, but it bypasses the
host physical machine page cache) and unsafe (host physical machine may cache all disk io,
and sync requests from guest virtual machine virtual machines are ignored).
The optional erro r_po l i cy attribute controls how the hypervisor behaves on a disk read or
write error, possible values are sto p, repo rt, i g no re, and eno space. The default setting of
erro r_po l i cy is repo rt. There is also an optional rerro r_po l i cy that controls behavior for
read errors only. If no rerro r_po l i cy is given, erro r_po l i cy is used for both read and write
errors. If rerro r_po l i cy is given, it overrides the erro r_po l i cy for read errors. Also note that
eno space is not a valid policy for read errors, so if erro r_po l i cy is set to eno space and no
rerro r_po l i cy is given, the read error the default setting, repo rt will be used.
The optional i o attribute controls specific policies on I/O; q emu guest virtual machine virtual
machines support thread s and nati ve. The optional i o eventfd attribute allows users to set
domain I/O asynchronous handling for disk device. The default is left to the discretion of the
hypervisor. Accepted values are o n and o ff. Enabling this allows the guest virtual machine
virtual machine to be executed while a separate thread handles I/O. Typically guest virtual
machine virtual machines experiencing high system CPU utilization during I/O will benefit from
this. On the other hand, an overloaded host physical machine can increase guest virtual machine
virtual machine I/O latency. Unless you are absolutely certian that the i o needs to be
manipulated, it is highly recommended that you not change the default setting and allow the
hypervisor to dictate the setting.
The optional event_i d x attribute controls some aspects of device event processing and can be
set to either o n or o ff - if it is on, it will reduce the number of interrupts and exits for the guest
virtual machine virtual machine. The default is determined by the hypervisor and the default
setting is o n. In cases that there is a situation where this behavior is suboptimal, this attribute

366

Chapt er 2 0 . Manipulat ing t he Domain XML

provides a way to force the feature o ff. Unless you are absolutely certian that the event_i d x
needs to be manipulated, it is highly recommended that you not change the default setting and
allow the hypervisor to dictate the setting.
The optional co py_o n_read attribute controls whether to copy the read backing file into the
image file. The accepted values can be either o n or <o ff>. co py-o n-read avoids accessing the
same backing file sectors repeatedly and is useful when the backing file is over a slow network.
By default co py-o n-read is o ff.

2 0 .1 6 .1 .7 . Addit io nal de vice e le m e nt s

The following attributes may be used within the d evi ce element:
<bo o t> - Specifies that the disk is bootable.
Ad d it io n al b o o t valu es
<o rd er> - D etermines the order in which devices will be tried during boot sequence.
<per-d evi ce> boot elements cannot be used together with general boot elements in BIOS
bootloader section
<encrypti o n> - Specifies how the volume is encrypted. See the Storage Encryption page for
more information.
<read o nl y> - Indicates the device cannot be modified by the guest virtual machine virtual
machine. This setting is the default for disks with attri bute d evi ce= ' cd ro m' .
shareabl e Indicates the device is expected to be shared between domains (as long as
hypervisor and OS support this). If shareabl e is used, cache= ' no ' should be used for that
device.
<transi ent>- Indicates that changes to the device contents should be reverted automatically
when the guest virtual machine virtual machine exits. With some hypervisors, marking a disk
transi ent prevents the domain from participating in migration or snapshots.
<seri al >- Specifies the serial number of guest virtual machine virtual machine virtual machine's
hard drive. For example, <seri al >WD -WMAP9A966149</seri al >.
<wwn> - Specifies the WWN (World Wide Name) of a virtual hard disk or CD -ROM drive. It must be
composed of 16 hexadecimal digits.
<vend o r> - Specifies the vendor of a virtual hard disk or CD -ROM device. It must not be longer
than 8 printable characters.
<pro d uct> - Specifies the product of a virtual hard disk or CD -ROM device. It must not be longer
than 16 printable characters
<ho st> - Supports 4 attributes: vi z, name, po rt, transpo rt and so cket, which specify the
hostname, the port number, transport type and path to socket, respectively. The meaning of this
element and the number of the elements depend on the pro to co l attribute as shown here:
Ad d it io n al h o st at t rib u t es
nbd - Specifies a server running nbd-server and may only be used for only one host physical
machine
rbd - Monitors servers of RBD type and may be used for one or more host physical machines

367

Virt ualiz at ion Administ rat ion G uide

sheepd o g - Specifies one of the sheepdog servers (default is localhost:7000) and can be
used one or none of the host physical machines
g l uster - Specifies a server running a glusterd daemon and may be used for only only one
host physical machine. The valid values for transport attribute are tcp, rd ma or uni x. If
nothing is specified, tcp is assumed. If transport is uni x, the so cket attribute specifies path
to unix socket.
<ad d ress> - Ties the disk to a given slot of a controller. The actual <co ntro l l er> device can
often be inferred by but it can also be explicitly specified. The type attribute is mandatory, and is
typically pci or d ri ve. For a pci controller, additional attributes for bus, sl o t, and functi o n
must be present, as well as optional d o mai n and mul ti functi o n. mul ti fun cti o n defaults
to o ff. For a d ri ve controller, additional attributes co ntro l l er, bus, targ et, and uni t are
available, each with a default setting of 0 .
auth - Provides the authentication credentials needed to access the source. It includes a
mandatory attribute username, which identifies the username to use during authentication, as well
as a sub-element secret with mandatory attribute type. More information can be found here at
D evice Elements
g eo metry - Provides the ability to override geometry settings. This mostly useful for S390 D ASD disks or older D OS-disks.
cyl s - Specifies the number of cylinders.
head s - Specifies the number of heads.
secs - Specifies the number of sectors per track.
trans - Specifies the BIOS-Translation-Modus and can have the following values:no ne, l ba or
auto
bl o cki o - Allows the block device to be overridden with any of the block device properties listed
below:
b lo ckio o p t io n s
l o g i cal _bl o ck_si ze- reports to the guest virtual machine virtual machine OS and
describes the smallest units for disk I/O.
physi cal _bl o ck_si ze - reports to the guest virtual machine virtual machine OS and
describes the disk's hardware sector size which can be relevant for the alignment of disk data.

20.16.2. Filesyst ems

A filesystems directory on the host physical machine that can be accessed directly from the guest
virtual machine virtual machine

...
<devices>
<filesystem type='template'>
<source name='my-vm-template'/>
<target dir='/'/>
</filesystem>
<filesystem type='mount' accessmode='passthrough'>

368

Chapt er 2 0 . Manipulat ing t he Domain XML

<driver type='path' wrpolicy='immediate'/>

<source dir='/export/to/guest'/>
<target dir='/import/from/host'/>
<readonly/>
</filesystem>
...
</devices>
...

Fig u re 20.24 . D evices - f ilesyst ems

The fi l esystem attribute has the following possible values:
type= ' mo unt' - Specifies the host physical machine directory to mount in the guest virtual
machine. This is the default type if one is not specified. This mode also has an optional subelement d ri ver, with an attribute type= ' path' or type= ' hand l e' . The driver block has an
optional attribute wrpo l i cy that further controls interaction with the host physical machine page
cache; omitting the attribute reverts to the default setting, while specifying a value immediate
means that a host physical machine writeback is immediately triggered for all pages touched
during a guest virtual machine file write operation
type= ' templ ate' - Specifies the OpenVZ filesystem template and is only used by OpenVZ
driver.
type= ' fi l e' - Specifies that a host physical machine file will be treated as an image and
mounted in the guest virtual machine. This filesystem format will be autodetected and is only used
by LXC driver.
type= ' bl o ck' - Specifies the host physical machine block device to mount in the guest virtual
machine. The filesystem format will be autodetected and is only used by LXC driver.
type= ' ram' - Specifies that an in-memory filesystem, using memory from the host physical
machine OS will be used. The source element has a single attribute usag e which gives the
memory usage limit in kibibytes and is only used by LXC driver.
type= ' bi nd ' - Specifies a directory inside the guest virtual machine which will be bound to
another directory inside the guest virtual machine. This element is only used by LXC driver.
accessmo d e which specifies the security mode for accessing the source. Currently this only
works with type='mount' for the QEMU/KVM driver. The possible values are:
passthro ug h - Specifies that the source is accessed with the User's permission settings that
are set from inside the guest virtual machine. This is the default accessmode if one is not
specified.
mapped - Specifies that the source is accessed with the permission settings of the hypervisor.
sq uash - Similar to ' passthro ug h' , the exception is that failure of privileged operations like
cho wn are ignored. This makes a passthrough-like mode usable for people who run the
hypervisor as non-root.
<so urce> - Specifies that the resource on the host physical machine that is being accessed in
the guest virtual machine. The name attribute must be used with <type= ' templ ate' >, and the
d i r attribute must be used with <type= ' mo unt' >. The usag e attribute is used with
<type= ' ram' > to set the memory limit in KB.

369

Virt ualiz at ion Administ rat ion G uide

targ et - D ictates where the source drivers can be accessed in the guest virtual machine. For
most drivers this is an automatic mount point, but for QEMU-KVM this is merely an arbitrary string
tag that is exported to the guest virtual machine as a hint for where to mount.
read o nl y - Enables exporting the filesytem as a readonly mount for guest virtual machine, by
default read -wri te access is given.
space_hard _l i mi t - Specifies the maximum space available to this guest virtual machine's
filesystem
space_so ft_l i mi t - Specifies the maximum space available to this guest virtual machine's
filesystem. The container is permitted to exceed its soft limits for a grace period of time. Afterwards
the hard limit is enforced.

20.16.3. Device Addresses

Many devices have an optional <ad d ress> sub-element to describe where the device placed on the
virtual bus is presented to the guest virtual machine. If an address (or any optional attribute within an
address) is omitted on input, libvirt will generate an appropriate address; but an explicit address is
required if more control over layout is required. See below for device examples including an address
element.
Every address has a mandatory attribute type that describes which bus the device is on. The choice
of which address to use for a given device is constrained in part by the device and the architecture of
the guest virtual machine. For example, a disk device uses type= ' d i sk' , while a console device
would use type= ' pci ' on i686 or x86_64 guest virtual machines, or type= ' spapr-vi o ' on
PowerPC64 pseries guest virtual machines. Each address <type> has additional optional attributes
that control where on the bus the device will be placed. The additional attributes are as follows:
type= ' pci ' - PCI addresses have the following additional attributes:
d o mai n (a 2-byte hex integer, not currently used by qemu)
bus (a hex value between 0 and 0xff, inclusive)
sl o t (a hex value between 0x0 and 0x1f, inclusive)
functi o n (a value between 0 and 7, inclusive)
Also available is the mul ti functi o n attribute, which controls turning on the multifunction bit
for a particular slot/function in the PCI control register. This multifunction attribute defaults to
' o ff' , but should be set to ' o n' for function 0 of a slot that will have multiple functions
used.
type= ' d ri ve - drive addresses have the following additional attributes:
co ntro l l er- (a 2-digit controller number)
bus - (a 2-digit bus number)
targ et - (a 2-digit bus number)
uni t - (a 2-digit unit number on the bus)
type= ' vi rti o -seri al ' - Each virtio-serial address has the following additional attributes:
co ntro l l er - (a 2-digit controller number)
bus - (a 2-digit bus number)

370

Chapt er 2 0 . Manipulat ing t he Domain XML

sl o t - (a 2-digit slot within the bus)

type= ' cci d ' - A CCID address, used for smart-cards, has the following additional attributes:
bus - (a 2-digit bus number)
sl o t attribute - (a 2-digit slot within the bus)
type= ' usb' - USB addresses have the following additional attributes:
bus - (a hex value between 0 and 0xfff, inclusive)
po rt - (a dotted notation of up to four octets, such as 1.2 or 2.1.3.1)
type= ' spapr-vi o - On PowerPC pseries guest virtual machines, devices can be assigned to
the SPAPR-VIO bus. It has a flat 64-bit address space; by convention, devices are generally
assigned at a non-zero multiple of 0x1000, but other addresses are valid and permitted by libvirt.
The additional attribute: reg (the hex value address of the starting register) can be assigned to
this attribute.

20.16.4 . Cont rollers

D epending on the guest virtual machine architecture, it is possible to assign many virtual devices to
a single bus. Under normal circumstances libvirt can automatically infer which controller to use for the
bus. However, it may be necessary to provide an explicit <co ntro l l er> element in the guest virtual
machine XML:

...
<devices>
<controller type='ide' index='0'/>
<controller type='virtio-serial' index='0' ports='16' vectors='4'/>
<controller type='virtio-serial' index='1'>
<address type='pci' domain='0x0000' bus='0x00' slot='0x0a'
function='0x0'/>
<controller type='scsi' index='0' model='virtio-scsi'
num_queues='8'/>
</controller>
...
</devices>
...

Fig u re 20.25. C o n t ro ller Elemen t s

Each controller has a mandatory attribute type, which must be one of "i d e", "fd c", "scsi ",
"sata", "usb", "cci d ", o r "vi rti o -seri al ", and a mandatory attribute i nd ex which is the
decimal integer describing in which order the bus controller is encountered (for use in controller
attributes of ad d ress elements). The "vi rti o -seri al " controller has two additional optional
attributes, po rts and vecto rs, which control how many devices can be connected through the
controller.
A <co ntro l l er type= ' scsi ' > has an optional attribute mo d el , which is one of "auto ",
"busl o g i c", "i bmvscsi ", "l si l o g i c", "l si as10 6 8", "vi rti o -scsi or "vmpvscsi ". It

371

Virt ualiz at ion Administ rat ion G uide

should be noted that virtio-scsi controllers and drivers will work on both KVM and Windows guest
virtual machines. The <co ntro l l er type= ' scsi ' > also has an attribute num_q ueues which
enables multi-queue support for the number of queues specified.
A "usb" controller has an optional attribute mo d el , which is one of "pi i x3-uhci ", "pi i x4 uhci ", "ehci ", "i ch9 -ehci 1", "i ch9 -uhci 1", "i ch9 -uhci 2", "i ch9 -uhci 3", "vt82c6 86 buhci ", "pci -o hci " or "nec-xhci ". Additionally, if the USB bus needs to be explicitly disabled for
the guest virtual machine, mo d el = ' no ne' may be used. The PowerPC64 " spapr-vio" addresses do
not have an associated controller.
For controllers that are themselves devices on a PCI or USB bus, an optional sub-element ad d ress
can specify the exact relationship of the controller to its master bus, with semantics given above.
USB companion controllers have an optional sub-element master to specify the exact relationship of
the companion to its master controller. A companion controller is on the same bus as its master, so
the companion index value should be equal.

...
<devices>
<controller type='usb' index='0'
<address type='pci' domain='0'
</controller>
<controller type='usb' index='0'
<master startport='0'/>
<address type='pci' domain='0'
multifunction='on'/>
</controller>
...
</devices>
...

model='ich9-ehci1'>
bus='0' slot='4' function='7'/>
model='ich9-uhci1'>
bus='0' slot='4' function='0'

Fig u re 20.26 . D evices - co n t ro llers - U SB

20.16.5. Device Leases

When using a lock manager, you have the option to record device leases against a guest virtual
machine. The lock manager will ensure that the guest virtual machine doesn't start unless the leases
can be acquired. When configured using conventional management tools, the following section of
the domain xml is effected:

...
<devices>
...
<lease>
<lockspace>somearea</lockspace>
<key>somekey</key>
<target path='/some/lease/path' offset='1024'/>

372

Chapt er 2 0 . Manipulat ing t he Domain XML

</lease>
...
</devices>
...
Fig u re 20.27. D evices - d evice leases
The l ease section can have the following arguements:
l o ckspace - an arbitrary string that identifies lockspace within which the key is held. Lock
managers may impose extra restrictions on the format, or length of the lockspace name.
key - an arbitrary string, that uniquely identies the lease to be acquired. Lock managers may
impose extra restrictions on the format, or length of the key.
targ et - the fully qualified path of the file associated with the lockspace. The offset specifies
where the lease is stored within the file. If the lock manager does not require a offset, set this value
to 0 .

20.16.6. Host Physical Machine Device Assignment

This section provides information about host physical machine device assignment.

2 0 .1 6 .6 .1 . USB / PCI De vice s

The host physical machine's USB and PCI devices can be passed through to the guest virtual
machine using the ho std ev element, by modifying the host physical machine using a management
tool the following section of the domain xml file is configured:

...
<devices>
<hostdev mode='subsystem' type='usb'>
<source startupPolicy='optional'>
<vendor id='0x1234'/>
<product id='0xbeef'/>
</source>
<boot order='2'/>
</hostdev>
</devices>
...
Fig u re 20.28. D evices - h o st p h ysical mach in e d evice assig n men t
Alternatively the following can also be done:

...
<devices>
<hostdev mode='subsystem' type='pci' managed='yes'>

373

Virt ualiz at ion Administ rat ion G uide

<source>
<address bus='0x06' slot='0x02' function='0x0'/>
</source>
<boot order='1'/>
<rom bar='on' file='/etc/fake/boot.bin'/>
</hostdev>
</devices>
...
Fig u re 20.29 . D evices - h o st p h ysical mach in e d evice assig n men t alt ern at ive
The components of this section of the domain XML are as follows:
T ab le 20.13. H o st p h ysical mach in e d evice assig n men t elemen t s
Paramet er

D escrip t io n

ho std ev

This is the main container for describing host

physical machine devices. For USB device
passthrough mo d e is always subsystem and
type is usb for a USB device and pci for a PCI
device. When manag ed is yes for a PCI device,
it is detached from the host physical machine
before being passed on to the guest virtual
machine, and reattached to the host physical
machine after the guest virtual machine exits. If
manag ed is omitted or no for PCI and for USB
devices, the user is responsible to use the
argument vi rNo d eD evi ceD ettach (or vi rsh
no d ed ev-d ettach) before starting the guest
virtual machine or hot-plugging the device, and
vi rNo d eD evi ceR eAttach (or vi rsh
no d ed ev-reattach) after hot-unplug or
stopping the guest virtual machine.
D escribes the device as seen from the host
physical machine. The USB device can either be
addressed by vendor / product id using the
vend o r and pro d uct elements or by the
device's address on the host physical machines
using the ad d ress element. PCI devices on the
other hand can only be described by their
address. Note that the source element of USB
devices may contain a startupP o l i cy
attribute which can be used to define a rule for
what to do if the specified host physical machine
USB device is not found. The attribute accepts
the following values:

so urce

mand ato ry - fails if missing for any reason

(the default)
req ui si te - fails if missing on boot up,
drops if missing on migrate/restore/revert
o pti o nal - drops if missing at any start
attempt

374

Chapt er 2 0 . Manipulat ing t he Domain XML

Paramet er

D escrip t io n

vend o r, pro d uct

These elements each have an i d attribute that

specifies the USB vendor and product id. The
ID s can be given in decimal, hexadecimal
(starting with 0x) or octal (starting with 0) form.
Specifies that the device is bootable. The
attribute's order determines the order in which
devices will be tried during boot sequence. The
per-device boot elements cannot be used
together with general boot elements in BIOS
bootloader section.
Used to change how a PCI device's ROM is
presented to the guest virtual machine. The
optional bar attribute can be set to o n or o ff,
and determines whether or not the device's ROM
will be visible in the guest virtual machine's
memory map. (In PCI documentation, the
ro mbar setting controls the presence of the
Base Address Register for the ROM). If no rom
bar is specified, the default setting will be used.
The optional fi l e attribute is used to point to a
binary file to be presented to the guest virtual
machine as the device's ROM BIOS. This can be
useful, for example, to provide a PXE boot ROM
for a virtual function of an sr-iov capable
ethernet device (which has no boot ROMs for the
VFs).
Also has a bus and d evi ce attribute to specify
the USB bus and device number the device
appears at on the host physical machine. The
values of these attributes can be given in
decimal, hexadecimal (starting with 0x) or octal
(starting with 0) form. For PCI devices the
element carries 3 attributes allowing to
designate the device as can be found with
l spci or with vi rsh no d ed ev-l i st

bo o t

ro m

ad d ress

2 0 .1 6 .6 .2 . Blo ck / charact e r de vice s

The host physical machine's block / character devices can be passed through to the guest virtual
machine by using management tools to modify the domain xml ho std ev element. Note that this is
only possible with container based virtualization.

...
<hostdev mode='capabilities' type='storage'>
<source>
<block>/dev/sdf1</block>
</source>
</hostdev>
...

375

Virt ualiz at ion Administ rat ion G uide

Fig u re 20.30. D evices - h o st p h ysical mach in e d evice assig n men t b lo ck ch aract er

d evices
An alternative approach is this:

...
<hostdev mode='capabilities' type='misc'>
<source>
<char>/dev/input/event3</char>
</source>
</hostdev>
...

Fig u re 20.31. D evices - h o st p h ysical mach in e d evice assig n men t b lo ck ch aract er

d evices alt ern at ive 1
Another alternative approach is this:

...
<hostdev mode='capabilities' type='net'>
<source>
<interface>eth0</interface>
</source>
</hostdev>
...

Fig u re 20.32. D evices - h o st p h ysical mach in e d evice assig n men t b lo ck ch aract er

d evices alt ern at ive 2
The components of this section of the domain XML are as follows:
T ab le 20.14 . B lo ck / ch aract er d evice elemen t s
Paramet er

D escrip t io n

ho std ev

This is the main container for describing host

physical machine devices. For block/character
devices passthrough mo d e is always
capabi l i ti es and type is bl o ck for a block
device and char for a character device.
This describes the device as seen from the host
physical machine. For block devices, the path to
the block device in the host physical machine
OS is provided in the nested bl o ck element,
while for character devices the char element is
used

so urce

376

Chapt er 2 0 . Manipulat ing t he Domain XML

20.16.7. Redirect ed Devices

USB device redirection through a character device is supported by configuring it with management
tools that modify the following section of the domain xml:

...
<devices>
<redirdev bus='usb' type='tcp'>
<source mode='connect' host='localhost' service='4000'/>
<boot order='1'/>
</redirdev>
<redirfilter>
<usbdev class='0x08' vendor='0x1234' product='0xbeef'
version='2.00' allow='yes'/>
<usbdev allow='no'/>
</redirfilter>
</devices>
...
Fig u re 20.33. D evices - red irect ed d evices
The components of this section of the domain XML are as follows:
T ab le 20.15. R ed irect ed d evice elemen t s
Paramet er

D escrip t io n

red i rd ev

This is the main container for describing

redirected devices. bus must be usb for a USB
device. An additional attribute type is required,
matching one of the supported serial device
types, to describe the host physical machine
side of the tunnel; type= ' tcp' or
type= ' spi cevmc' (which uses the usbredir
channel of a SPICE graphics device) are typical.
The redirdev element has an optional subelement ad d ress which can tie the device to a
particular controller. Further sub-elements, such
as so urce, may be required according to the
given type, although atarg et sub-element is
not required (since the consumer of the
character device is the hypervisor itself, rather
than a device visible in the guest virtual
machine).
Specifies that the device is bootable. The order
attribute determines the order in which devices
will be tried during boot sequence. The perdevice boot elements cannot be used together
with general boot elements in BIOS bootloader
section.

bo o t

377

Virt ualiz at ion Administ rat ion G uide

Paramet er

D escrip t io n

red i rfi l ter

This is used for creating the filter rule to filter out

certain devices from redirection. It uses subelement usbd ev to define each filter rule. The
cl ass attribute is the USB Class code.

20.16.8. Smart card Devices

A virtual smartcard device can be supplied to the guest virtual machine via the smartcard element. A
USB smartcard reader device on the host machine cannot be used on a guest with simple device
passthrough, as it cannot be made available to both the host and guest, and can lock the host
computer when it is removed from the guest. Therefore, some hypervisors provide a specialized
virtual device that can present a smartcard interface to the guest virtual machine, with several modes
for describing how the credentials are obtained from the host machine, or from a channel created by
a third-party smartcard provider. To set parameters for USB device redirection through a character
device, edit the following section of the domain XML:

...
<devices>
<smartcard mode='host'/>
<smartcard mode='host-certificates'>
<certificate>cert1</certificate>
<certificate>cert2</certificate>
<certificate>cert3</certificate>
<database>/etc/pki/nssdb/</database>
</smartcard>
<smartcard mode='passthrough' type='tcp'>
<source mode='bind' host='127.0.0.1' service='2001'/>
<protocol type='raw'/>
<address type='ccid' controller='0' slot='0'/>
</smartcard>
<smartcard mode='passthrough' type='spicevmc'/>
</devices>
...
Fig u re 20.34 . D evices - smart card d evices
The smartcard element has a mandatory attribute mo d e. The following modes are supported; in
each mode, the guest virtual machine sees a device on its USB bus that behaves like a physical USB
CCID (Chip/Smart Card Interface D evice) card.
The mode attributes are as follows:
T ab le 20.16 . Smart card mo d e elemen t s
Paramet er

378

D escrip t io n

Chapt er 2 0 . Manipulat ing t he Domain XML

Paramet er

D escrip t io n

mo d e= ' ho st'

In this mode, the hypervisor relays all direct

access requests from the guest virtual machine
to the host physical machine's smartcard via
NSS. No other attributes or sub-elements are
required. See below about the use of an
optional ad d ress sub-element.
This mode allows you to provide three NSS
certificate names residing in a database on the
host physical machine, rather than requiring a
smartcard to be plugged into the host physical
machine. These certificates can be generated
via the command certuti l -d
/etc/pki /nssd b -x -t C T ,C T ,C T -S -s
C N= cert1 -n cert1, and the resulting three
certificate names must be supplied as the
content of each of three certi fi cate subelements. An additional sub-element d atabase
can specify the absolute path to an alternate
directory (matching the -d option of the
certuti l command when creating the
certificates); if not present, it defaults to
/etc/pki /nssd b.
This mode allows you to tunnel all requests
through a secondary character device to a thirdparty provider (which may in turn be talking to a
smartcard or using three certificate files), rather
than having the hypervisor directly
communicate with the host physical machine. In
this mode, an additional attribute type is
required, matching one of the supported serial
device types, to describe the host physical
machine side of the tunnel; type= ' tcp' or
type= ' spi cevmc' (which uses the smartcard
channel of a SPICE graphics device) are typical.
Further sub-elements, such as so urce, may be
required according to the given type, although a
targ et sub-element is not required (since the
consumer of the character device is the
hypervisor itself, rather than a device visible in
the guest virtual machine).

mo d e= ' ho st-certi fi cates'

mo d e= ' passthro ug h'

Each mode supports an optional sub-element ad d ress, which fine-tunes the correlation between the
smartcard and a ccid bus controller (Refer to Section 20.16.3, D evice Addresses ).

20.16.9. Net work Int erfaces

The network interface devices are modified using management tools that will configure the following
part of the D omain XML:

...
<devices>

379

Virt ualiz at ion Administ rat ion G uide

<interface type='bridge'>
<source bridge='xenbr0'/>
<mac address='00:16:3e:5d:c7:9e'/>
<script path='vif-bridge'/>
<boot order='1'/>
<rom bar='off'/>
</interface>
</devices>
...
Fig u re 20.35. D evices - n et wo rk in t erf aces
There are several possibilities for specifying a network interface visible to the guest virtual machine.
Each subsection below provides more details about common setup options. Additionally, each
<i nterface> element has an optional <ad d ress> sub-element that can tie the interface to a
particular pci slot, with attribute type= ' pci ' (Refer to Section 20.16.3, D evice Addresses ).

2 0 .1 6 .9 .1 . Virt ual ne t wo rks

This is the recommended configuration for general guest virtual machine connectivity on host
physical machines with dynamic / wireless networking configurations (or multi-host physical
machine environments where the host physical machine hardware details are described separately in
a <netwo rk> definition). In addition, it provides a connection whose details are described by the
named network definition. D epending on the virtual network's fo rward mo d e configuration, the
network may be totally isolated (no <fo rward > element given), NAT'ing to an explicit network device
or to the default route (fo rward mo d e= ' nat' ), routed with no NAT (fo rward mo d e= ' ro ute' /),
or connected directly to one of the host physical machine's network interfaces (via macvtap) or
bridge devices (fo rward mo d e= ' bri d g e| pri vate| vepa| passthro ug h' /)
For networks with a forward mode of bridge, private, vepa, and passthrough, it is assumed that the
host physical machine has any necessary D NS and D HCP services already setup outside the scope
of libvirt. In the case of isolated, nat, and routed networks, D HCP and D NS are provided on the
virtual network by libvirt, and the IP range can be determined by examining the virtual network config
with vi rsh net-d umpxml [netwo rkname]. There is one virtual network called 'default' setup out
of the box which does NAT'ing to the default route and has an IP range of
192.168.122.0/255.255.255.0. Each guest virtual machine will have an associated tun device
created with a name of vnetN, which can also be overridden with the <targ et> element (refer to
Section 20.16.9.11, Overriding the target element ).
When the source of an interface is a network, a portgroup can be specified along with the name of the
network; one network may have multiple portgroups defined, with each portgroup containing slightly
different configuration information for different classes of network connections. Also, similar to
<d i rect> network connections (described below), a connection of type netwo rk may specify a
<vi rtual po rt> element, with configuration data to be forwarded to a vepa (802.1Qbg) or
802.1Qbh compliant switch, or to an Open vSwitch virtual switch.
Since the actual type of switch may vary depending on the configuration in the <netwo rk> on the
host physical machine, it is acceptable to omit the virtualport type attribute, and specify attributes
from multiple different virtualport types (and also to leave out certain attributes); at domain startup
time, a complete <vi rtual po rt> element will be constructed by merging together the type and
attributes defined in the network and the portgroup referenced by the interface. The newlyconstructed virtualport is a combination of both. The attributes from lower virtualport can't make
changes on the ones defined in higher virtualport. Interfaces take the highest priority, portgroup is
lowest priority.
For example, in order to work properly with both an 802.1Qbh switch and an Open vSwitch switch,

380

Chapt er 2 0 . Manipulat ing t he Domain XML

you may choose to specify no type, but both an pro fi l ei d (in case the switch is 802.1Qbh) and
an i nterfacei d (in case the switch is Open vSwitch) (you may also omit the other attributes, such
as manag eri d , typei d , or pro fi l ei d , to be filled in from the network's vi rtual po rt). If you
want to limit a guest virtual machine to connecting only to certain types of switches, you can specify
the virtualport type, but still omit some/all of the parameters - in this case if the host physical
machine's network has a different type of virtualport, connection of the interface will fail. The virtual
network parameters are defined using management tools that modify the following part of the domain
XML:

...
<devices>
<interface type='network'>
<source network='default'/>
</interface>
...
<interface type='network'>
<source network='default' portgroup='engineering'/>
<target dev='vnet7'/>
<mac address="00:11:22:33:44:55"/>
<virtualport>
<parameters instanceid='09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f'/>
</virtualport>
</interface>
</devices>
...
Fig u re 20.36 . D evices - n et wo rk in t erf aces- virt u al n et wo rks

2 0 .1 6 .9 .2 . Bridge t o LAN
Note that this is the recommended configuration setting for general guest virtual machine connectivity
on host physical machines with static wired networking configurations.
Bridge to LAN provides a bridge from the guest virtual machine directly onto the LAN. This assumes
there is a bridge device on the host physical machine which has one or more of the host physical
machines physical NICs enslaved. The guest virtual machine will have an associated tun device
created with a name of <vnetN>, which can also be overridden with the <targ et> element (refer to
Section 20.16.9.11, Overriding the target element ). The <tun> device will be enslaved to the bridge.
The IP range / network configuration is whatever is used on the LAN. This provides the guest virtual
machine full incoming and outgoing net access just like a physical machine.
On Linux systems, the bridge device is normally a standard Linux host physical machine bridge. On
host physical machines that support Open vSwitch, it is also possible to connect to an open vSwitch
bridge device by adding a vi rtual po rt type= ' o penvswi tch' / to the interface definition. The
Open vSwitch type virtualport accepts two parameters in its parameters element - an i nterfacei d
which is a standard uuid used to uniquely identify this particular interface to Open vSwitch (if you do
no specify one, a random i nterfacei d will be generated for you when you first define the
interface), and an optional pro fi l ei d which is sent to Open vSwitch as the interfaces <po rtpro fi l e>. To set the bridge to LAN settings, use a management tool that will configure the following
part of the domain XML:

381

Virt ualiz at ion Administ rat ion G uide

...
<devices>
...
<interface type='bridge'>
<source bridge='br0'/>
</interface>
<interface type='bridge'>
<source bridge='br1'/>
<target dev='vnet7'/>
<mac address="00:11:22:33:44:55"/>
</interface>
<interface type='bridge'>
<source bridge='ovsbr'/>
<virtualport type='openvswitch'>
<parameters profileid='menial' interfaceid='09b11c53-8b5c-4eeb8f00-d84eaa0aaa4f'/>
</virtualport>
</interface>
...
</devices>
Fig u re 20.37. D evices - n et wo rk in t erf aces- b rid g e t o LAN

2 0 .1 6 .9 .3. Se t t ing a po rt m asque rading range

In cases where you want to set the port masquerading range, the port can be set as follows:

<forward mode='nat'>
<address start='192.0.2.1' end='192.0.2.10'/>
</forward> ...
Fig u re 20.38. Po rt Masq u erad in g R an g e
These values should be set using the i ptabl es commands as shown in Section 18.3, Network
Address Translation Mode

2 0 .1 6 .9 .4 . Use rspace SLIRP st ack

Setting the userspace SLIRP stack parameters provides a virtual LAN with NAT to the outside world.
The virtual network has D HCP and D NS services and will give the guest virtual machine an IP
addresses starting from 10.0.2.15. The default router will be 10.0.2.2 and the D NS server will be
10.0.2.3. This networking is the only option for unprivileged users who need their guest virtual
machines to have outgoing access.
The userspace SLIP stack parameters are defined in the following part of the domain XML::

...

382

Chapt er 2 0 . Manipulat ing t he Domain XML

<devices>
<interface type='user'/>
...
<interface type='user'>
<mac address="00:11:22:33:44:55"/>
</interface>
</devices>
...
Fig u re 20.39 . D evices - n et wo rk in t erf aces- U sersp ace SLIR P st ack

2 0 .1 6 .9 .5 . Ge ne ric Et he rne t co nne ct io n

Provides a means for the administrator to execute an arbitrary script to connect the guest virtual
machine's network to the LAN. The guest virtual machine will have a tun device created with a name
of vnetN, which can also be overridden with the targ et element. After creating the tun device a shell
script will be run which is expected to do whatever host physical machine network integration is
required. By default this script is called /etc/q emu-i fup but can be overridden (refer to
Section 20.16.9.11, Overriding the target element ).
The generic Ethernet connection parameters are defined in the following part of the domain XML:

...
<devices>
<interface type='ethernet'/>
...
<interface type='ethernet'>
<target dev='vnet7'/>
<script path='/etc/qemu-ifup-mynet'/>
</interface>
</devices>
...
Fig u re 20.4 0. D evices - n et wo rk in t erf aces- g en eric Et h ern et co n n ect io n

2 0 .1 6 .9 .6 . Dire ct at t achm e nt t o physical int e rface s

Using <i nterface type= ' d i rect' > attaches a virtual machine's NIC to a specified physical
interface on the host.
This set up requires the Linux macvtap driver to be available. One of the following modes can be
chosen for the operation mode of the macvtap device: vepa ( 'Virtual Ethernet Port Aggregator'),
which is the default mode, bri d g e or pri vate.
To set up direct attachment to physical interface, use the following parameters in the domain XML:

...
<devices>
...

383

Virt ualiz at ion Administ rat ion G uide

<interface type='direct'>
<source dev='eth0' mode='vepa'/>
</interface>
</devices>
...
Fig u re 20.4 1. D evices - n et wo rk in t erf aces- d irect at t ach men t t o p h ysical in t erf aces
The individual modes cause the delivery of packets to behave as shown in Table 20.17, D irect
attachment to physical interface elements :
T ab le 20.17. D irect at t ach men t t o p h ysical in t erf ace elemen t s
Elemen t

D escrip t io n

vepa

All of the guest virtual machines' packets are

sent to the external bridge. Packets whose
destination is a guest virtual machine on the
same host physical machine as where the
packet originates from are sent back to the host
physical machine by the VEPA capable bridge
(today's bridges are typically not VEPA
capable).
Packets whose destination is on the same host
physical machine as where they originate from
are directly delivered to the target macvtap
device. Both origin and destination devices
need to be in bridge mode for direct delivery. If
either one of them is in vepa mode, a VEPA
capable bridge is required.
All packets are sent to the external bridge and
will only be delivered to a target VM on the same
host physical machine if they are sent through
an external router or gateway and that device
sends them back to the host physical machine.
This procedure is followed if either the source or
destination device is in private mode.
This feature attaches a virtual function of a
SRIOV capable NIC directly to a guest virtual
machine without losing the migration capability.
All packets are sent to the VF/IF of the
configured network device. D epending on the
capabilities of the device additional
prerequisites or limitations may apply; for
example, this requires kernel 2.6.38 or newer.

bri d g e

pri vate

passthro ug h

The network access of direct attached virtual machines can be managed by the hardware switch to
which the physical interface of the host physical machine machine is connected to.
The interface can have additional parameters as shown below, if the switch is conforming to the IEEE
802.1Qbg standard. The parameters of the virtualport element are documented in more detail in the
IEEE 802.1Qbg standard. The values are network specific and should be provided by the network
administrator. In 802.1Qbg terms, the Virtual Station Interface (VSI) represents the virtual interface of
a virtual machine.
Note that IEEE 802.1Qbg requires a non-zero value for the VLAN ID .

384

Chapt er 2 0 . Manipulat ing t he Domain XML

Additional elements that can be manipulated are described in Table 20.18, D irect attachment to
physical interface additional elements :
T ab le 20.18. D irect at t ach men t t o p h ysical in t erf ace ad d it io n al elemen t s
Elemen t

D escrip t io n

manag eri d

The VSI Manager ID identifies the database

containing the VSI type and instance definitions.
This is an integer value and the value 0 is
reserved.
The VSI Type ID identifies a VSI type
characterizing the network access. VSI types are
typically managed by network administrator.
This is an integer value.
The VSI Type Version allows multiple versions of
a VSI Type. This is an integer value.
The VSI Instance ID Identifier is generated when
a VSI instance (i.e. a virtual interface of a virtual
machine) is created. This is a globally unique
identifier.
The profile ID contains the name of the port
profile that is to be applied onto this interface.
This name is resolved by the port profile
database into the network parameters from the
port profile, and those network parameters will
be applied to this interface.

typei d

typei d versi o n
i nstancei d

pro fi l ei d

Additional parameters in the domain XML include:

...
<devices>
...
<interface type='direct'>
<source dev='eth0.2' mode='vepa'/>
<virtualport type="802.1Qbg">
<parameters managerid="11" typeid="1193047" typeidversion="2"
instanceid="09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f"/>
</virtualport>
</interface>
</devices>
...
Fig u re 20.4 2. D evices - n et wo rk in t erf aces- d irect at t ach men t t o p h ysical in t erf aces
ad d it io n al p aramet ers
The interface can have additional parameters as shown below if the switch is conforming to the IEEE
802.1Qbh standard. The values are network specific and should be provided by the network
administrator.
Additional parameters in the domain XML include:

385

Virt ualiz at ion Administ rat ion G uide

...
<devices>
...
<interface type='direct'>
<source dev='eth0' mode='private'/>
<virtualport type='802.1Qbh'>
<parameters profileid='finance'/>
</virtualport>
</interface>
</devices>
...
Fig u re 20.4 3. D evices - n et wo rk in t erf aces- d irect at t ach men t t o p h ysical in t erf aces
mo re ad d it io n al p aramet ers
The pro fi l ei d attribute, contains the name of the port profile that is to be applied to this interface.
This name is resolved by the port profile database into the network parameters from the port profile,
and those network parameters will be applied to this interface.

2 0 .1 6 .9 .7 . PCI passt hro ugh

A PCI network device (specified by the so urce element) is directly assigned to the guest virtual
machine using generic device passthrough, after first optionally setting the device's MAC address to
the configured value, and associating the device with an 802.1Qbh capable switch using an
optionally specified vi rtual po rt element (see the examples of virtualport given above for
type='direct' network devices). Note that - due to limitations in standard single-port PCI ethernet card
driver design - only SR-IOV (Single Root I/O Virtualization) virtual function (VF) devices can be
assigned in this manner; to assign a standard single-port PCI or PCIe ethernet card to a guest virtual
machine, use the traditional ho std ev device definition
Note that this " intelligent passthrough" of network devices is very similar to the functionality of a
standard ho std ev device, the difference being that this method allows specifying a MAC address
and vi rtual po rt for the passed-through device. If these capabilities are not required, if you have a
standard single-port PCI, PCIe, or USB network card that doesn't support SR-IOV (and hence would
anyway lose the configured MAC address during reset after being assigned to the guest virtual
machine domain), or if you are using a version of libvirt older than 0.9.11, you should use standard
ho std ev to assign the device to the guest virtual machine instead of i nterface
type= ' ho std ev' /.

...
<devices>
<interface type='hostdev'>
<driver name='vfio'/>
<source>
<address type='pci' domain='0x0000' bus='0x00' slot='0x07'
function='0x0'/>
</source>
<mac address='52:54:00:6d:90:02'>
<virtualport type='802.1Qbh'>
<parameters profileid='finance'/>

386

Chapt er 2 0 . Manipulat ing t he Domain XML

</virtualport>
</interface>
</devices>
...
Fig u re 20.4 4 . D evices - n et wo rk in t erf aces- PC I p asst h ro u g h

2 0 .1 6 .9 .8 . Mult icast t unne l

A multicast group may be used to represent a virtual network. Any guest virtual machine whose
network devices are within the same multicast group will talk to each other, even if they reside across
miltiple physical host physical machines. This mode may be used as an unprivileged user. There is
no default D NS or D HCP support and no outgoing network access. To provide outgoing network
access, one of the guest virtual machines should have a second NIC which is connected to one of
the first 4 network types in order to provide appropriate routing. The multicast protocol is compatible
with protocols used by user mo d e linux guest virtual machines as well. Note that the source
address used must be from the multicast address block. A multicast tunnel is created by manipulating
the i nterface type using a management tool and setting/changing it to mcast, and providing a
mac and source address. The result is shown in changes made to the domain XML:

...
<devices>
<interface type='mcast'>
<mac address='52:54:00:6d:90:01'>
<source address='230.0.0.1' port='5558'/>
</interface>
</devices>
...
Fig u re 20.4 5. D evices - n et wo rk in t erf aces- mu lt icast t u n n el

2 0 .1 6 .9 .9 . T CP t unne l
Creating a TCP client/server architecture is another way to provide a virtual network wher one guest
virtual machine provides the server end of the network and all other guest virtual machines are
configured as clients. All network traffic between the guest virtual machines is routed via the guest
virtual machine that is configured as the server. This model is also available for use to unprivileged
users. There is no default D NS or D HCP support and no outgoing network access. To provide
outgoing network access, one of the guest virtual machines should have a second NIC which is
connected to one of the first 4 network types thereby providing the appropriate routing. A TCP tunnel
is created by manipulating the i nterface type using a management tool and setting/changing it
to server or cl i ent, and providing a mac and source address. The result is shown in changes
made to the domain XML:

...
<devices>
<interface type='server'>
<mac address='52:54:00:22:c9:42'>

387

Virt ualiz at ion Administ rat ion G uide

<source address='192.168.0.1' port='5558'/>

</interface>
...
<interface type='client'>
<mac address='52:54:00:8b:c9:51'>
<source address='192.168.0.1' port='5558'/>
</interface>
</devices>
...
Fig u re 20.4 6 . D evices - n et wo rk in t erf aces- T C P t u n n el

2 0 .1 6 .9 .1 0 . Se t t ing NIC drive r-spe cific o pt io ns

Some NICs may have tunable driver-specific options. These options are set as attributes of the
d ri ver sub-element of the interface definition. These options are set by using management tools to
configuring the following sections of the domain XML:

<devices>
<interface type='network'>
<source network='default'/>
<target dev='vnet1'/>
<model type='virtio'/>
<driver name='vhost' txmode='iothread' ioeventfd='on'
event_idx='off'/>
</interface>
</devices>
...
Fig u re 20.4 7. D evices - n et wo rk in t erf aces- set t in g N IC d river- sp ecif ic o p t io n s
Currently the following attributes are available for the " virtio" NIC driver:
T ab le 20.19 . virt io N IC d river elemen t s
Paramet er

D escrip t io n

name

The optional name attribute forces which type of

backend driver to use. The value can be either
q emu (a user-space backend) or vho st (a
kernel backend, which requires the vhost
module to be provided by the kernel); an attempt
to require the vhost driver without kernel support
will be rejected. The default setting is vho st if
the vhost driver present, but will silently fall back
to q emu if not.

388

Chapt er 2 0 . Manipulat ing t he Domain XML

Paramet er

D escrip t io n

txmo d e

Specifies how to handle transmission of packets

when the transmit buffer is full. The value can be
either i o thread or ti mer. If set to i o thread ,
packet tx is all done in an iothread in the bottom
half of the driver (this option translates into
adding "tx= bh" to the q emu commandline device virtio-net-pci option). If set to ti mer, tx
work is done in qemu, and if there is more tx
data than can be sent at the present time, a timer
is set before qemu moves on to do other things;
when the timer fires, another attempt is made to
send more data. In general you should leave
this option alone, unless you are very certain
you that changing it is an absolute necessity.
Allows users to set domain I/O asynchronous
handling for interface device. The default is left
to the discretion of the hypervisor. Accepted
values are o n and o ff . Enabling this option
allows qemu to execute a guest virtual machine
while a separate thread handles I/O. Typically
guest virtual machines experiencing high
system CPU utilization during I/O will benefit
from this. On the other hand, overloading the
physical host physical machine may also
increase guest virtual machine I/O latency.
Therefore, you should leave this option alone,
unless you are very certain you that changing it
is an absolute necessity.
The event_idx attribute controls some aspects of
device event processing. The value can be
either o n or o ff. Choosing o n, reduces the
number of interrupts and exits for the guest
virtual machine. The default is o n. In case there
is a situation where this behavior is suboptimal,
this attribute provides a way to force the feature
off. You should leave this option alone, unless
you are very certain you that changing it is an
absolute necessity.

i o eventfd

event_i d x

2 0 .1 6 .9 .1 1 . Ove rriding t he t arge t e le m e nt

To override the target element, use a management tool to make the following changes to the domain
XML:

...
<devices>
<interface type='network'>
<source network='default'/>

389

Virt ualiz at ion Administ rat ion G uide

<target dev='vnet1'/>
</interface>
</devices>
...
Fig u re 20.4 8. D evices - n et wo rk in t erf aces- o verrid in g t h e t arg et elemen t
If no target is specified, certain hypervisors will automatically generate a name for the created tun
device. This name can be manually specified, however the name must not start with either 'vnet' or
'vif', which are prefixes reserved by libvirt and certain hypervisors. Manually specified targets using
these prefixes will be ignored.

2 0 .1 6 .9 .1 2 . Spe cifying bo o t o rde r

To specify the boot order, use a management tool to make the following changes to the domain XML:

...
<devices>
<interface type='network'>
<source network='default'/>
<target dev='vnet1'/>
<boot order='1'/>
</interface>
</devices>
...
Fig u re 20.4 9 . Sp ecif yin g b o o t o rd er
For hypervisors which support it, you can set a specific NIC to be used for the network boot. The
order of attributes determine the order in which devices will be tried during boot sequence. Note that
the per-device boot elements cannot be used together with general boot elements in BIOS bootloader
section.

2 0 .1 6 .9 .1 3. Int e rface ROM BIOS co nfigurat io n

To specify the ROM BIOS configuration settings, use a management tool to make the following
changes to the domain XML:

...
<devices>
<interface type='network'>
<source network='default'/>
<target dev='vnet1'/>
<rom bar='on' file='/etc/fake/boot.bin'/>
</interface>
</devices>
...

390

Chapt er 2 0 . Manipulat ing t he Domain XML

Fig u re 20.50. In t erf ace R O M B IO S co n f ig u rat io n

For hypervisors which support it, you can change how a PCI Network device's ROM is presented to
the guest virtual machine. The bar attribute can be set to o n or o ff, and determines whether or not
the device's ROM will be visible in the guest virtual machine's memory map. (In PCI documentation,
the " rombar" setting controls the presence of the Base Address Register for the ROM). If no rom bar is
specified, the qemu default will be used (older versions of qemu used a default of o ff, while newer
qemus have a default of o n). The optional fi l e attribute is used to point to a binary file to be
presented to the guest virtual machine as the device's ROM BIOS. This can be useful to provide an
alternative boot ROM for a network device.

2 0 .1 6 .9 .1 4 . Qualit y o f se rvice
This section of the domain XML provides setting quality of service. Incoming and outgoing traffic can
be shaped independently. The band wi d th element can have at most one inbound and at most one
outbound child elements. Leaving any of these children element out results in no QoS being applied
on that traffic direction. Therefore, when you want to shape only domain's incoming traffic, use
inbound only, and vice versa.
Each of these elements has one mandatory attribute averag e (or fl o o r as described below).
averag e specifies average bit rate on the interface being shaped. Then there are two optional
attributes: peak, which specifies maximum rate at which interface can send data, and burst, which
specifies the amount of bytes that can be burst at peak speed. Accepted values for attributes are
integer numbers.
The units for averag e and peak attributes are kilobytes per second, whereas burst is only set in
kilobytes. In addtion, inbound traffic can optionally have a fl o o r attribute. This guarantees minimal
throughput for shaped interfaces. Using the fl o o r requires that all traffic goes through one point
where QoS decisions can take place. As such it may only be used in cases where the i nterface
type= ' netwo rk' / with a fo rward type of ro ute, nat, or no forward at all). It should be noted that
within a virtual network, all connected interfaces are required to have at least the inbound QoS set
(averag e at least) but the floor attribute doesn't require specifying averag e. However, peak and
burst attributes still require averag e. At the present time, ingress qdiscs may not have any classes,
and therefore fl o o r may only be applied only on inbound and not outbound traffic.
To specify the QoS configuration settings, use a management tool to make the following changes to
the domain XML:

...
<devices>
<interface type='network'>
<source network='default'/>
<target dev='vnet0'/>
<bandwidth>
<inbound average='1000' peak='5000' floor='200' burst='1024'/>
<outbound average='128' peak='256' burst='256'/>
</bandwidth>
</interface>
<devices>
...
Fig u re 20.51. Q u alit y o f service

391

Virt ualiz at ion Administ rat ion G uide

2 0 .1 6 .9 .1 5 . Se t t ing VLAN t ag (o n suppo rt e d ne t wo rk t ype s o nly)

To specify the VLAN tag configuration settings, use a management tool to make the following
changes to the domain XML:

...
<devices>
<interface type='bridge'>
<vlan>
<tag id='42'/>
</vlan>
<source bridge='ovsbr0'/>
<virtualport type='openvswitch'>
<parameters interfaceid='09b11c53-8b5c-4eeb-8f00-d84eaa0aaa4f'/>
</virtualport>
</interface>
<devices>
...
Fig u re 20.52. Set t in g VLAN t ag ( o n su p p o rt ed n et wo rk t yp es o n ly)
If (and only if) the network connection used by the guest virtual machine supports vlan tagging
transparent to the guest virtual machine, an optional vl an element can specify one or more vlan
tags to apply to the guest virtual machine's network traffic (openvswitch and type= ' ho std ev' SRIOV interfaces do support transparent vlan tagging of guest virtual machine traffic; everything else,
including standard Linux bridges and libvirt's own virtual networks, do not support it. 802.1Qbh (vnlink) and 802.1Qbg (VEPA) switches provide their own way (outside of libvirt) to tag guest virtual
machine traffic onto specific vlans.) To allow for specification of multiple tags (in the case of vlan
trunking), a subelement, tag , specifies which vlan tag to use (for example: tag i d = ' 4 2' /. If an
interface has more than one vl an element defined, it is assumed that the user wants to do VLAN
trunking using all the specified tags. In the case that vlan trunking with a single tag is desired, the
optional attribute trunk= ' yes' can be added to the toplevel vlan element.

2 0 .1 6 .9 .1 6 . Mo difying virt ual link st at e

This element provides means of setting state of the virtual network link. Possible values for attribute
state are up and d o wn. If d o wn is specified as the value, the interface behaves as if it had the
network cable disconnected. D efault behavior if this element is unspecified is to have the link state
up.
To specify the virtual link state configuration settings, use a management tool to make the following
changes to the domain XML:

...
<devices>
<interface type='network'>
<source network='default'/>
<target dev='vnet0'/>

392

Chapt er 2 0 . Manipulat ing t he Domain XML

<link state='down'/>
</interface>
<devices>
...
Fig u re 20.53. Mo d if yin g virt u al lin k st at e

20.16.10. Input Devices

Input devices allow interaction with the graphical framebuffer in the guest virtual machine virtual
machine. When enabling the framebuffer, an input device is automatically provided. It may be
possible to add additional devices explicitly, for example, to provide a graphics tablet for absolute
cursor movement.
To specify the input devices configuration settings, use a management tool to make the following
changes to the domain XML:

...
<devices>
<input type='mouse' bus='usb'/>
</devices>
...
Fig u re 20.54 . In p u t d evices
The <i nput> element has one mandatory attribute: type which can be set to: mo use or tabl et. The
latter provides absolute cursor movement, while the former uses relative movement. The optional bus
attribute can be used to refine the exact device type and can be set to: xen (paravirtualized), ps2,
and usb.
The input element has an optional sub-element <ad d ress>, which can tie the device to a particular
PCI slot, as documented above.

20.16.11. Hub Devices

A hub is a device that expands a single port into several so that there are more ports available to
connect devices to a host physical machine system.
To specify the hub devices configuration settings, use a management tool to make the following
changes to the domain XML:

...
<devices>
<hub type='usb'/>
</devices>
...
Fig u re 20.55. H u b d evices

393

Virt ualiz at ion Administ rat ion G uide

The hub element has one mandatory attribute, the type whose value can only be usb. The hub
element has an optional sub-element ad d ress with type= ' usb' which can tie the device to a
particular controller.

20.16.12. Graphical framebuffers

A graphics device allows for graphical interaction with the guest virtual machine OS. A guest virtual
machine will typically have either a framebuffer or a text console configured to allow interaction with
the admin.
To specify the graphical framebuffer devices configuration settings, use a management tool to make
the following changes to the domain XML:

...
<devices>
<graphics type='sdl' display=':0.0'/>
<graphics type='vnc' port='5904'>
<listen type='address' address='192.0.2.1'/>
</graphics>
<graphics type='rdp' autoport='yes' multiUser='yes' />
<graphics type='desktop' fullscreen='yes'/>
<graphics type='spice'>
<listen type='network' network='rednet'/>
</graphics>
</devices>
...
Fig u re 20.56 . G rap h ical Frameb u f f ers
The g raphi cs element has a mandatory type attribute which takes the value sd l , vnc, rd p or
d eskto p as explained below:
T ab le 20.20. G rap h ical f rameb u f f er elemen t s
Paramet er

D escrip t io n

sd l

This displays a window on the host physical

machine desktop, it can take 3 optional
arguments: a d i spl ay attribute for the display
to use, an xauth attribute for the authentication
identifier, and an optional ful l screen
attribute accepting values yes or no

394

Chapt er 2 0 . Manipulat ing t he Domain XML

Paramet er

D escrip t io n

vnc

Starts a VNC server. The po rt attribute specifies

the TCP port number (with -1 as legacy syntax
indicating that it should be auto-allocated). The
auto po rt attribute is the new preferred syntax
for indicating autoallocation of the TCP port to
use. The l i sten attribute is an IP address for
the server to listen on. The passwd attribute
provides a VNC password in clear text. The
keymap attribute specifies the keymap to use. It
is possible to set a limit on the validity of the
password be giving an ti mestamp
passwd Val i d T o = ' 20 10 -0 4 0 9 T 15: 51: 0 0 ' assumed to be in UTC. The
co nnected attribute allows control of
connected client during password changes.
VNC accepts keep value only and note that it
may not be supported by all hypervisors. Rather
than using listen/port, QEMU supports a socket
attribute for listening on a unix domain socket
path.
Starts a SPICE server. The po rt attribute
specifies the TCP port number (with -1 as legacy
syntax indicating that it should be autoallocated), while tl sP o rt gives an alternative
secure port number. The auto po rt attribute is
the new preferred syntax for indicating autoallocation of both port numbers. The l i sten
attribute is an IP address for the server to listen
on. The passwd attribute provides a SPICE
password in clear text. The keymap attribute
specifies the keymap to use. It is possible to set
a limit on the validity of the password be giving
an ti mestamp passwd Val i d T o = ' 20 10 0 4 -0 9 T 15: 51: 0 0 ' assumed to be in UTC.
The co nnected attribute allows control of
connected client during password changes.
SPICE accepts keep to keep client connected,
disconnect to disconnect client and fail to fail
changing password. Note it is not be supported
by all hypervisors. The d efaul tMo d e attribute
sets the default channel security policy, valid
values are secure, i nsecure and the default
any (which is secure if possible, but falls back
to i nsecure rather than erroring out if no
secure path is available).

spi ce

When SPICE has both a normal and TLS secured TCP port configured, it may be desirable to restrict
what channels can be run on each port. This is achieved by adding one or more channel elements
inside the main g raphi cs element. Valid channel names include mai n, d i spl ay, i nputs, curso r,
pl ayback, reco rd ; smartcard ; and usbred i r.
To specify the SPICE configuration settings, use a mangement tool to make the following changes to
the domain XML:

395

Virt ualiz at ion Administ rat ion G uide

<graphics type='spice' port='-1' tlsPort='-1' autoport='yes'>

<channel name='main' mode='secure'/>
<channel name='record' mode='insecure'/>
<image compression='auto_glz'/>
<streaming mode='filter'/>
<clipboard copypaste='no'/>
<mouse mode='client'/>
</graphics>
Fig u re 20.57. SPIC E co n f ig u rat io n
SPICE supports variable compression settings for audio, images and streaming. These settings are
accessible via the compression attribute in all following elements: i mag e to set image compression
(accepts auto_glz, auto_lz, quic, glz, lz, off), jpeg for JPEG compression for images over wan
(accepts auto, never, always), zl i b for configuring wan image compression (accepts auto, never,
always) and pl ayback for enabling audio stream compression (accepts on or off).
Streaming mode is set by the streami ng element, settings its mo d e attribute to one of fi l ter, al l
or o ff.
In addition, Copy and paste functionality (via the SPICE agent) is set by the cl i pbo ard element. It
is enabled by default, and can be disabled by setting the co pypaste property to no .
Mouse mode is set by the mo use element, setting its mo d e attribute to one of server or cl i ent. If no
mode is specified, the qemu default will be used (cl i ent mode).
Additional elements include:
T ab le 20.21. Ad d it io n al g rap h ical f rameb u f f er elemen t s
Paramet er

D escrip t io n

rd p

Starts a RD P server. The port attribute specifies

the TCP port number (with -1 as legacy syntax
indicating that it should be auto-allocated). The
autoport attribute is the new preferred syntax for
indicating autoallocation of the TCP port to use.
The replaceUser attribute is a boolean deciding
whether multiple simultaneous connections to
the VM are permitted. The multiUser whether the
existing connection must be dropped and a new
connection must be established by the VRD P
server, when a new client connects in single
connection mode.
This value is reserved for VirtualBox domains
for the moment. It displays a window on the host
physical machine desktop, similarly to " sdl" , but
uses the VirtualBox viewer. Just like " sdl" , it
accepts the optional attributes display and
fullscreen.

d eskto p

396

Chapt er 2 0 . Manipulat ing t he Domain XML

Paramet er

D escrip t io n

l i sten

Rather than putting the address information

used to set up the listening socket for graphics
types vnc and spice in the g raphi cs, the
l i sten attribute, a separate subelement of
g raphi cs, called l i sten can be specified (see
the examples above). l i sten accepts the
following attributes:
type - Set to either address or network. This
tells whether this listen element is specifying
the address to be used directly, or by naming
a network (which will then be used to
determine an appropriate address for
listening).
ad d ress - this attribute will contain either an
IP address or hostname (which will be
resolved to an IP address via a D NS query)
to listen on. In the " live" XML of a running
domain, this attribute will be set to the IP
address used for listening, even if
type= ' netwo rk' .
network - if type= ' netwo rk' , the network
attribute will contain the name of a network in
libvirt's list of configured networks. The
named network configuration will be
examined to determine an appropriate listen
address. For example, if the network has an
IPv4 address in its configuration (e.g. if it has
a forward type of route, nat, or no forward
type (isolated)), the first IPv4 address listed
in the network's configuration will be used. If
the network is describing a host physical
machine bridge, the first IPv4 address
associated with that bridge device will be
used, and if the network is describing one of
the 'direct' (macvtap) modes, the first IPv4
address of the first forward dev will be used.

20.16.13. Video Devices

A video device.
To specify the video devices configuration settings, use a management tool to make the following
changes to the domain XML:

...
<devices>
<video>
<model type='vga' vram='8192' heads='1'>
<acceleration accel3d='yes' accel2d='yes'/>

397

Virt ualiz at ion Administ rat ion G uide

</model>
</video>
</devices>
...
Fig u re 20.58. Vid eo d evices
The g raphi cs element has a mandatory type attribute which takes the value " sdl" , " vnc" , " rdp" or
" desktop" as explained below:
T ab le 20.22. G rap h ical f rameb u f f er elemen t s
Paramet er

D escrip t io n

vi d eo

The vi d eo element is the container for

describing video devices. For backwards
compatibility, if no video is set but there is a
g raphi cs element in domain xml, then libvirt
will add a default vi d eo according to the guest
virtual machine type. If " ram" or " vram" are not
supplied a default value is used.
This has a mandatory type attribute which
takes the value vg a, ci rrus, vmvg a, xen,
vbo x, or q xl depending on the hypervisor
features available. You can also provide the
amount of video memory in kibibytes (blocks of
1024 bytes) using vram and the number of figure
with heads.
If acceleration is supported it should be enabled
using the accel 3d and accel 2d attributes in
the accel erati o n element.
The optional address sub-element can be used
to tie the video device to a particular PCI slot.

mo d el

acceleration

ad d ress

20.16.14 . Consoles, Serial, Parallel, and Channel Devices

A character device provides a way to interact with the virtual machine. Paravirtualized consoles,
serial ports, parallel ports and channels are all classed as character devices and so represented
using the same syntax.
To specify the consols, channel and other devices configuration settings, use a management tool to
make the following changes to the domain XML:

...
<devices>
<parallel type='pty'>
<source path='/dev/pts/2'/>
<target port='0'/>
</parallel>
<serial type='pty'>
<source path='/dev/pts/3'/>
<target port='0'/>

398

Chapt er 2 0 . Manipulat ing t he Domain XML

</serial>
<console type='pty'>
<source path='/dev/pts/4'/>
<target port='0'/>
</console>
<channel type='unix'>
<source mode='bind' path='/tmp/guestfwd'/>
<target type='guestfwd' address='10.0.2.1' port='4600'/>
</channel>
</devices>
...
Fig u re 20.59 . C o n so les, serial, p arallel, an d ch an n el d evices
In each of these directives, the top-level element name (parallel, serial, console, channel) describes
how the device is presented to the guest virtual machine. The guest virtual machine interface is
configured by the target element. The interface presented to the host physical machine is given in the
type attribute of the top-level element. The host physical machine interface is configured by the
source element. The source element may contain an optional seclabel to override the way that
labelling is done on the socket path. If this element is not present, the security label is inherited from
the per-domain setting. Each character device element has an optional sub-element ad d ress which
can tie the device to a particular controller or PCI slot.

20.16.15. Guest Virt ual Machine Int erfaces

A character device presents itself to the guest virtual machine as one of the following types.
To set the parallel port, use a management tool to make the following change to the domain XML

...
<devices>
<parallel type='pty'>
<source path='/dev/pts/2'/>
<target port='0'/>
</parallel>
</devices>
...
Fig u re 20.6 0. G u est virt u al mach in e in t erf ace Parallel Po rt
<targ et> can have a po rt attribute, which specifies the port number. Ports are numbered starting
from 0. There are usually 0, 1 or 2 parallel ports.
To set the serial port use a management tool to make the following change to the domain XML:

...
<devices>
<serial type='pty'>

399

Virt ualiz at ion Administ rat ion G uide

<source path='/dev/pts/3'/>
<target port='0'/>
</serial>
</devices>
...
Fig u re 20.6 1. G u est virt u al mach in e in t erf ace serial p o rt
<targ et> can have a po rt attribute, which specifies the port number. Ports are numbered starting
from 0. There are usually 0, 1 or 2 serial ports. There is also an optional type attribute, which has
two choices for its value, one is i sa-seri al , the other is usb-seri al . If type is missing, i saseri al will be used by default. For usb-serial an optional sub-element <ad d ress> with
type= ' usb' can tie the device to a particular controller, documented above.
The <co nso l e> element is used to represent interactive consoles. D epending on the type of guest
virtual machine in use, the consoles might be paravirtualized devices, or they might be a clone of a
serial device, according to the following rules:
If no targ etT ype attribute is set, then the default device type is according to the hypervisor's
rules. The default type will be added when re-querying the XML fed into libvirt. For fully virtualized
guest virtual machines, the default device type will usually be a serial port.
If the targ etT ype attribute is seri al , and if no <seri al > element exists, the console element
will be copied to the <seri al > element. If a <seri al > element does already exist, the console
element will be ignored.
If the targ etT ype attribute is not seri al , it will be treated normally.
Only the first <co nso l e> element may use a targ etT ype of seri al . Secondary consoles must
all be paravirtualized.
On s390, the console element may use a targetType of sclp or sclplm (line mode). SCLP is the
native console type for s390. There's no controller associated to SCLP consoles.
In the example below, a virtio console device is exposed in the guest virtual machine as /dev/hvc[0-7]
(for more information, see http://fedoraproject.org/wiki/Features/VirtioSerial):

...
<devices>
<console type='pty'>
<source path='/dev/pts/4'/>
<target port='0'/>
</console>
<!-- KVM virtio console -->
<console type='pty'>
<source path='/dev/pts/5'/>
<target type='virtio' port='0'/>
</console>
</devices>
...
...
<devices>

4 00

Chapt er 2 0 . Manipulat ing t he Domain XML

<!-- KVM s390 sclp console -->

<console type='pty'>
<source path='/dev/pts/1'/>
<target type='sclp' port='0'/>
</console>
</devices>
...
Fig u re 20.6 2. G u est virt u al mach in e in t erf ace - virt io co n so le d evice
If the console is presented as a serial port, the <targ et> element has the same attributes as for a
serial port. There is usually only one console.

20.16.16. Channel
This represents a private communication channel between the host physical machine and the guest
virtual machine and is manipulated by making changes to your guest virtual machine virtual
machine using a management tool that results in changes made to the following section of the
domain xml

...
<devices>
<channel type='unix'>
<source mode='bind' path='/tmp/guestfwd'/>
<target type='guestfwd' address='10.0.2.1' port='4600'/>
</channel>
<!-- KVM virtio channel -->
<channel type='pty'>
<target type='virtio' name='arbitrary.virtio.serial.port.name'/>
</channel>
<channel type='unix'>
<source mode='bind' path='/var/lib/libvirt/qemu/f16x86_64.agent'/>
<target type='virtio' name='org.qemu.guest_agent.0'/>
</channel>
<channel type='spicevmc'>
<target type='virtio' name='com.redhat.spice.0'/>
</channel>
</devices>
...
Fig u re 20.6 3. C h an n el
This can be implemented in a variety of ways. The specific type of <channel > is given in the type
attribute of the <targ et> element. D ifferent channel types have different target attributes as follows:
g uestfwd - D ictates that TCP traffic sent by the guest virtual machine to a given IP address and
port is forwarded to the channel device on the host physical machine. The targ et element must
have address and port attributes.
vi rti o - Paravirtualized virtio channel. <channel > is exposed in the guest virtual machine

4 01

Virt ualiz at ion Administ rat ion G uide

under /d ev/vpo rt*, and if the optional element nameis specified, /d ev/vi rti o -po rts/$name
(for more info, please see http://fedoraproject.org/wiki/Features/VirtioSerial). The optional element
ad d ress can tie the channel to a particular type= ' vi rti o -seri al ' controller, documented
above. With QEMU, if name is " org.qemu.guest_agent.0" , then libvirt can interact with a guest
virtual machine agent installed in the guest virtual machine, for actions such as guest virtual
machine shutdown or file system quiescing.
spi cevmc - Paravirtualized SPICE channel. The domain must also have a SPICE server as a
graphics device, at which point the host physical machine piggy-backs messages across the
main channel. The targ et element must be present, with attribute type= ' vi rti o ' ; an optional
attribute name controls how the guest virtual machine will have access to the channel, and
defaults to name= ' co m. red hat. spi ce. 0 ' . The optional <ad d ress> element can tie the
channel to a particular type= ' vi rti o -seri al ' controller.

20.16.17. Host Physical Machine Int erface

A character device presents itself to the host physical machine as one of the following types:
T ab le 20.23. C h aract er d evice elemen t s
Paramet er

D escrip t io n

D omain logfile

D isables all input on the

character device, and sends
output into the virtual
machine's logfile

XML sn ip p et

<d evi ces>

<co nso l e
type= ' std i o ' >
<targ et po rt= ' 1' />
</co nso l e>
</d evi ces>

D evice logfile

A file is opened and all data

sent to the character device is
written to the file.

<d evi ces>

<seri al type= "fi l e">
<so urce
path= "/var/l o g /vm/vm
-seri al . l o g "/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>

4 02

Chapt er 2 0 . Manipulat ing t he Domain XML

Paramet er

D escrip t io n

Virtual console

Connects the character device

to the graphical framebuffer in
a virtual console. This is
typically accessed via a special
hotkey sequence such as
" ctrl+alt+3"

XML sn ip p et

<d evi ces>

<seri al type= ' vc' >
<targ et po rt= "1"/>
</seri al >
</d evi ces>

Null device

Connects the character device

to the void. No data is ever
provided to the input. All data
written is discarded.

<d evi ces>

<seri al
type= ' nul l ' >
<targ et po rt= "1"/>
</seri al >
</d evi ces>

Pseudo TTY

A Pseudo TTY is allocated

using /d ev/ptmx. A suitable
client such as vi rsh
co nso l e can connect to
interact with the serial port
locally.

<d evi ces>

<seri al type= "pty">
<so urce
path= "/d ev/pts/3"/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>

NB Special case

NB special case if <co nso l e

type= ' pty' >, then the TTY
path is also duplicated as an
attribute tty= ' /d ev/pts/3'
on the top level <co nso l e>
tag. This provides compat with
existing syntax for <co nso l e>
tags.

4 03

Virt ualiz at ion Administ rat ion G uide

Paramet er

D escrip t io n

Host physical machine device

proxy

The character device is passed

through to the underlying
physical character device. The
device types must match, eg the
emulated serial port should
only be connected to a host
physical machine serial port don't connect a serial port to a
parallel port.

XML sn ip p et

<d evi ces>

<seri al type= "d ev">
<so urce
path= "/d ev/ttyS0 "/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>

Named pipe

The character device writes

output to a named pipe. See
pipe(7) man page for more info.

<d evi ces>

<seri al type= "pi pe">
<so urce
path= "/tmp/mypi pe"/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>

TCP client/server

The character device acts as a

TCP client connecting to a
remote server.

<d evi ces>

<seri al type= "tcp">
<so urce
mo d e= "co nnect"
ho st= "0 . 0 . 0 . 0 "
servi ce= "24 4 5"/>
<pro to co l
type= "raw"/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>

4 04

Chapt er 2 0 . Manipulat ing t he Domain XML

Paramet er

D escrip t io n

Or
assn
a TCP
XML
ip p etserver waiting for
a client connection.

<d evi ces>

<seri al type= "tcp">
<so urce mo d e= "bi nd "
ho st= "127. 0 . 0 . 1"
servi ce= "24 4 5"/>
<pro to co l
type= "raw"/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>
Alternatively you can use telnet
instead of raw TCP. In addition,
you can also use telnets
(secure telnet) and tls.

<d evi ces>

<seri al type= "tcp">
<so urce
mo d e= "co nnect"
ho st= "0 . 0 . 0 . 0 "
servi ce= "24 4 5"/>
<pro to co l
type= "tel net"/>
<targ et po rt= "1"/>
</seri al >
<seri al type= "tcp">
<so urce mo d e= "bi nd "
ho st= "127. 0 . 0 . 1"
servi ce= "24 4 5"/>
<pro to co l
type= "tel net"/>
<targ et po rt= "1"/>

4 05

Virt ualiz at ion Administ rat ion G uide

Paramet er

D escrip t io n

UD P network console

The character device acts as a

UD P netconsole service,
sending and receiving packets.
This is a lossy service.

XML sn ip p et
</seri al >
</d evi ces>
<d evi ces>
<seri al type= "ud p">
<so urce mo d e= "bi nd "
ho st= "0 . 0 . 0 . 0 "
servi ce= "24 4 5"/>
<so urce
mo d e= "co nnect"
ho st= "0 . 0 . 0 . 0 "
servi ce= "24 4 5"/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>

UNIX domain socket

client/server

The character device acts as a

UNIX domain socket server,
accepting connections from
local clients.

<d evi ces>

<seri al type= "uni x">
<so urce mo d e= "bi nd "
path= "/tmp/fo o "/>
<targ et po rt= "1"/>
</seri al >
</d evi ces>

20.17. Sound Devices

A virtual sound card can be attached to the host physical machine via the sound element.

...
<devices>
<sound model='es1370'/>
</devices>
...
Fig u re 20.6 4 . Virt u al so u n d card

4 06

Chapt er 2 0 . Manipulat ing t he Domain XML

The sound element has one mandatory attribute, mo d el , which specifies what real sound device is
emulated. Valid values are specific to the underlying hypervisor, though typical choices are
' es1370 ' , ' sb16 ' , ' ac9 7' , and ' i ch6 ' . In addition, a sound element with ich6 model can have
optional sub-elements co d ec to attach various audio codecs to the audio device. If not specified, a
default codec will be attached to allow playback and recording. Valid values are ' d upl ex'
(advertises a line-in and a line-out) and ' mi cro ' (advertises a speaker and a microphone).

...
<devices>
<sound model='ich6'>
<codec type='micro'/>
<sound/>
</devices>
...
Fig u re 20.6 5. So u n d d evices
Each sound element has an optional sub-element <ad d ress> which can tie the device to a
particular PCI slot, documented above.

20.18. Wat chdog Device

A virtual hardware watchdog device can be added to the guest virtual machine via the <watchd o g >
element. The watchdog device requires an additional driver and management daemon in the guest
virtual machine. As merely enabling the watchdog in the libvirt configuration does not do anything
useful on its own. Currently there is no support notification when the watchdog fires.

...
<devices>
<watchdog model='i6300esb'/>
</devices>
...
...
<devices>
<watchdog model='i6300esb' action='poweroff'/>
</devices>
</domain>
Fig u re 20.6 6 . Wat ch d o g D evice
The following attributes are declared in this XML:
mo d el - The required mo d el attribute specifies what real watchdog device is emulated. Valid
values are specific to the underlying hypervisor.
The mo d el attribute may take the following values:

4 07

Virt ualiz at ion Administ rat ion G uide

i 6 30 0 esb the recommended device, emulating a PCI Intel 6300ESB

i b70 0 emulates an ISA iBase IB700
acti o n - The optional acti o n attribute describes what action to take when the watchdog
expires. Valid values are specific to the underlying hypervisor. The acti o n attribute can have the
following values:
reset default setting, forcefully resets the guest virtual machine
shutd o wn gracefully shuts down the guest virtual machine (not recommended)
po wero ff forcefully powers off the guest virtual machine
pause pauses the guest virtual machine
no ne does nothing
d ump automatically dumps the guest virtual machine.
Note that the 'shutdown' action requires that the guest virtual machine is responsive to ACPI signals.
In the sort of situations where the watchdog has expired, guest virtual machines are usually unable
to respond to ACPI signals. Therefore using 'shutdown' is not recommended. In addition, the
directory to save dump files can be configured by auto_dump_path in file /etc/libvirt/qemu.conf.

20.19. Memory Balloon Device

A virtual memory balloon device is added to all Xen and KVM/QEMU guest virtual machines. It will be
seen as <membal l o o n> element. It will be automatically added when appropriate, so there is no
need to explicitly add this element in the guest virtual machine XML unless a specific PCI slot needs
to be assigned. Note that if the memballoon device needs to be explicitly disabled, mo d el = ' no ne'
may be used.
The following example automatically added device with KVM

...
<devices>
<memballoon model='virtio'/>
</devices>
...
Fig u re 20.6 7. Memo ry b allo o n d evice
Here is an example where the device is added manually with static PCI slot 2 requested

...
<devices>
<memballoon model='virtio'>
<address type='pci' domain='0x0000' bus='0x00' slot='0x02'

4 08

Chapt er 2 0 . Manipulat ing t he Domain XML

function='0x0'/>
</memballoon>
</devices>
</domain>
Fig u re 20.6 8. Memo ry b allo o n d evice ad d ed man u ally
The required mo d el attribute specifies what type of balloon device is provided. Valid values are
specific to the virtualization platform are: ' vi rti o ' which is the default setting with the KVM
hypervisor or ' xen' which is the default setting with the Xen hypervisor.

20.20. Securit y Label

The <secl abel > element allows control over the operation of the security drivers. There are three
basic modes of operation, ' d ynami c' where libvirt automatically generates a unique security label,
' stati c' where the application/administrator chooses the labels, or ' no ne' where confinement is
disabled. With dynamic label generation, libvirt will always automatically relabel any resources
associated with the virtual machine. With static label assignment, by default, the administrator or
application must ensure labels are set correctly on any resources. However, automatic relabeling can
be enabled if desired.
If more than one security driver is used by libvirt, multiple seclabel tags can be used, one for each
driver and the security driver referenced by each tag can be defined using the attribute mo d el Valid
input XML configurations for the top-level security label are:

<seclabel type='dynamic' model='selinux'/>

<seclabel type='dynamic' model='selinux'>
<baselabel>system_u:system_r:my_svirt_t:s0</baselabel>
</seclabel>
<seclabel type='static' model='selinux' relabel='no'>
<label>system_u:system_r:svirt_t:s0:c392,c662</label>
</seclabel>
<seclabel type='static' model='selinux' relabel='yes'>
<label>system_u:system_r:svirt_t:s0:c392,c662</label>
</seclabel>
<seclabel type='none'/>
Fig u re 20.6 9 . Secu rit y lab el
If no ' type' attribute is provided in the input XML, then the security driver default setting will be
used, which may be either ' no ne' or ' d ynami c' . If a <basel abel > is set but no ' type' is set,
then the type is presumed to be ' d ynami c' . When viewing the XML for a running guest virtual
machine with automatic resource relabeling active, an additional XML element, i mag el abel , will be
included. This is an output-only element, so will be ignored in user supplied XML documents.
The following elements can be manipulated with the following values:

4 09

Virt ualiz at ion Administ rat ion G uide

type - Either stati c, d ynami c or no ne to determine whether libvirt automatically generates a

unique security label or not.
mo d el - A valid security model name, matching the currently activated security model
rel abel - Either yes or no . This must always be yes if dynamic label assignment is used. With
static label assignment it will default to no .
<l abel > - If static labelling is used, this must specify the full security label to assign to the virtual
domain. The format of the content depends on the security driver in use:
SELi nux: a SELinux context.
AppArmo r: an AppArmor profile.
D AC : owner and group separated by colon. They can be defined both as user/group names or
uid/gid. The driver will first try to parse these values as names, but a leading plus sign can
used to force the driver to parse them as uid or gid.
<basel abel > - If dynamic labelling is used, this can optionally be used to specify the base
security label. The format of the content depends on the security driver in use.
<i mag el abel > - This is an output only element, which shows the security label used on
resources associated with the virtual domain. The format of the content depends on the security
driver in use When relabeling is in effect, it is also possible to fine-tune the labeling done for
specific source file names, by either disabling the labeling (useful if the file lives on NFS or other
file system that lacks security labeling) or requesting an alternate label (useful when a
management application creates a special label to allow sharing of some, but not all, resources
between domains). When a seclabel element is attached to a specific path rather than the top-level
domain assignment, only the attribute relabel or the sub-element label are supported.

20.21. Example Domain XML Configurat ion

QEMU emulated guest virtual machine on x86_64

<domain type='qemu'>
<name>QEmu-fedora-i686</name>
<uuid>c7a5fdbd-cdaf-9455-926a-d65c16db1809</uuid>
<memory>219200</memory>
<currentMemory>219200</currentMemory>
<vcpu>2</vcpu>
<os>
<type arch='i686' machine='pc'>hvm</type>
<boot dev='cdrom'/>
</os>
<devices>
<emulator>/usr/bin/qemu-system-x86_64</emulator>
<disk type='file' device='cdrom'>
<source file='/home/user/boot.iso'/>
<target dev='hdc'/>
<readonly/>
</disk>
<disk type='file' device='disk'>
<source file='/home/user/fedora.img'/>

4 10

Chapt er 2 0 . Manipulat ing t he Domain XML

<target dev='hda'/>
</disk>
<interface type='network'>
<source network='default'/>
</interface>
<graphics type='vnc' port='-1'/>
</devices>
</domain>
Fig u re 20.70. Examp le d o main XML co n f ig
KVM hardware accelerated guest virtual machine on i686

<domain type='kvm'>
<name>demo2</name>
<uuid>4dea24b3-1d52-d8f3-2516-782e98a23fa0</uuid>
<memory>131072</memory>
<vcpu>1</vcpu>
<os>
<type arch="i686">hvm</type>
</os>
<clock sync="localtime"/>
<devices>
<emulator>/usr/bin/qemu-kvm</emulator>
<disk type='file' device='disk'>
<source file='/var/lib/libvirt/images/demo2.img'/>
<target dev='hda'/>
</disk>
<interface type='network'>
<source network='default'/>
<mac address='24:42:53:21:52:45'/>
</interface>
<graphics type='vnc' port='-1' keymap='de'/>
</devices>
</domain>
Fig u re 20.71. Examp le d o main XML co n f ig

4 11

Virt ualiz at ion Administ rat ion G uide

Chapter 21. Troubleshooting

This chapter covers common problems and solutions for Red Hat Enterprise Linux 6 virtualization
issues.
Read this chapter to develop an understanding of some of the common problems associated with
virtualization technologies. Troubleshooting takes practice and experience which are difficult to
learn from a book. It is recommended that you experiment and test virtualization on Red Hat
Enterprise Linux 6 to develop your troubleshooting skills.
If you cannot find the answer in this document there may be an answer online from the virtualization
community. Refer to Section B.1, Online Resources for a list of Linux virtualization websites.

21.1. Debugging and T roubleshoot ing T ools

This section summarizes the System Administrator applications, the networking utilities, and
debugging tools. You can employ these standard System administration tools and logs to assist with
troubleshooting:
kvm_stat - refer to Section 21.4, kvm_stat
trace-cmd
ftrace Refer to the Red Hat Enterprise Linux Developer Guide
vmstat
i o stat
l so f
systemtap
crash
sysrq
sysrq t
sysrq w
These networking tools can assist with troubleshooting virtualization networking problems:
i fco nfi g
tcpd ump
The tcpd ump command 'sniffs' network packets. tcpd ump is useful for finding network
abnormalities and problems with network authentication. There is a graphical version of tcpd ump
named wi reshark.
brctl
brctl is a networking tool that inspects and configures the Ethernet bridge configuration in the
Linux kernel. You must have root access before performing these example commands:
# brctl show
bridge-name

4 12

bridge-id

STP

enabled

interfaces

Chapt er 2 1 . T roubleshoot ing

---------------------------------------------------------------------------virtbr0
8000.feffffff
yes
eth0
# brctl showmacs virtbr0
port-no
mac-addr
1
fe:ff:ff:ff:ff:
2
fe:ff:ff:fe:ff:
# brctl showstp virtbr0
virtbr0
bridge-id
8000.fefffffffff
designated-root
8000.fefffffffff
root-port
0
max-age
20.00
hello-time
2.00
forward-delay
0.00
aging-time
300.01
hello-timer
1.43
topology-change-timer 0.00

local?
yes
yes

aging timer
0.00
0.00

path-cost
bridge-max-age
bridge-hello-time
bridge-forward-delay

0
20.00
2.00
0.00

tcn-timer
gc-timer

0.00
0.02

Listed below are some other useful commands for troubleshooting virtualization.
st race is a command which traces system calls and events received and used by another
process.
vn cviewer: connect to a VNC server running on your server or a virtual machine. Install
vn cviwer using the yum i nstal l vnc command.
vn cserver: start a remote desktop on your server. Gives you the ability to run graphical user
interfaces such as virt-manager via a remote session. Install vn cserver using the yum i nstal l
vnc-server command.

21.2. Preparing for Disast er Recovery

If at all possible, it is best to prepare for situations where your equipment will be compromised due to
weather, or other reasons. It is highly recommended that you perform a backup of the following files
and directories on your host physical machine:
From the /etc/l i bvi rt directory, all files.
From the /var/l i b/l i bvi rt directory, back up the following items:
The current dnsmasq D HCP leases found in /var/l i b/l i bvi rt/d nsmasq
The running virtual network configuration files found in /var/l i b/l i bvi rt/netwo rk
Guest virtual machine files created by virt - man ag er when saving the current state of a guest,
if there are any. These can be found in the /var/l i b/l i bvi rt/q emu/save/ directory. If the
virsh save co mmand was used to create virtual machines, the files can be found in the
location specified for vi rsh save by the user.
The guest virtual machine snapshot files, created by the q emu-i mg create and vi rsh
snapsho t-create commands and found in the locations specified for the commands by the
user.
Guest virtual machine disk images created by virt - man ag er, if any, which can be found in the
/var/l i b/l i bvi rt/i mag es/ directory. If the vi rsh po o l -d efi ne command was used

4 13

Virt ualiz at ion Administ rat ion G uide

to create virtual storage, the image file can be found in the location specified for vi rsh
po o l -d efi ne by the user. For instructions on how to back up the guest image files, use the
steps described in Procedure 21.1, Creating a backup of the guest virtual machine's disk
image for disaster recovery purposes .
If you are using bridges, you will also need to back up the files located in
/etc/sysco nfi g /netwo rk-scri pts/i fcfg -<bri d g e_name>
Optionally, the guest virtual machine core dump files found in /var/l i b/l i bvi rt/q emu/d ump
can also be backed up to be used for analysing the causes of the failure. Note, however, that
these files can be very large for some systems.
Pro ced u re 21.1. C reat in g a b acku p o f t h e g u est virt u al mach in e' s d isk imag e f o r
d isast er reco very p u rp o ses
This procedure will cover how to back up several different disk image types.
1. To back up only the guest virtual machine disk image, back up the files located in
/var/l i b/l i bvi rt/i mag es. To back up guest virtual machine disk images with LVM
logical volumes, run the following command:
# lvcreate --snapshot --name snap --size 8G /dev/vg0/data
This command creates a snapshot volume named snap with a size of 8G as part of a 64G
volume.
2. Create a file for the snapshots using a command similar to this one:
# mkdir /mnt/virt.snapshot
3. Mount the directory you created and the snapshot volume using the following command:
# mount /dev/vg0/snap /mnt/virt.snapshot
4. Use one of the following commands to back up the volume:
a.

# tar -pzc -f /mnt/backup/virt-snapshot-MM-DD-YYYY.tgz

/mnt/virt.snapshot++++++++++++

b.

# rsync -a /mnt/virt.snapshot/ /mnt/backup/virt-snapshot.MMDD-YYYY/

21.3. Creat ing virsh Dump Files

Executing a vi rsh d ump command sends a request to dump the core of a guest virtual machine to a
file so errors in the virtual machine can be diagnosed. Running this command may require you to
manually ensure proper permissions on file and path specified by the argument co refi l epath. The
vi rsh d ump command is similar to a coredump (or the crash utility). To create the vi rsh d ump
file, run:
#vi rsh d ump <d o mai n> <co refi l epath> [--bypass-cache] { [--l i ve] | [-crash] | [--reset] } [--verbo se] [--memo ry-o nl y]

4 14

Chapt er 2 1 . T roubleshoot ing

While the domain (guest virtual machine domain name) and corefilepath (location of the newly
created core dump file) are mandatory, the following arguments are optional:
--l i ve creates a dump file on a running machine and doesn't pause it.
--crash stops the guest virtual machine and generates the dump file. The main difference is that
the guest virtual machine will not be listed as Stopped, with the reason as Crashed. Note that in
virt - man ag er the status will be listed as Paused.
--reset will reset the guest virtual machine following a successful dump. Note, these three
switches are mutually exclusive.
--bypass-cache uses O_D IRECT to bypass the file system cache.
--memo ry-o nl y the dump file will be saved as an elf file, and will only include domains memory
and cpu common register value. This option is very useful if the domain uses host devices
directly.
--verbo se displays the progress of the dump
The entire dump process may be monitored using vi rsh d o mjo bi nfo command and can be
canceled by running vi rsh d o mjo babo rt.

21.4 . kvm_st at
The kvm_stat command is a python script which retrieves runtime statistics from the kvm kernel
module. The kvm_stat command can be used to diagnose guest behavior visible to kvm. In
particular, performance related issues with guests. Currently, the reported statistics are for the entire
system; the behavior of all running guests is reported. To run this script you need to install the qemukvm-tools package.
The kvm_stat command requires that the kvm kernel module is loaded and d ebug fs is mounted. If
either of these features are not enabled, the command will output the required steps to enable
d ebug fs or the kvm module. For example:
# kvm_stat
Please mount debugfs ('mount -t debugfs debugfs /sys/kernel/debug')
and ensure the kvm modules are loaded
Mount d ebug fs if required:
# mount -t debugfs debugfs /sys/kernel/debug
kvm_st at O u t p u t
The kvm_stat command outputs statistics for all guests and the host. The output is updated until the
command is terminated (using C trl +c or the q key).
# kvm_stat
kvm statistics
efer_reload
exits
fpu_reload
halt_exits

94
4003074
1313881
14050

0
31272
10796
259

4 15

Virt ualiz at ion Administ rat ion G uide

halt_wakeup
4496
203
host_state_reload 1638354
24893
hypercalls
0
0
insn_emulation
1093850
1909
insn_emulation_fail
0
0
invlpg
75569
0
io_exits
1596984
24509
irq_exits
21013
363
irq_injections
48039
1222
irq_window
24656
870
largepages
0
0
mmio_exits
11873
0
mmu_cache_miss
42565
8
mmu_flooded
14752
0
mmu_pde_zapped
58730
0
mmu_pte_updated
6
0
mmu_pte_write
138795
0
mmu_recycled
0
0
mmu_shadow_zapped
40358
0
mmu_unsync
793
0
nmi_injections
0
0
nmi_window
0
0
pf_fixed
697731
3150
pf_guest
279349
0
remote_tlb_flush
5
0
request_irq
0
0
signal_exits
1
0
tlb_flush
200190
0
Exp lan at io n o f variab les:
ef er_relo ad
The number of Extended Feature Enable Register (EFER) reloads.
exit s
The count of all VMEXIT calls.
f p u _relo ad
The number of times a VMENT R Y reloaded the FPU state. The fpu_rel o ad is incremented
when a guest is using the Floating Point Unit (FPU).
h alt _exit s
Number of guest exits due to hal t calls. This type of exit is usually seen when a guest is
idle.
h alt _wakeu p
Number of wakeups from a hal t.
h o st _st at e_relo ad
Count of full reloads of the host state (currently tallies MSR setup and guest MSR reads).
h yp ercalls

4 16

Chapt er 2 1 . T roubleshoot ing

Number of guest hypervisor service calls.

in sn _emu lat io n
Number of guest instructions emulated by the host.
in sn _emu lat io n _f ail
Number of failed i nsn_emul ati o n attempts.
io _exit s
Number of guest exits from I/O port accesses.
irq _exit s
Number of guest exits due to external interrupts.
irq _in ject io n s
Number of interrupts sent to guests.
irq _win d o w
Number of guest exits from an outstanding interrupt window.
larg ep ag es
Number of large pages currently in use.
mmio _exit s
Number of guest exits due to memory mapped I/O (MMIO) accesses.
mmu _cach e_miss
Number of KVM MMU shadow pages created.
mmu _f lo o d ed
D etection count of excessive write operations to an MMU page. This counts detected write
operations not of individual write operations.
mmu _p d e_z ap p ed
Number of page directory entry (PD E) destruction operations.
mmu _p t e_u p d at ed
Number of page table entry (PTE) destruction operations.
mmu _p t e_writ e
Number of guest page table entry (PTE) write operations.
mmu _recycled
Number of shadow pages that can be reclaimed.
mmu _sh ad o w_z ap p ed
Number of invalidated shadow pages.

4 17

Virt ualiz at ion Administ rat ion G uide

mmu _u n syn c
Number of non-synchronized pages which are not yet unlinked.
n mi_in ject io n s
Number of Non-maskable Interrupt (NMI) injections to the guest.
n mi_win d o w
Number of guest exits from (outstanding) Non-maskable Interrupt (NMI) windows.
p f _f ixed
Number of fixed (non-paging) page table entry (PTE) maps.
p f _g u est
Number of page faults injected into guests.
remo t e_t lb _f lu sh
Number of remote (sibling CPU) Translation Lookaside Buffer (TLB) flush requests.
req u est _irq
Number of guest interrupt window request exits.
sig n al_exit s
Number of guest exits due to pending signals from the host.
t lb _f lu sh
Number of tl b_fl ush operations performed by the hypervisor.

Note
The output information from the kvm_stat command is exported by the KVM hypervisor as
pseudo files located in the /sys/kernel /d ebug /kvm/ directory.

21.5. Guest Virt ual Machine Fails t o Shut down

Traditionally, executing a vi rsh shutd o wn command causes a power button ACPI event to be sent,
thus copying the same action as when someone presses a power button on a physical machine.
Within every physical machine, it is up to the OS to handle this event. In the past operating systems
would just silently shutdown. Today, the most usual action is to show a dialog asking what should
be done. Some operating systems even ignore this event completely, especially when no users are
logged in. When such operating systems are installed on a guest virtual machine, running vi rsh
shutd o wn just does not work (it is either ignored or a dialog is shown on a virtual display). However,
if a qemu-guest-agent channel is added to a guest virtual machine and this agent is running inside
the guest virtual machine's OS, the vi rsh shutd o wn command will ask the agent to shutdown the
guest OS instead of sending the ACPI event. The agent will call for a shutdown from inside the guest
virtual machine OS and everything works as expected.
Pro ced u re 21.2. C o n f ig u rin g t h e g u est ag en t ch an n el in a g u est virt u al mach in e

4 18

Chapt er 2 1 . T roubleshoot ing

1. Stop the guest virtual machine.

2. Open the D omain XML for the guest virtual machine and add the following snippet:

<channel type='unix'>
<source mode='bind'/>
<target type='virtio' name='org.qemu.guest_agent.0'/>
</channel>
Fig u re 21.1. C o n f ig u rin g t h e g u est ag en t ch an n el
3. Start the guest virtual machine, by running vi rsh start [d o mai n].
4. Install qemu-guest-agent on the guest virtual machine (yum i nstal l q emu-g uest-ag ent)
and make it run automatically at every boot as a service (qemu-guest-agent.service). Refer to
Chapter 10, QEMU-img and QEMU Guest Agent for more information.

21.6. T roubleshoot ing wit h Serial Consoles

Linux kernels can output information to serial ports. This is useful for debugging kernel panics and
hardware issues with video devices or headless servers. The subsections in this section cover setting
up serial console output for host physical machines using the KVM hypervisor.
This section covers how to enable serial console output for fully virtualized guests.
Fully virtualized guest serial console output can be viewed with the vi rsh co nso l e command.
Be aware fully virtualized guest serial consoles have some limitations. Present limitations include:
output data may be dropped or scrambled.
The serial port is called ttyS0 on Linux or C O M1 on Windows.
You must configure the virtualized operating system to output information to the virtual serial port.
To output kernel information from a fully virtualized Linux guest into the domain, modify the
/bo o t/g rub/g rub. co nf file. Append the following to the kernel line: console=tty0
console=ttyS0,115200.
title Red Hat Enterprise Linux Server (2.6.32-36.x86-64)
root (hd0,0)
kernel /vmlinuz-2.6.32-36.x86-64 ro root=/dev/volgroup00/logvol00 \
console=tty0 console=ttyS0,115200
initrd /initrd-2.6.32-36.x86-64.img
Reboot the guest.
On the host, access the serial console with the following command:
# virsh console

4 19

Virt ualiz at ion Administ rat ion G uide

You can also use vi rt-manag er to display the virtual text console. In the guest console window,
select Seri al 1 in T ext C o nso l es from the Vi ew menu.

21.7. Virt ualiz at ion Log Files

Each fully virtualized guest log is in the /var/l o g /l i bvi rt/q emu/ directory. Each guest log is
named as GuestName.log and will be periodically compressed once a size limit is reached.
If you encounter any errors with the Virtual Machine Manager, you can review the generated data in
the vi rt-manag er. l o g file that resides in the $HO ME/. vi rt-manag er directory.

21.8. Loop Device Errors

If file-based guest images are used you may have to increase the number of configured loop devices.
The default configuration allows up to eight active loop devices. If more than eight file-based guests
or loop devices are needed the number of loop devices configured can be adjusted in the
/etc/mo d pro be. d /directory. Add the following line:
options loop max_loop=64
This example uses 64 but you can specify another number to set the maximum loop value. You may
also have to implement loop device backed guests on your system. To use a loop device backed
guests for a full virtualized system, use the phy: d evi ce or fi l e: fi l e commands.

21.9. Live Migrat ion Errors

There may be cases where a live migration causes the memory contents to be re-transferred over and
over again. This process causes the guest to be in a state where it is constantly writing to memory
and therefore will slow down migration. If this should occur, and the guest is writing more than
several tens of MBs per second, then live migration may fail to finish (converge). This issue is not
scheduled to be resolved at the moment for Red Hat Enterprise Linux 6, and is scheduled to be fixed
in Red Hat Enterprise Linux 7.
The current live-migration implementation has a default migration time configured to 30ms. This
value determines the guest pause time at the end of the migration in order to transfer the leftovers.
Higher values increase the odds that live migration will converge

21.10. Enabling Int el VT -x and AMD-V Virt ualiz at ion Hardware

Ext ensions in BIOS
Note
To expand your expertise, you might also be interested in the Red Hat Enterprise Virtualization
(RH318) training course.
This section describes how to identify hardware virtualization extensions and enable them in your
BIOS if they are disabled.
The Intel VT-x extensions can be disabled in the BIOS. Certain laptop vendors have disabled the
Intel VT-x extensions by default in their CPUs.

4 20

Chapt er 2 1 . T roubleshoot ing

The virtualization extensions cannot be disabled in the BIOS for AMD -V.
Refer to the following section for instructions on enabling disabled virtualization extensions.
Verify the virtualization extensions are enabled in BIOS. The BIOS settings for Intel VT or AMD -V are
usually in the C h ip set or Pro cesso r menus. The menu names may vary from this guide, the
virtualization extension settings may be found in Securi ty Setti ng s or other non standard menu
names.
Pro ced u re 21.3. En ab lin g virt u aliz at io n ext en sio n s in B IO S
1. Reboot the computer and open the system's BIOS menu. This can usually be done by
pressing the d el ete key, the F1 key or Al t and F4 keys depending on the system.
2. En ab lin g t h e virt u aliz at io n ext en sio n s in B IO S

Note
Many of the steps below may vary depending on your motherboard, processor type,
chipset and OEM. Refer to your system's accompanying documentation for the correct
information on configuring your system.
a. Open the P ro cesso r submenu The processor settings menu may be hidden in the
C hi pset, Ad vanced C P U C o nfi g urati o n or No rthbri d g e.
b. Enable Intel Vi rtual i zati o n T echno l o g y (also known as Intel VT-x). AMD -V
extensions cannot be disabled in the BIOS and should already be enabled. The
virtualization extensions may be labeled Vi rtual i zati o n Extensi o ns,
Vand erpo o l or various other names depending on the OEM and system BIOS.
c. Enable Intel VT-d or AMD IOMMU, if the options are available. Intel VT-d and AMD
IOMMU are used for PCI device assignment.
d. Select Save & Exi t.
3. Reboot the machine.
4. When the machine has booted, run cat /pro c/cpui nfo | g rep -E "vmx| svm".
Specifying --color is optional, but useful if you want the search term highlighted. If the
command outputs, the virtualization extensions are now enabled. If there is no output your
system may not have the virtualization extensions or the correct BIOS setting enabled.

21.11. KVM Net working Performance

By default, KVM virtual machines are assigned a virtual Realtek 8139 (rtl8139) NIC (network interface
controller). Whereas Red Hat Enterprise Linux guests are assigned a virtio NIC by default, Windows
guests or the guest type is not specified.
The rtl8139 virtualized NIC works fine in most environments,but this device can suffer from
performance degradation problems on some networks, such as a 10 Gigabit Ethernet.
To improve performance, you can switch to the paravirtualized network driver.

4 21

Virt ualiz at ion Administ rat ion G uide

Note
Note that the virtualized Intel PRO/1000 (e10 0 0 ) driver is also supported as an emulated
driver choice. To use the e10 0 0 driver, replace vi rti o in the procedure below with e10 0 0 .
For the best performance it is recommended to use the vi rti o driver.

Pro ced u re 21.4 . Swit ch in g t o t h e virt io d river

1. Shutdown the guest operating system.
2. Edit the guest's configuration file with the vi rsh command (where GUEST is the guest's
name):
# virsh edit GUEST
The vi rsh ed i t command uses the $ED IT O R shell variable to determine which editor to
use.
3. Find the network interface section of the configuration. This section resembles the snippet
below:
<interface type='network'>
[output truncated]
<model type='rtl8139' />
</interface>
4. Change the type attribute of the model element from 'rtl8139' to 'virtio'. This will
change the driver from the rtl8139 driver to the e1000 driver.
<interface type='network'>
[output truncated]
<model type='virtio' />
</interface>
5. Save the changes and exit the text editor
6. Restart the guest operating system.
C reat in g N ew G u est s U sin g O t h er N et wo rk D rivers
Alternatively, new guests can be created with a different network driver. This may be required if you
are having difficulty installing guests over a network connection. This method requires you to have at
least one guest already created (possibly installed from CD or D VD ) to use as a template.
1. Create an XML template from an existing guest (in this example, named Guest1):
# virsh dumpxml Guest1 > /tmp/guest-template.xml
2. Copy and edit the XML file and update the unique fields: virtual machine name, UUID , disk
image, MAC address, and any other unique parameters. Note that you can delete the UUID
and MAC address lines and virsh will generate a UUID and MAC address.

4 22

Chapt er 2 1 . T roubleshoot ing

# cp /tmp/guest-template.xml /tmp/new-guest.xml
# vi /tmp/new-guest.xml
Add the model line in the network interface section:
<interface type='network'>
[output truncated]
<model type='virtio' />
</interface>
3. Create the new virtual machine:
# virsh define /tmp/new-guest.xml
# virsh start new-guest

21.12. Workaround for Creat ing Ext ernal Snapshot s wit h libvirt
There are two classes of snapshots for QEMU guests. Internal snapshots are contained completely
within a qcow2 file, and fully supported by libvirt, allowing for creating, deleting, and reverting of
snapshots. This is the default setting used by libvirt when creating a snapshot, especially when no
option is specified. Although this file type takes a bit longer than others in creating the the snapshot,
it is required by libvirt to use qcow2 disks. Another drawback to this file type is that qcow2 disks are
not subject to receive improvements from QEMU.
External snapshots, on the other hand work with any type of original disk image, can be taken with
no guest downtime, and are able to receive active improvements from QEMU. In libvirt, they are created
when using the --d i sk-o nl y option to snapsho t-create-as (or when specifying an explicit XML
file to snapsho t-create that does the same). At the moment external snapshots are a one-way
operation as libvirt can create them but cannot do anything further with them.

21.13. Missing Charact ers on Guest Console wit h Japanese Keyboard

On a Red Hat Enterprise Linux 6 host, connecting a Japanese keyboard locally to a machine may
result in typed characters such as the underscore (the _ character) not being displayed correctly in
guest consoles. This occurs because the required keymap is not set correctly by default.
With Red Hat Enterprise Linux 3 and Red Hat Enterprise Linux 6 guests, there is usually no error
message produced when pressing the associated key. However, Red Hat Enterprise Linux 4 and Red
Hat Enterprise Linux 5 guests may display an error similar to the following:
atkdb.c: Unknown key pressed (translated set 2, code 0x0 on
isa0060/serio0).
atkbd.c: Use 'setkeycodes 00 <keycode>' to make it known.
To fix this issue in vi rt-manag er, perform the following steps:
Open the affected guest in vi rt-manag er.
Click View D et ails.
Select D i spl ay VNC in the list.
Change Au t o to ja in the Keymap pull-down menu.

4 23

Virt ualiz at ion Administ rat ion G uide

Click the Appl y button.

Alternatively, to fix this issue using the vi rsh ed i t command on the target guest:
Run vi rsh ed i t <targ et g uest>
Add the following attribute to the <graphics> tag: keymap = ' ja' . For example:
<graphics type='vnc' port='-1' autoport='yes' keymap= ' ja' />

21.14 . Verifying Virt ualiz at ion Ext ensions

Use this section to determine whether your system has the hardware virtualization extensions.
Virtualization extensions (Intel VT-x or AMD -V) are required for full virtualization.
1. Run the following command to verify the CPU virtualization extensions are available:
$ grep -E 'svm|vmx' /proc/cpuinfo
2. Analyze the output.
The following output contains a vmx entry indicating an Intel processor with the Intel VT-x
extension:
flags
: fpu tsc msr pae mce cx8 apic mtrr mca cmov pat pse36
clflush
dts acpi mmx fxsr sse sse2 ss ht tm syscall lm constant_tsc pni
monitor ds_cpl
vmx est tm2 cx16 xtpr lahf_lm
The following output contains an svm entry indicating an AMD processor with the AMD -V
extensions:
flags
: fpu tsc msr pae mce cx8 apic mtrr mca cmov pat pse36
clflush
mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt lm 3dnowext
3dnow pni cx16
lahf_lm cmp_legacy svm cr8legacy ts fid vid ttp tm stc
If any output is received, the processor has the hardware virtualization extensions. However
in some circumstances manufacturers disable the virtualization extensions in BIOS.
The " fl ag s: " output content may appear multiple times, once for each hyperthread, core or
CPU on the system.
The virtualization extensions may be disabled in the BIOS. If the extensions do not appear or
full virtualization does not work refer to Procedure 21.3, Enabling virtualization extensions
in BIOS .
3. En su re K VM su b syst em is lo ad ed
As an additional check, verify that the kvm modules are loaded in the kernel:
# lsmod | grep kvm

4 24

Chapt er 2 1 . T roubleshoot ing

If the output includes kvm_i ntel or kvm_amd then the kvm hardware virtualization modules
are loaded and your system meets requirements.

Note
If the libvirt package is installed, the vi rsh command can output a full list of virtualization
system capabilities. Run vi rsh capabi l i ti es as root to receive the complete list.

4 25

Virt ualiz at ion Administ rat ion G uide

Appendix A. The Virtual Host Metrics Daemon (vhostmd)

vh o st md (the Virtual Host Metrics D aemon) allows virtual machines to see limited information about
the host they are running on. This daemon is only supplied with Red Hat Enterprise Linux for SAP.
In the host, a daemon (vh o st md ) runs which writes metrics periodically into a disk image. This disk
image is exported read-only to guest virtual machines. Guest virtual machines can read the disk
image to see metrics. Simple synchronization stops guest virtual machines from seeing out of date or
corrupt metrics.
The system administrator chooses which metrics are available for use on a per guest virtual machine
basis. In addition, the system administrator may block one or more guest virtual machines from
having any access to metric configurations.
Customers who want to use vh o st md and vm- d u mp - met rics therefore need subscriptions for
" RHEL for SAP Business Applications" to be able to subscribe their RHEL systems running SAP to
the " RHEL for SAP" channel on the Customer Portal or Red Hat Subscription Management to install
the packages. The following kbase article in the customer portal describes the the setup of vhostmd
on RHEL: https://access.redhat.com/knowledge/solutions/41566

4 26

Appendix B. Addit ional Resources

Appendix B. Additional Resources

To learn more about virtualization and Red Hat Enterprise Linux, refer to the following resources.

B.1. Online Resources

http://www.libvirt.org/ is the official website for the l i bvi rt virtualization API.
http://virt-manager.et.redhat.com/ is the project website for the Virt u al Mach in e Man ag er (virtmanager), the graphical application for managing virtual machines.
Open Virtualization Center
http://www.openvirtualization.com
Red Hat D ocumentation
http://www.redhat.com/docs/
Virtualization technologies overview
http://virt.kernelnewbies.org
Red Hat Emerging Technologies group
http://et.redhat.com

B.2. Inst alled Document at ion

man vi rsh and /usr/share/d o c/l i bvi rt-<version-number> Contains sub commands
and options for the vi rsh virtual machine management utility as well as comprehensive
information about the l i bvi rt virtualization library API.
/usr/share/d o c/g no me-appl et-vm-<version-number> D ocumentation for the GNOME
graphical panel applet that monitors and manages locally-running virtual machines.
/usr/share/d o c/l i bvi rt-pytho n-<version-number> Provides details on the Python
bindings for the l i bvi rt library. The l i bvi rt-pytho n package allows python developers to
create programs that interface with the l i bvi rt virtualization management library.
/usr/share/d o c/pytho n-vi rti nst-<version-number> Provides documentation on the
vi rt-i nstal l command that helps in starting installations of Fedora and Red Hat Enterprise
Linux related distributions inside of virtual machines.
/usr/share/d o c/vi rt-manag er-<version-number> Provides documentation on the
Virtual Machine Manager, which provides a graphical tool for administering virtual machines.

4 27

Virt ualiz at ion Administ rat ion G uide

Appendix C. Revision History

R evisio n 1- 503
Several post-release updates

T u e Au g 9 2016

Jiri H errman n

R evisio n 1- 501
Mo n May 02 2016
Updates for the 6.8 GA release

Jiri H errman n

R evisio n 1- 500
T h u Mar 01 2016
Multiple updates for the 6.8 beta release

Jiri H errman n

R evisio n 1- 4 4 9
T h u O ct 08 2015
Cleaned up the Revision History

Jiri H errman n

R evisio n 1- 4 4 7
Fri Ju l 10 2015
Updates for the 6.7 GA release.

D ayle Parker

4 28