[

Lev
el-I
nte
rme
dia
te]

Lin
uxS e
rver
Ad
minis
tra
tio
n
S
tud
entMa
ter
ial
(
Vol
ume2-No
tesa
ndWo
rkb
ook
)

L
INUX
h
ttp
://
mic
rol
in
k.e
du.
et/
Table of Contents
1. Process management ......................................................................................... 1
1.1. terminology .............................................................................................. 1
1.2. basic process management ....................................................................... 2
1.3. signalling processes ................................................................................. 6
1.4. practice : basic process management ....................................................... 9
1.5. solution : basic process management ..................................................... 10
1.6. priority and nice values ......................................................................... 12
1.7. practice : process priorities .................................................................... 15
1.8. solution : process priorities .................................................................... 16
1.9. background processes ............................................................................ 18
1.10. practice : background processes ........................................................... 20
1.11. solution : background processes ........................................................... 21
2. Disk management ............................................................................................ 23
2.1. hard disk devices ................................................................................... 24
2.2. practice: hard disk devices ..................................................................... 31
2.3. solution: hard disk devices .................................................................... 32
2.4. partitions ................................................................................................. 34
2.5. practice: partitions .................................................................................. 39
2.6. solution: partitions ................................................................................. 40
2.7. file systems ............................................................................................ 41
2.8. practice: file systems .............................................................................. 45
2.9. solution: file systems ............................................................................. 46
2.10. mounting .............................................................................................. 47
2.11. practice: mounting file systems ........................................................... 52
2.12. solution: mounting file systems ........................................................... 53
2.13. uuid and filesystems ............................................................................ 55
2.14. practice: uuid and filesystems .............................................................. 57
2.15. solution: uuid and filesystems .............................................................. 58
2.16. RAID .................................................................................................... 59
2.17. practice: RAID ..................................................................................... 65
3. Logical volume management ......................................................................... 66
3.1. introduction to lvm ................................................................................ 67
3.2. lvm terminology ..................................................................................... 68
3.3. example: using lvm ................................................................................ 69
3.4. example: extend a logical volume ......................................................... 71
3.5. example: resize a physical Volume ....................................................... 73
3.6. example: mirror a logical volume .......................................................... 75
3.7. example: snapshot a logical volume ...................................................... 76
3.8. verifying existing physical volumes ...................................................... 77
3.9. verifying existing volume groups .......................................................... 79
3.10. verifying existing logical volumes ....................................................... 81
3.11. manage physical volumes .................................................................... 82
3.12. manage volume groups ........................................................................ 84
3.13. manage logical volumes ....................................................................... 86
3.14. practice : lvm ....................................................................................... 89
4. Booting the system .......................................................................................... 90

iii
 Linux System Administration

4.1. boot terminology .................................................................................... 90

4.2. grub ........................................................................................................ 93
4.3. lilo .......................................................................................................... 98
4.4. practice : bootloader ............................................................................... 99
4.5. solution : bootloader ............................................................................. 100
5. init ................................................................................................................... 101
5.1. about sysv init ...................................................................................... 101
5.2. system init(ialization) ........................................................................... 101
5.3. daemon or demon ? .............................................................................. 106
5.4. starting and stopping daemons ............................................................. 106
5.5. chkconfig .............................................................................................. 106
5.6. update-rc.d ............................................................................................ 108
5.7. bum ....................................................................................................... 109
5.8. runlevels ............................................................................................... 110
5.9. practice: init ......................................................................................... 113
5.10. solution : init ...................................................................................... 114
6. Linux Kernel ................................................................................................. 116
6.1. about the Linux kernel ......................................................................... 116
6.2. Linux kernel source ............................................................................. 118
6.3. kernel boot files ................................................................................... 122
6.4. Linux kernel modules .......................................................................... 123
6.5. compiling a kernel ............................................................................... 127
6.6. compiling one module ......................................................................... 130
7. Introduction to network sniffing ................................................................. 133
7.1. about sniffing ....................................................................................... 133
7.2. wireshark .............................................................................................. 133
7.3. tcpdump ................................................................................................ 135
7.4. practice: network sniffing .................................................................... 136
7.5. solution: network sniffing .................................................................... 137
8. Introduction to networking .......................................................................... 138
8.1. introduction to computer networks ...................................................... 139
8.2. about tcp/ip .......................................................................................... 144
8.3. practice : about tcp/ip ........................................................................... 146
8.4. solution : about tcp/ip .......................................................................... 147
8.5. using tcp/ip ........................................................................................... 148
8.6. practice : using tcp/ip ........................................................................... 154
8.7. solution : using tcp/ip ........................................................................... 155
8.8. multiple ip-addresses ............................................................................ 157
8.9. practice : multiple ip-addresses ............................................................ 157
8.10. solution : multiple ip-addresses .......................................................... 158
8.11. multihomed hosts ............................................................................... 159
8.12. practice : multihomed hosts ............................................................... 161
8.13. solution : multihomed hosts ............................................................... 162
8.14. introduction to iptables ...................................................................... 164
8.15. practice : iptables ............................................................................... 165
8.16. solution : iptables ............................................................................... 166
8.17. xinetd and inetd ................................................................................. 167
8.18. practice : inetd and xinetd .................................................................. 169

iv
 Linux System Administration

8.19. openssh ............................................................................................... 170

8.20. practice: ssh ........................................................................................ 174
8.21. network file system ............................................................................ 175
8.22. practice : network file system ............................................................ 177
9. Scheduling ...................................................................................................... 178
9.1. about scheduling .................................................................................. 178
9.2. one time jobs with at ........................................................................... 178
9.3. cron ....................................................................................................... 180
9.4. practice : scheduling ............................................................................. 182
9.5. solution : scheduling ............................................................................ 183
10. Logging ......................................................................................................... 184
10.1. about logging ..................................................................................... 184
10.2. login logging ...................................................................................... 185
10.3. syslogd daemon .................................................................................. 187
10.4. logger ................................................................................................. 189
10.5. watching logs ..................................................................................... 190
10.6. rotating logs ....................................................................................... 190
10.7. practice : logging ................................................................................ 191
10.8. solution : logging ............................................................................... 192
11. Library Management .................................................................................. 194
11.1. introduction ........................................................................................ 194
11.2. /lib and /usr/lib ................................................................................... 194
11.3. ldd ....................................................................................................... 194
11.4. ltrace ................................................................................................... 195
11.5. dpkg -S and debsums ......................................................................... 195
11.6. rpm -qf and rpm -V ........................................................................... 196
12. Memory management ................................................................................. 197
12.1. about memory .................................................................................... 197
12.2. /proc/meminfo .................................................................................... 197
12.3. swap space ......................................................................................... 198
12.4. practice : memory ............................................................................... 199
13. Installing Linux ........................................................................................... 200
13.1. about ................................................................................................... 200
13.2. installation by cdrom ......................................................................... 200
13.3. installation with rarp and tftp ............................................................. 200
13.4. about Red Hat kickstart ..................................................................... 201
13.5. using kickstart .................................................................................... 202
14. Package management ................................................................................. 203
14.1. terminology ........................................................................................ 203
14.2. Red Hat package manager ................................................................. 204
14.3. yum ..................................................................................................... 205
14.4. rpm2cpio ............................................................................................ 210
14.5. Debian package management ............................................................. 210
14.6. alien .................................................................................................... 212
14.7. Downloading software ....................................................................... 213
14.8. Compiling software ............................................................................ 213
14.9. Practice: Installing software ............................................................... 213
14.10. Solution: Installing software ............................................................ 214

v
 Linux System Administration

15. Backup .......................................................................................................... 215

15.1. About tape devices ............................................................................. 215
15.2. Compression ....................................................................................... 216
15.3. tar ....................................................................................................... 217
15.4. Backup Types ..................................................................................... 219
15.5. dump and restore ............................................................................... 219
15.6. cpio ..................................................................................................... 220
15.7. dd ........................................................................................................ 220
15.8. split ..................................................................................................... 222
15.9. Practice backup .................................................................................. 222
16. Performance monitoring ............................................................................ 224
16.1. About Monitoring .............................................................................. 224
16.2. top ....................................................................................................... 224
16.3. free ..................................................................................................... 225
16.4. watch .................................................................................................. 225
16.5. vmstat ................................................................................................. 225
16.6. iostat ................................................................................................... 226
16.7. mpstat ................................................................................................. 227
16.8. sadc and sar ........................................................................................ 227
16.9. ntop ..................................................................................................... 228
16.10. iftop .................................................................................................. 228

vi
List of Tables
2.1. ide device naming .......................................................................................... 25
2.2. scsi device naming ......................................................................................... 26
2.3. primary, extended and logical partitions ........................................................ 34
2.4. Partition naming ............................................................................................. 34
3.1. disk partitioning example ............................................................................... 67
3.2. LVM Example ............................................................................................... 67

vii
Chapter 1. Process management

Table of Contents
1.1. terminology ...................................................................................................... 1
1.2. basic process management ............................................................................... 2
1.3. signalling processes ......................................................................................... 6
1.4. practice : basic process management ............................................................... 9
1.5. solution : basic process management ............................................................. 10
1.6. priority and nice values ................................................................................. 12
1.7. practice : process priorities ............................................................................ 15
1.8. solution : process priorities ............................................................................ 16
1.9. background processes .................................................................................... 18
1.10. practice : background processes ................................................................... 20
1.11. solution : background processes .................................................................. 21

1.1. terminology

1.1.1. process
A process is compiled source code that is currently running on the system.

1.1.2. PID
All processes have a process id or PID.

1.1.3. PPID
Every process has a parent process (with a PPID). The child process is often started
by the parent process.

1.1.4. init
The init process always has process ID 1. The init process is started by the kernel
itself so technically it does not have a parent process. init serves as a foster parent
for orphaned processes.

1.1.5. kill
When a process stops running, the process dies, when you want a process to die, you
kill it.

1
 Process management

1.1.6. daemon
Processes that start at system startup and keep running forever are called daemon
processes or daemons. These daemons never die.

1.1.7. zombie
When a process is killed, but it still shows up on the system, then the process is
referred to as zombie. You cannot kill zombies, because they are already dead.

1.2. basic process management

1.2.1. $$ and $PPID

Some shell environment variables contain information about processes. The $$
variable will hold your current process ID, and $PPID contains the parent PID.
Actually $$ is a shell parameter and not a variable, you cannot assign a value to it.

Below we use echo to display the values of $$ and $PPID.

[paul@RHEL4b ~]$ echo $$ $PPID

4224 4223

1.2.2. pidof
You can find all process id's by name using the pidof command.

root@rhel53 ~# pidof mingetty

2819 2798 2797 2796 2795 2794

1.2.3. parent and child

Processes have a parent-child relationship. Every process has a parent process.

When starting a new bash you can use echo to verify that the pid from before is the
ppid of the new shell. The child process from above is now the parent process.

[paul@RHEL4b ~]$ bash

[paul@RHEL4b ~]$ echo $$ $PPID
4812 4224

Typing exit will end the current process and brings us back to our original values
for $$ and $PPID.

2
 Process management

[paul@RHEL4b ~]$ echo $$ $PPID

4812 4224
[paul@RHEL4b ~]$ exit
exit
[paul@RHEL4b ~]$ echo $$ $PPID
4224 4223
[paul@RHEL4b ~]$

1.2.4. fork and exec

A process starts another process in two phases. First the process creates a fork of
itself, an identical copy. Then the forked process executes an exec to replace the
forked process with the target child process.

[paul@RHEL4b ~]$ echo $$

4224
[paul@RHEL4b ~]$ bash
[paul@RHEL4b ~]$ echo $$ $PPID
5310 4224
[paul@RHEL4b ~]$

1.2.5. exec
With the exec command, you can execute a process without forking a new process.
In the following screenshot a Korn shell (ksh) is started and is being replaced with a
bash shell using the exec command. The pid of the bash shell is the same as the pid
of the Korn shell. Exiting the child bash shell will get me back to the parent bash,
not to the Korn shell (which does not exist anymore).

[paul@RHEL4b ~]$ echo $$

4224 # PID of bash
[paul@RHEL4b ~]$ ksh
$ echo $$ $PPID
5343 4224 # PID of ksh and bash
$ exec bash
[paul@RHEL4b ~]$ echo $$ $PPID
5343 4224 # PID of bash and bash
[paul@RHEL4b ~]$ exit
exit
[paul@RHEL4b ~]$ echo $$
4224

1.2.6. ps
One of the most common tools on Linux to look at processes is ps. The following
screenshot shows the parent child relationship between three bash processes.

[paul@RHEL4b ~]$ echo $$ $PPID

3
 Process management

4224 4223
[paul@RHEL4b ~]$ bash
[paul@RHEL4b ~]$ echo $$ $PPID
4866 4224
[paul@RHEL4b ~]$ bash
[paul@RHEL4b ~]$ echo $$ $PPID
4884 4866
[paul@RHEL4b ~]$ ps fx
PID TTY STAT TIME COMMAND
4223 ? S 0:01 sshd: paul@pts/0
4224 pts/0 Ss 0:00 \_ -bash
4866 pts/0 S 0:00 \_ bash
4884 pts/0 S 0:00 \_ bash
4902 pts/0 R+ 0:00 \_ ps fx
[paul@RHEL4b ~]$ exit
exit
[paul@RHEL4b ~]$ ps fx
PID TTY STAT TIME COMMAND
4223 ? S 0:01 sshd: paul@pts/0
4224 pts/0 Ss 0:00 \_ -bash
4866 pts/0 S 0:00 \_ bash
4903 pts/0 R+ 0:00 \_ ps fx
[paul@RHEL4b ~]$ exit
exit
[paul@RHEL4b ~]$ ps fx
PID TTY STAT TIME COMMAND
4223 ? S 0:01 sshd: paul@pts/0
4224 pts/0 Ss 0:00 \_ -bash
4904 pts/0 R+ 0:00 \_ ps fx
[paul@RHEL4b ~]$

On Linux, ps fax is often used. On Solaris ps -ef (which also works on Linux) is
common. Here is a partial output from ps fax.

[paul@RHEL4a ~]$ ps fax

PID TTY STAT TIME COMMAND
1 ? S 0:00 init [5]

...

3713 ? Ss 0:00 /usr/sbin/sshd

5042 ? Ss 0:00 \_ sshd: paul [priv]
5044 ? S 0:00 \_ sshd: paul@pts/1
5045 pts/1 Ss 0:00 \_ -bash
5077 pts/1 R+ 0:00 \_ ps fax

1.2.7. pgrep
Similar to the ps -C, you can also use pgrep to search for a process by its command
name.

[paul@RHEL5 ~]$ sleep 1000 &

[1] 32558
[paul@RHEL5 ~]$ pgrep sleep
32558
[paul@RHEL5 ~]$ ps -C sleep
PID TTY TIME CMD
32558 pts/3 00:00:00 sleep

4
 Process management

You can also list the command name of the process with pgrep.

paul@laika:~$ pgrep -l sleep

9661 sleep

1.2.8. top
Another popular tool on Linux is top. The top tool can order processes according to
cpu usage or other properties. You can also kill processes from within top. Press h
inside top for help.

In case of trouble, top is often the first tool to fire up, since it also provides you
memory and swap space information.

5
 Process management

1.3. signalling processes

1.3.1. kill
The kill command will kill (or stop) a process. The screenshot shows how to use a
standard kill to stop the process with pid 1942.

paul@ubuntu910:~$ kill 1942

paul@ubuntu910:~$

By using the kill we are sending a signal to the process.

1.3.2. list signals

Running processes can receive signals from each other or from the users. You can
have a list of signals by typing kill -l, that is a letter l, not the number 1.

[paul@RHEL4a ~]$ kill -l

1) SIGHUP 2) SIGINT 3) SIGQUIT 4) SIGILL
5) SIGTRAP 6) SIGABRT 7) SIGBUS 8) SIGFPE
9) SIGKILL 10) SIGUSR1 11) SIGSEGV 12) SIGUSR2
13) SIGPIPE 14) SIGALRM 15) SIGTERM 17) SIGCHLD
18) SIGCONT 19) SIGSTOP 20) SIGTSTP 21) SIGTTIN
22) SIGTTOU 23) SIGURG 24) SIGXCPU 25) SIGXFSZ
26) SIGVTALRM 27) SIGPROF 28) SIGWINCH 29) SIGIO
30) SIGPWR 31) SIGSYS 34) SIGRTMIN 35) SIGRTMIN+1
36) SIGRTMIN+2 37) SIGRTMIN+3 38) SIGRTMIN+4 39) SIGRTMIN+5
40) SIGRTMIN+6 41) SIGRTMIN+7 42) SIGRTMIN+8 43) SIGRTMIN+9
44) SIGRTMIN+10 45) SIGRTMIN+11 46) SIGRTMIN+12 47) SIGRTMIN+13
48) SIGRTMIN+14 49) SIGRTMIN+15 50) SIGRTMAX-14 51) SIGRTMAX-13
52) SIGRTMAX-12 53) SIGRTMAX-11 54) SIGRTMAX-10 55) SIGRTMAX-9
56) SIGRTMAX-8 57) SIGRTMAX-7 58) SIGRTMAX-6 59) SIGRTMAX-5
60) SIGRTMAX-4 61) SIGRTMAX-3 62) SIGRTMAX-2 63) SIGRTMAX-1
64) SIGRTMAX
[paul@RHEL4a ~]$

1.3.3. kill -1 (SIGHUP)

It is common on Linux to use the first signal SIGHUP (or HUP or 1) to tell a process
that it should re-read its configuration file. Thus, the kill -1 1 command forces the
init process (init always runs with pid 1) to re-read its configuration file.

root@deb503:~# kill -1 1
root@deb503:~#

It is up to the developer of the process to decide whether the process can do this
running, or whether it needs to stop and start. It is up to the user to read the
documentation of the program.

6
 Process management

1.3.4. kill -15 (SIGTERM)

The SIGTERM signal is also called a standard kill. Whenever kill is executed
without specifying the signal, a kill -15 is assumed.

Both commands in the screenshot below are identical.

paul@ubuntu910:~$ kill 1942

paul@ubuntu910:~$ kill -15 1942

1.3.5. kill -9 (SIGKILL)

The SIGKILL is different from most other signals in that it is not being sent to the
process, but to the Linux kernel. A kill -9 is also called a sure kill. The kernel will
shoot down the process. As a developer you have no means to intercept a kill -9 signal.

root@rhel53 ~# kill -9 3342

1.3.6. killall
The killall command will also default to sending a signal 15 to the processes.

This command and its SysV counterpart killall5 can by used when shutting down
the system. This screenshot shows how Red Hat Enterprise Linux 5.3 uses killall5
when halting the system.

root@rhel53 ~# grep killall /etc/init.d/halt

action $"Sending all processes the TERM signal..." /sbin/killall5 -15
action $"Sending all processes the KILL signal..." /sbin/killall5 -9

1.3.7. pkill
You can use the pkill command to kill a process by its command name.

[paul@RHEL5 ~]$ sleep 1000 &

[1] 30203
[paul@RHEL5 ~]$ pkill sleep
[1]+ Terminated sleep 1000
[paul@RHEL5 ~]$

1.3.8. top
Inside top the k key allows you to select a signal and pid to kill. Below is a partial
screenshot of the line just below the summary in top after pressing k.

7
 Process management

PID to kill: 1932

Kill PID 1932 with signal [15]: 9

1.3.9. SIGSTOP and SIGCONT

A running process can be suspended when it receives a SIGSTOP signal. This is the
same as kill -19 on Linux, but might have a different number in other Unix systems.

A suspended process does not use any cpu cycles, but it stays in memory and can be
re-animated with a SIGCONT signal (kill -18 on Linux).

Both signals will be used in the section about background processes.

8
 Process management

1.4. practice : basic process management

1. Use ps to search for the init process by name.

2. What is the process id of the init process ?

3. Use the who am i command to determine your terminal name.

4. Using your terminal name from above, use ps to find all processes associated with
your terminal.

5. What is the process id of your shell ?

6. What is the parent process id of your shell ?

7. Start two instances of the sleep 3342 in background.

8. Locate the process id of all sleep commands.

9. Display only those two sleep processes in top. Then quit top.

10. Use a standard kill to kill one of the sleep processes.

11. Use one command to kill all sleep processes.

9
 Process management

1.5. solution : basic process management

1. Use ps to search for the init process by name.

root@rhel53 ~# ps -C init
PID TTY TIME CMD
1 ? 00:00:04 init

2. What is the process id of the init process ?

1

3. Use the who am i command to determine your terminal name.

root@rhel53 ~# who am i
paul pts/0 2010-04-12 17:44 (192.168.1.38)

4. Using your terminal name from above, use ps to find all processes associated with
your terminal.

oot@rhel53 ~# ps fax | grep pts/0

2941 ? S 0:00 \_ sshd: paul@pts/0
2942 pts/0 Ss 0:00 \_ -bash
2972 pts/0 S 0:00 \_ su -
2973 pts/0 S 0:00 \_ -bash
3808 pts/0 R+ 0:00 \_ ps fax
3809 pts/0 R+ 0:00 \_ grep pts/0

or also

root@rhel53 ~# ps -ef | grep pts/0

paul 2941 2939 0 17:44 ? 00:00:00 sshd: paul@pts/0
paul 2942 2941 0 17:44 pts/0 00:00:00 -bash
root 2972 2942 0 17:45 pts/0 00:00:00 su -
root 2973 2972 0 17:45 pts/0 00:00:00 -bash
root 3816 2973 0 21:25 pts/0 00:00:00 ps -ef
root 3817 2973 0 21:25 pts/0 00:00:00 grep pts/0

5. What is the process id of your shell ?

2973 in the screenshot above, probably different for you

echo $$ should display same number as the one you found

6. What is the parent process id of your shell ?

2972 in the screenshot above, probably different for you

in this example the PPID is from the su - command, but when inside gnome then for
example gnome-terminal can be the parent process

7. Start two instances of the sleep 3342 in background.

10
 Process management

sleep 3342 &

sleep 3342 &

8. Locate the process id of all sleep commands.

pidof sleep

9. Display only those two sleep processes in top. Then quit top.
top -p pidx,pidy (replace pidx pidy with the actual numbers)

10. Use a standard kill to kill one of the sleep processes.

kill pidx

11. Use one command to kill all sleep processes.

pkill sleep

11
 Process management

1.6. priority and nice values

1.6.1. introduction
All processes have a priority and a nice value. Higher priority processes will get
more cpu time than lower priority processes. You can influence this with the nice
and renice commands.

1.6.2. pipes (mkfifo)

Processes can communicate with each other via pipes. These pipes can be created
with the mkfifo command.

The screenshots shows the creation of four distinct pipes (in a new directory).

paul@ubuntu910:~$ mkdir procs

paul@ubuntu910:~$ cd procs/
paul@ubuntu910:~/procs$ mkfifo pipe33a pipe33b pipe42a pipe42b
paul@ubuntu910:~/procs$ ls -l
total 0
prw-r--r-- 1 paul paul 0 2010-04-12 13:21 pipe33a
prw-r--r-- 1 paul paul 0 2010-04-12 13:21 pipe33b
prw-r--r-- 1 paul paul 0 2010-04-12 13:21 pipe42a
prw-r--r-- 1 paul paul 0 2010-04-12 13:21 pipe42b
paul@ubuntu910:~/procs$

1.6.3. some fun with cat

To demonstrate the use of the top and renice commands we will make the cat
command use the previously created pipes to generate a full load on the cpu.

The cat is copied with a distinct name to the current directory. (This enables us to
easily recognize the processes within top. You could do the same exercise without
copying the cat command, but using different users. Or you could just look at the pid
of each process.)

paul@ubuntu910:~/procs$ cp /bin/cat proj33

paul@ubuntu910:~/procs$ cp /bin/cat proj42
paul@ubuntu910:~/procs$ echo -n x | ./proj33 - pipe33a > pipe33b &
[1] 1670
paul@ubuntu910:~/procs$ ./proj33 <pipe33b >pipe33a &
[2] 1671
paul@ubuntu910:~/procs$ echo -n z | ./proj42 - pipe42a > pipe42b &
[3] 1673
paul@ubuntu910:~/procs$ ./proj42 <pipe42b >pipe42a &
[4] 1674

The commands you see above will create two proj33 processes that use cat to bounce
the x character between pipe33a and pipe33b. And ditto for the z character and
proj42.

12
 Process management

1.6.4. top
Just running top without options or arguments will display all processes and an
overview of innformation. The top of the top screen might look something like this.

top - 13:59:29 up 48 min, 4 users, load average: 1.06, 0.25, 0.14

Tasks: 139 total, 3 running, 136 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.3%us, 99.7%sy, 0.0%ni, 0.0%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 509352k total, 460040k used, 49312k free, 66752k buffers
Swap: 746980k total, 0k used, 746980k free, 247324k cached

Notice the cpu idle time (0.0%id) is zero. This is because our cat processes are
consuming the whole cpu. Results can vary on systems with four or more cpu cores.

1.6.5. top -p
The top -p 1670,1671,1673,1674 screenshot below shows four processes, all of then
using approximately 25 percent of the cpu.

paul@ubuntu910:~$ top -p 1670,1671,1673,1674

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND

1674 paul 20 0 2972 616 524 S 26.6 0.1 0:11.92 proj42
1670 paul 20 0 2972 616 524 R 25.0 0.1 0:23.16 proj33
1671 paul 20 0 2972 616 524 S 24.6 0.1 0:23.07 proj33
1673 paul 20 0 2972 620 524 R 23.0 0.1 0:11.48 proj42

All four processes have an equal priority (PR), and are battling for cpu time. On
some systems the Linux kernel might attribute slightly varying priority values, but
the result will still be four processes fighting for cpu time.

1.6.6. renice
Since the processes are already running, we need to use the renice command to
change their nice value (NI).

The screenshot shows how to use renice on both the proj33 processes.

paul@ubuntu910:~$ renice +8 1670

1670: old priority 0, new priority 8
paul@ubuntu910:~$ renice +8 1671
1671: old priority 0, new priority 8

Normal users can attribute a nice value from zero to 20 to processes they own. Only
the root user can use negative nice values. Be very careful with negative nice values,
since they can make it impossible to use the keyboard or ssh to a system.

13
 Process management

1.6.7. impact of nice values

The impact of a nice value on running processes can vary. The screenshot below
shows the result of our renice +8 command. Look at the %CPU values.

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND

1674 paul 20 0 2972 616 524 S 46.6 0.1 0:22.37 proj42
1673 paul 20 0 2972 620 524 R 42.6 0.1 0:21.65 proj42
1671 paul 28 8 2972 616 524 S 5.7 0.1 0:29.65 proj33
1670 paul 28 8 2972 616 524 R 4.7 0.1 0:29.82 proj33

Important to remember is to always make less important processes nice to more

important processes. Using negative nice values can have a serere impact on a
system's usability.

1.6.8. nice
The nice works identical to the renice but it is used when starting a command.

The screenshot shows how to start a script with a nice value of five.

paul@ubuntu910:~$ nice -5 ./backup.sh

14
 Process management

1.7. practice : process priorities

1. Create a new directory and create six pipes in that directory.

2. Bounce a character between two pipes.

3. Use top and ps to display information (pid, ppid, priority, nice value, ...) about
these two cat processes.

4. Bounce another character between two other pipes, but this time start the
commands nice. Verify that all cat processes are battling for the cpu. (Feel free to
fire up two more cats with the remaining pipes).

5. Use ps to verify that the two new cat processes have a nice value. Use the -o and
-C options of ps for this.

6. Use renice te increase the nice value from 10 to 15. Notice the difference with
the usual commands.

15
 Process management

1.8. solution : process priorities

1. Create a new directory and create six pipes in that directory.
[paul@rhel53 ~]$ mkdir pipes ; cd pipes
[paul@rhel53 pipes]$ mkfifo p1 p2 p3 p4 p5 p6
[paul@rhel53 pipes]$ ls -l
total 0
prw-rw-r-- 1 paul paul 0 Apr 12 22:15 p1
prw-rw-r-- 1 paul paul 0 Apr 12 22:15 p2
prw-rw-r-- 1 paul paul 0 Apr 12 22:15 p3
prw-rw-r-- 1 paul paul 0 Apr 12 22:15 p4
prw-rw-r-- 1 paul paul 0 Apr 12 22:15 p5
prw-rw-r-- 1 paul paul 0 Apr 12 22:15 p6

2. Bounce a character between two pipes.

[paul@rhel53 pipes]$ echo -n x | cat - p1 > p2 &
[1] 4013
[paul@rhel53 pipes]$ cat <p2 >p1 &
[2] 4016

3. Use top and ps to display information (pid, ppid, priority, nice value, ...) about
these two cat processes.
top (probably the top two lines)

[paul@rhel53 pipes]$ ps -C cat

PID TTY TIME CMD
4013 pts/0 00:03:38 cat
4016 pts/0 00:01:07 cat

[paul@rhel53 pipes]$ ps fax | grep cat

4013 pts/0 R 4:00 | \_ cat - p1
4016 pts/0 S 1:13 | \_ cat
4044 pts/0 S+ 0:00 | \_ grep cat

4. Bounce another character between two other pipes, but this time start the
commands nice. Verify that all cat processes are battling for the cpu. (Feel free to
fire up two more cats with the remaining pipes).
echo -n y | nice cat - p3 > p4 &
nice cat <p4 >p3 &

5. Use ps to verify that the two new cat processes have a nice value. Use the -o and
-C options of ps for this.
[paul@rhel53 pipes]$ ps -C cat -o pid,ppid,pri,ni,comm
PID PPID PRI NI COMMAND
4013 3947 14 0 cat
4016 3947 21 0 cat
4025 3947 13 10 cat
4026 3947 13 10 cat

6. Use renice te increase the nice value from 10 to 15. Notice the difference with
the usual commands.
[paul@rhel53 pipes]$ renice +15 4025
4025: old priority 10, new priority 15
[paul@rhel53 pipes]$ renice +15 4026

16
 Process management

4026: old priority 10, new priority 15

[paul@rhel53 pipes]$ ps -C cat -o pid,ppid,pri,ni,comm

PID PPID PRI NI COMMAND
4013 3947 14 0 cat
4016 3947 21 0 cat
4025 3947 9 15 cat
4026 3947 8 15 cat

17
 Process management

1.9. background processes

1.9.1. jobs
Stuff that runs in background of your current shell can be displayed with the jobs
command. By default you will not have any jobs running in background.

root@rhel53 ~# jobs
root@rhel53 ~#

This jobs command will be used several times in this section.

1.9.2. control-Z
Some processes can be suspended with the Ctrl-Z key combination. This sends
a SIGSTOP signal to the Linux kernel, effectively freezing the operation of the
process.

When doing this in vi(m), then vi(m) goes to the background. The background vi(m)
can be seen with the jobs command.

[paul@RHEL4a ~]$ vi procdemo.txt

[5]+ Stopped vim procdemo.txt

[paul@RHEL4a ~]$ jobs
[5]+ Stopped vim procdemo.txt

1.9.3. & ampersand

Processes that are started in background using the & character at the end of the
command line are also visible with the jobs command.

[paul@RHEL4a ~]$ find / > allfiles.txt 2> /dev/null &

[6] 5230
[paul@RHEL4a ~]$ jobs
[5]+ Stopped vim procdemo.txt
[6]- Running find / >allfiles.txt 2>/dev/null &
[paul@RHEL4a ~]$

1.9.4. jobs -p
An interesting option is jobs -p to see the process id of background processes.

[paul@RHEL4b ~]$ sleep 500 &

18
 Process management

[1] 4902
[paul@RHEL4b ~]$ sleep 400 &
[2] 4903
[paul@RHEL4b ~]$ jobs -p
4902
4903
[paul@RHEL4b ~]$ ps `jobs -p`
PID TTY STAT TIME COMMAND
4902 pts/0 S 0:00 sleep 500
4903 pts/0 S 0:00 sleep 400
[paul@RHEL4b ~]$

1.9.5. fg
Running the fg command will bring a background job to the foreground. The number
of the background job to bring forward is the parameter of fg.

[paul@RHEL5 ~]$ jobs

[1] Running sleep 1000 &
[2]- Running sleep 1000 &
[3]+ Running sleep 2000 &
[paul@RHEL5 ~]$ fg 3
sleep 2000

1.9.6. bg
Jobs that are suspended in background can be started in background with bg. The
bg will send a SIGCONT signal.

Below an example of the sleep command (suspended with Ctrl-Z) being reactivated
in background with bg.

[paul@RHEL5 ~]$ jobs

[paul@RHEL5 ~]$ sleep 5000 &
[1] 6702
[paul@RHEL5 ~]$ sleep 3000

[2]+ Stopped sleep 3000

[paul@RHEL5 ~]$ jobs
[1]- Running sleep 5000 &
[2]+ Stopped sleep 3000
[paul@RHEL5 ~]$ bg 2
[2]+ sleep 3000 &
[paul@RHEL5 ~]$ jobs
[1]- Running sleep 5000 &
[2]+ Running sleep 3000 &
[paul@RHEL5 ~]$

19
 Process management

1.10. practice : background processes

1. Use the jobs command to verify whether you have any processes running in
background.

2. Use vi to create a little text file. Suspend vi in background.

3. Verify with jobs that vi is suspended in background.

4. Start find / > allfiles.txt 2>/dev/null in foreground. Suspend it in background

before it finishes.

5. Start two long sleep processes in background.

6. Display all jobs in background.

7. Use the kill command to suspend the last sleep process.

8. Continue the find process in background (make sure it runs again).

9. Put one of the sleep commands back in foreground.

10. (if time permits, a general review question...) Explain in detail where the numbers
come from in the next screenshot. When are the variables replaced by their value ?
By which shell ?
[paul@RHEL4b ~]$ echo $$ $PPID
4224 4223
[paul@RHEL4b ~]$ bash -c "echo $$ $PPID"
4224 4223
[paul@RHEL4b ~]$ bash -c 'echo $$ $PPID'
5059 4224
[paul@RHEL4b ~]$ bash -c `echo $$ $PPID`
4223: 4224: command not found

20
 Process management

1.11. solution : background processes

1. Use the jobs command to verify whether you have any processes running in
background.
jobs (maybe the catfun is still running?)

2. Use vi to create a little text file. Suspend vi in background.

vi text.txt
(inside vi press ctrl-z)

3. Verify with jobs that vi is suspended in background.

[paul@rhel53 ~]$ jobs
[1]+ Stopped vim text.txt

4. Start find / > allfiles.txt 2>/dev/null in foreground. Suspend it in background

before it finishes.

[paul@rhel53 ~]$ find / > allfiles.txt 2>/dev/null

(press ctrl-z)
[2]+ Stopped find / > allfiles.txt 2> /dev/null

5. Start two long sleep processes in background.

sleep 4000 & ; sleep 5000 &

6. Display all jobs in background.

[paul@rhel53 ~]$ jobs
[1]- Stopped vim text.txt
[2]+ Stopped find / > allfiles.txt 2> /dev/null
[3] Running sleep 4000 &
[4] Running sleep 5000 &

7. Use the kill command to suspend the last sleep process.

[paul@rhel53 ~]$ kill -SIGSTOP 4519
[paul@rhel53 ~]$ jobs
[1] Stopped vim text.txt
[2]- Stopped find / > allfiles.txt 2> /dev/null
[3] Running sleep 4000 &
[4]+ Stopped sleep 5000

8. Continue the find process in background (make sure it runs again).

bg 2 (verify the job-id in your jobs list)

9. Put one of the sleep commands back in foreground.

fg 3 (again verify your job-id)

10. (if time permits, a general review question...) Explain in detail where the numbers
come from in the next screenshot. When are the variables replaced by their value ?
By which shell ?

21
 Process management

[paul@RHEL4b ~]$ echo $$ $PPID

4224 4223
[paul@RHEL4b ~]$ bash -c "echo $$ $PPID"
4224 4223
[paul@RHEL4b ~]$ bash -c 'echo $$ $PPID'
5059 4224
[paul@RHEL4b ~]$ bash -c `echo $$ $PPID`
4223: 4224: command not found

The current bash shell will replace the $$ and $PPID while scanning the line, and
before executing the echo command.

[paul@RHEL4b ~]$ echo $$ $PPID

4224 4223

The variables are now double quoted, but the current bash shell will replace $$ and
$PPID while scanning the line, and before executing the bach -c command.

[paul@RHEL4b ~]$ bash -c "echo $$ $PPID"

4224 4223

The variables are now single quoted. The current bash shell will not replace the $$
and the $PPID. The bash -c command will be executed before the variables replaced
with their value. This latter bash is the one replacing the $$ and $PPID with their
value.

[paul@RHEL4b ~]$ bash -c 'echo $$ $PPID'

5059 4224

With backticks the shell will still replace both variable before the embedded echo is
executed. The result of this echo is the two process id's. These are given as commands
to bash -c. But two numbers are not commands!

[paul@RHEL4b ~]$ bash -c `echo $$ $PPID`

4223: 4224: command not found

22
Chapter 2. Disk management

Table of Contents
2.1. hard disk devices ........................................................................................... 24
2.2. practice: hard disk devices ............................................................................. 31
2.3. solution: hard disk devices ............................................................................ 32
2.4. partitions ......................................................................................................... 34
2.5. practice: partitions .......................................................................................... 39
2.6. solution: partitions ......................................................................................... 40
2.7. file systems .................................................................................................... 41
2.8. practice: file systems ..................................................................................... 45
2.9. solution: file systems ..................................................................................... 46
2.10. mounting ...................................................................................................... 47
2.11. practice: mounting file systems ................................................................... 52
2.12. solution: mounting file systems ................................................................... 53
2.13. uuid and filesystems .................................................................................... 55
2.14. practice: uuid and filesystems ...................................................................... 57
2.15. solution: uuid and filesystems ..................................................................... 58
2.16. RAID ............................................................................................................ 59
2.17. practice: RAID ............................................................................................. 65

23
 Disk management

2.1. hard disk devices

2.1.1. terminology

platter, head, track, cylinder, sector

Data is commonly stored on magnetic or optical disk platters. The platters are rotated
(at high speeds). Data is read by heads, which are very close to the surface of the
platter, without touching it! The heads are mounted on an arm (sometimes called a
comb or a fork).

Data is written in concentric circles called tracks. Track zero is (usually) on the
inside. The time it takes to position the head over a certain track is called the seek
time. Often the platters are stacked on top of each other, hence the set of tracks
accessible at a certain position of the comb forms a cylinder. Tracks are divided into
512 byte sectors, with more unused space (gap) between the sectors on the outside
of the platter.

When you break down the advertised access time of a hard drive, you will notice
that most of that time is taken by movement of the heads (about 65%) and rotational
latency (about 30%).

block device
Random access hard disk devices have an abstraction layer called block device to
enable formatting in fixed-size (usually 512 bytes) blocks. Blocks can be accessed
independent of access to other blocks. A block device has the letter b to denote the
file type in the output of ls -l.

[root@RHEL4b ~]# ls -l /dev/sda*

brw-rw---- 1 root disk 8, 0 Aug 4 22:55 /dev/sda
brw-rw---- 1 root disk 8, 1 Aug 4 22:55 /dev/sda1
brw-rw---- 1 root disk 8, 2 Aug 4 22:55 /dev/sda2
[root@RHEL4b ~]#

Note that a character device is a constant stream of characters, being denoted by a

c in ls -l.

Note also that the ISO 9660 standard for cdrom uses a 2048 byte block size.

Old hard disks (and floppy disks) use cylinder-head-sector addressing to access a
sector on the disk. Most current disks use LBA (Logical Block Addressing).

ide or scsi
Actually, the title should be ata or scsi, since ide is an ata compatible device. Most
desktops use ata devices, most servers use scsi.

24
 Disk management

ata
An ata controller allows two devices per bus, one master and one slave. Unless
your controller and devices support cable select, you have to set this manually with
jumpers.

With the introduction of sata (serial ata), the original ata was renamed to parallel
ata. Optical drives often use atapi, which is an ATA interface using the SCSI
communication protocol.

scsi
A scsi controller allows more than two devices. When using SCSI (small computer
system interface), each device gets a unique scsi id. The scsi controller also needs
a scsi id, do not use this id for a scsi-attached device.

Older 8-bit SCSI is now called narrow, whereas 16-bit is wide. When the bus
speeds was doubled to 10Mhz, this was known as fast SCSI. Doubling to 20Mhz
made it ultra SCSI. Take a look at http://en.wikipedia.org/wiki/SCSI for more SCSI
standards.

2.1.2. device naming

ata (ide) device naming

All ata drives on your system will start with /dev/hd followed by a unit letter. The
master hdd on the first ata controller is /dev/hda, the slave is /dev/hdb. For the second
controller, the names of the devices are /dev/hdc and /dev/hdd.

Table 2.1. ide device naming

controller connection device name
master /dev/hda
ide0
slave /dev/hdb
master /dev/hdc
ide1
slave /dev/hdd

It is possible to have only /dev/hda and /dev/hdd. The first one is a single ata hard
disk, the second one is the cdrom (by default configured as slave).

scsi device naming

scsi drives follow a similar scheme, but all start with /dev/sd. When you run out of
letters (after /dev/sdz), you can continue with /dev/sdaa and /dev/sdab and so on. (We
will see later on that lvm volumes are commonly seen as /dev/md0, /dev/md1 etc.)

25
 Disk management

Below a sample of how scsi devices on a linux can be named. Adding a scsi disk or
raid controller with a lower scsi address will change the naming scheme (shifting the
higher scsi addresses one letter further in the alphabet).

Table 2.2. scsi device naming

device scsi id device name

disk 0 0 /dev/sda
disk 1 1 /dev/sdb
raid controller 0 5 /dev/sdc
raid controller 1 6 /dev/sdd

2.1.3. discovering all disk devices

/sbin/fdisk
You can start by using /sbin/fdisk to find out what kind of disks are seen by the
kernel. Below the result on Debian, with two ata-ide disks present.

root@barry:~# fdisk -l | grep Disk

Disk /dev/hda: 60.0 GB, 60022480896 bytes
Disk /dev/hdb: 81.9 GB, 81964302336 bytes

And here an example of sata disks on a laptop with Ubuntu. Remember that sata
disks are presented to you with the scsi /dev/sdx notation.

root@laika:~# fdisk -l | grep Disk

Disk /dev/sda: 100.0 GB, 100030242816 bytes
Disk /dev/sdb: 100.0 GB, 100030242816 bytes

Here is an overview of disks on a RHEL4u3 server with two real 72GB scsi disks.
This server is attached to a NAS with four NAS disks of half a terabyte. On the NAS
disks, four LVM (/dev/mdx) software RAID devices are configured.

[root@tsvtl1 ~]# fdisk -l | grep Disk

Disk /dev/sda: 73.4 GB, 73407488000 bytes
Disk /dev/sdb: 73.4 GB, 73407488000 bytes
Disk /dev/sdc: 499.0 GB, 499036192768 bytes
Disk /dev/sdd: 499.0 GB, 499036192768 bytes
Disk /dev/sde: 499.0 GB, 499036192768 bytes
Disk /dev/sdf: 499.0 GB, 499036192768 bytes
Disk /dev/md0: 271 MB, 271319040 bytes
Disk /dev/md2: 21.4 GB, 21476081664 bytes
Disk /dev/md3: 21.4 GB, 21467889664 bytes
Disk /dev/md1: 21.4 GB, 21476081664 bytes

You can also use fdisk to obtain information about one specific hard disk device.

26
 Disk management

[root@rhel4 ~]# fdisk -l /dev/sda

Disk /dev/sda: 12.8 GB, 12884901888 bytes

255 heads, 63 sectors/track, 1566 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sda1 * 1 13 104391 83 Linux
/dev/sda2 14 1566 12474472+ 8e Linux LVM

Later we will use fdisk to do dangerous stuff like creating and deleting partitions.

/bin/dmesg

Kernel boot messages can be seen after boot with dmesg. Since hard disk devices
are detected by the kernel during boot, you can also use dmesg to find information
about disk devices.

root@barry:~# dmesg | grep "[hs]d[a-z]"

Kernel command line: root=/dev/hda1 ro
ide0: BM-DMA at 0xfc00-0xfc07, BIOS settings: hda:DMA, hdb:DMA
ide1: BM-DMA at 0xfc08-0xfc0f, BIOS settings: hdc:DMA, hdd:DMA
hda: ST360021A, ATA DISK drive
hdb: Maxtor 6Y080L0, ATA DISK drive
hdc: SONY DVD RW DRU-510A, ATAPI CD/DVD-ROM drive
hdd: SONY DVD RW DRU-810A, ATAPI CD/DVD-ROM drive
hda: max request size: 128KiB
hda: 117231408 sectors (60022 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA
hda: hda1 hda2
hdb: max request size: 128KiB
hdb: 160086528 sectors (81964 MB) w/2048KiB Cache, CHS=65535/16/63, UDMA
hdb: hdb1 hdb2
hdc: ATAPI 32X DVD-ROM DVD-R CD-R/RW drive, 8192kB Cache, UDMA(33)
hdd: ATAPI 40X DVD-ROM DVD-R CD-R/RW drive, 2048kB Cache, UDMA(33)
...

Here's another example of dmesg (same computer as above, but with extra 200gb
disk now).

paul@barry:~$ dmesg | grep -i "ata disk"

[ 2.624149] hda: ST360021A, ATA DISK drive
[ 2.904150] hdb: Maxtor 6Y080L0, ATA DISK drive
[ 3.472148] hdd: WDC WD2000BB-98DWA0, ATA DISK drive

Third and last example of dmesg running on RHEL5.3.

root@rhel53 ~# dmesg | grep -i "scsi disk"

sd 0:0:2:0: Attached scsi disk sda
sd 0:0:3:0: Attached scsi disk sdb
sd 0:0:6:0: Attached scsi disk sdc

27
 Disk management

/sbin/lsscsi
The /sbin/lsscsi will gve you a nice readable output of all scsi (and scsi emulated
devices). This first screenshot shows lsscsi on a SPARC system.

root@shaka:~# lsscsi
[0:0:0:0] disk Adaptec RAID5 V1.0 /dev/sda
[1:0:0:0] disk SEAGATE ST336605FSUN36G 0438 /dev/sdb
root@shaka:~#

Here is the same command, but run on a laptop with scsi emulated dvd writer and
scsi emulated usb.

paul@laika:~$ lsscsi
[0:0:0:0] disk ATA HTS721010G9SA00 MCZO /dev/sda
[1:0:0:0] disk ATA HTS721010G9SA00 MCZO /dev/sdb
[3:0:0:0] cd/dvd _NEC DVD_RW ND-7551A 1-02 /dev/scd0
[4:0:0:0] disk GENERIC USB Storage-CFC 019A /dev/sdc
[4:0:0:1] disk GENERIC USB Storage-SDC 019A /dev/sdd
[4:0:0:2] disk GENERIC USB Storage-SMC 019A /dev/sde
[4:0:0:3] disk GENERIC USB Storage-MSC 019A /dev/sdf

/proc/scsi/scsi
Another way to locate scsi devices is via the /proc/scsi/scsi file.

root@shaka:~# cat /proc/scsi/scsi

Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: Adaptec Model: RAID5 Rev: V1.0
Type: Direct-Access ANSI SCSI revision: 02
Host: scsi1 Channel: 00 Id: 00 Lun: 00
Vendor: SEAGATE Model: ST336605FSUN36G Rev: 0438
Type: Direct-Access ANSI SCSI revision: 03
root@shaka:~#

/sbin/scsi_info and /sbin/scsiinfo

There is also a scsi_info command, but this is not always installed by default.

root@shaka:~# scsi_info /dev/sdb

SCSI_ID="0,0,0"
HOST="1"
MODEL="SEAGATE ST336605FSUN36G"
FW_REV="0438"
root@shaka:~#

Another simple tool is scsiinfo which is a part of scsitools (also not installed by
default).

28
 Disk management

root@deb503:~# scsiinfo -l
/dev/sda /dev/sdb /dev/sdc

2.1.4. erasing a hard disk

Before selling your old hard disk on the internet, it might be a good idea to erase it. By
simply repartitioning, by using the Microsoft Windows format utility, or even after
an mkfs command, some people will still be able to read most of the data on the disk.

Although technically the /sbin/badblocks tool is meant to look for bad blocks, you
can use it to completely erase all data from a disk. Since this is really writing to every
sector of the disk, it can take a long time!

root@RHELv4u2:~# badblocks -ws /dev/sdb

Testing with pattern 0xaa: done
Reading and comparing: done
Testing with pattern 0x55: done
Reading and comparing: done
Testing with pattern 0xff: done
Reading and comparing: done
Testing with pattern 0x00: done
Reading and comparing: done

2.1.5. advanced hard disk settings

Tweaking of hard disk settings (dma, gap, ...) are not covered in this course. Several
tools exists, hdparm is just one of them.

/sbin/hdparm can be used to display or set information and parameters about an

ATA (or SATA) hard disk device. The -i and -I options will give you even more
information about the physical properties of the device.

root@laika:~# hdparm /dev/sdb

/dev/sdb:
IO_support = 0 (default 16-bit)
readonly = 0 (off)
readahead = 256 (on)
geometry = 12161/255/63, sectors = 195371568, start = 0

Below hdparm info about a 200GB IDE disk.

root@barry:~# hdparm /dev/hdd

/dev/hdd:
multcount = 0 (off)
IO_support = 0 (default)
unmaskirq = 0 (off)
using_dma = 1 (on)

29
 Disk management

keepsettings = 0 (off)
readonly = 0 (off)
readahead = 256 (on)
geometry = 24321/255/63, sectors = 390721968, start = 0

Use hdparm with care.

30
 Disk management

2.2. practice: hard disk devices

About this lab: To practice working with hard disks, you will need some hard disks.
When there are no physical hard disk available, you can use virtual disks in vmware
or VirtualBox. The teacher will help you in attaching a couple of ATA and/or SCSI
disks to a virtual machine. The results of this lab can be used in the next three labs
(partitions, file systems, mounting). It is adviced to attach at least one ide and three
equally sized scsi disks to the virtual machine.

1. Use dmesg to make a list of hard disk devices detected at boot-up.

2. Use fdisk to find the total size of all hard disk devices on your system.

3. Stop a virtual machine, add three virtual 1 gigabyte scsi hard disk devices and one
virtual 400 megabyte ide hard disk device. If possible, also add another virtual 400
megabyte ide disk.

4. Use dmesg to verify that all the new disks are properly detected at boot-up.

5. Verify that you can see the disk devices in /dev.

6. Use fdisk (with grep and /dev/null) to display the total size of the new disks.

7. Use badblocks to completely erase one of the smaller hard disks.

8. Look at /proc/scsi/scsi.

9. If possible, install lsscsi and use it to list the scsi disks.

31
 Disk management

2.3. solution: hard disk devices

1. Use dmesg to make a list of hard disk devices detected at boot-up.
Some possible answers...

dmesg | grep -i disk

Looking for ATA disks: dmesg | grep hd[abcd]

Looking for ATA disks: dmesg | grep -i "ata disk"

Looking for SCSI disks: dmesg | grep sd[a-f]

Looking for SCSI disks: dmesg | grep -i "scsi disk"

2. Use fdisk to find the total size of all hard disk devices on your system.
fdisk -l

3. Stop a virtual machine, add three virtual 1 gigabyte scsi hard disk devices and one
virtual 400 megabyte ide hard disk device. If possible, also add another virtual 400
megabyte ide disk.
This exercise happens in the settings of vmware or VirtualBox.

4. Use dmesg to verify that all the new disks are properly detected at boot-up.
See 1.

5. Verify that you can see the disk devices in /dev.

SCSI+SATA: ls -l /dev/sd*

ATA: ls -l /dev/hd*

6. Use fdisk (with grep and /dev/null) to display the total size of the new disks.

root@rhel53 ~# fdisk -l 2>/dev/null | grep [MGT]B

Disk /dev/hda: 21.4 GB, 21474836480 bytes
Disk /dev/hdb: 1073 MB, 1073741824 bytes
Disk /dev/sda: 2147 MB, 2147483648 bytes
Disk /dev/sdb: 2147 MB, 2147483648 bytes
Disk /dev/sdc: 2147 MB, 2147483648 bytes

7. Use badblocks to completely erase one of the smaller hard disks.

#Verify the device (/dev/sdc??) you want to erase before typing this.
#
root@rhel53 ~# badblocks -ws /dev/sdc
Testing with pattern 0xaa: done
Reading and comparing: done
Testing with pattern 0x55: done
Reading and comparing: done
Testing with pattern 0xff: done
Reading and comparing: done
Testing with pattern 0x00: done
Reading and comparing: done

32
 Disk management

8. Look at /proc/scsi/scsi.

root@rhel53 ~# cat /proc/scsi/scsi

Attached devices:
Host: scsi0 Channel: 00 Id: 02 Lun: 00
Vendor: VBOX Model: HARDDISK Rev: 1.0
Type: Direct-Access ANSI SCSI revision: 05
Host: scsi0 Channel: 00 Id: 03 Lun: 00
Vendor: VBOX Model: HARDDISK Rev: 1.0
Type: Direct-Access ANSI SCSI revision: 05
Host: scsi0 Channel: 00 Id: 06 Lun: 00
Vendor: VBOX Model: HARDDISK Rev: 1.0
Type: Direct-Access ANSI SCSI revision: 05

9. If possible, install lsscsi and use it to list the scsi disks.

Debian,Ubuntu: aptitude install lsscsi

Fedora: yum install lsscsi

root@rhel53 ~# lsscsi
[0:0:2:0] disk VBOX HARDDISK 1.0 /dev/sda
[0:0:3:0] disk VBOX HARDDISK 1.0 /dev/sdb
[0:0:6:0] disk VBOX HARDDISK 1.0 /dev/sdc

33
 Disk management

2.4. partitions

2.4.1. about partitions

Linux requires you to create one or more partitions. The next paragraphs will explain
how to create and use partitions.

A partition's geometry and size is usually defined by a starting and ending cylinder
(sometimes by sector). Partitions can be of type primary (maximum four), extended
(maximum one) or logical (contained within the extended partition). Each partition
has a type field that contains a code. This determines the computers operating system
or the partitions file system.

Table 2.3. primary, extended and logical partitions

Partition Type naming
Primary (max 4) 1-4
Extended (max 1) 1-4
Logical 5-

2.4.2. partition naming

We saw before that hard disk devices are named /dev/hdx or /dev/sdx with x
depending on the hardware configuration. Next is the partition number, starting the
count at 1. Hence the four (possible) primary partitions are numbered 1 to 4. Logical
partition counting always starts at 5. Thus /dev/hda2 is the second partition on the
first ATA hard disk device, and /dev/hdb5 is the first logical partition on the second
ATA hard disk device. Same for SCSI, /dev/sdb3 is the third partition on the second
SCSI disk.

Table 2.4. Partition naming

partition device
/dev/hda1 first primary partition on /dev/hda
/dev/hda2 second primary or extended partition on /dev/hda
/dev/sda5 first logical drive on /dev/sda
/dev/sdb6 second logical on /dev/sdb

2.4.3. discovering all partitions

fdisk -l
In the fdisk -l example below you can see that two partitions exist on /dev/sdb2. The
first partition spans 31 cylinders and contains a Linux swap partition. The second
partition is much bigger.

34
 Disk management

root@laika:~# fdisk -l /dev/sdb

Disk /dev/sdb: 100.0 GB, 100030242816 bytes

255 heads, 63 sectors/track, 12161 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sdb1 1 31 248976 82 Linux swap / Solaris
/dev/sdb2 32 12161 97434225 83 Linux
root@laika:~#

/proc/partitions

The /proc/partitions file contains a table with major and minor number of partitioned
devices, their number of blocks and the device name in /dev. Verify with /proc/
devices to link the major number to the proper device.

paul@RHELv4u4:~$ cat /proc/partitions

major minor #blocks name

3 0 524288 hda
3 64 734003 hdb
8 0 8388608 sda
8 1 104391 sda1
8 2 8281507 sda2
8 16 1048576 sdb
8 32 1048576 sdc
8 48 1048576 sdd
253 0 7176192 dm-0
253 1 1048576 dm-1

The major number corresponds to the device type (or driver) and can be found in
/proc/devices. In this case 3 corresponds to ide and 8 to sd. The major number
determines the device driver to be used with this device.

The minor number is a unique identification of an instance of this device type. The
devices.txt file in the kernel tree contains a full list of major and minor numbers.

other tools

You might be interested in alternatives to fdisk like parted, cfdisk, sfdisk and
gparted. This course mainly uses fdisk to partition hard disks.

2.4.4. partitioning new disks

In the example below, we bought a new disk for our system. After the new hardware
is properly attached, you can use fdisk and parted to create the necessary partition(s).
This example uses fdisk, but there is nothing wrong with using parted.

35
 Disk management

First, we check with fdisk -l whether Linux can see the new disk. Yes it does, the
new disk is seen as /dev/sdb, but it does not have any partitions yet.

root@RHELv4u2:~# fdisk -l

Disk /dev/sda: 12.8 GB, 12884901888 bytes

255 heads, 63 sectors/track, 1566 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sda1 * 1 13 104391 83 Linux
/dev/sda2 14 1566 12474472+ 8e Linux LVM

Disk /dev/sdb: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Disk /dev/sdb doesn't contain a valid partition table

Then we create a partition with fdisk on /dev/sdb. First we start the fdisk tool with /
dev/sdb as argument. Be very very careful not to partition the wrong disk!!

root@RHELv4u2:~# fdisk /dev/sdb

Device contains neither a valid DOS partition table, nor Sun, SGI...
Building a new DOS disklabel. Changes will remain in memory only,
until you decide to write them. After that, of course, the previous
content won't be recoverable.

Warning: invalid flag 0x0000 of partition table 4 will be corrected...

Inside the fdisk tool, we can issue the p command to see the current disks partition
table.

Command (m for help): p

Disk /dev/sdb: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

No partitions exist yet, so we issue n to create a new partition. We choose p for

primary, 1 for the partition number, 1 for the start cylinder and 14 for the end cylinder.

Command (m for help): n

Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 1
First cylinder (1-130, default 1):
Using default value 1
Last cylinder or +size or +sizeM or +sizeK (1-130, default 130): 14

36
 Disk management

We can now issue p again to verify our changes, but they are not yet written to disk.
This means we can still cancel this operation! But it looks good, so we use w to write
the changes to disk, and then quit the fdisk tool.

Command (m for help): p

Disk /dev/sdb: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sdb1 1 14 112423+ 83 Linux

Command (m for help): w

The partition table has been altered!

Calling ioctl() to re-read partition table.

Syncing disks.
root@RHELv4u2:~#

Let's verify again with fdisk -l to make sure reality fits our dreams. Indeed, the
screenshot below now shows a partition on /dev/sdb.

root@RHELv4u2:~# fdisk -l

Disk /dev/sda: 12.8 GB, 12884901888 bytes

255 heads, 63 sectors/track, 1566 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sda1 * 1 13 104391 83 Linux
/dev/sda2 14 1566 12474472+ 8e Linux LVM

Disk /dev/sdb: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sdb1 1 14 112423+ 83 Linux
root@RHELv4u2:~#

2.4.5. about the partition table

master boot record

The partition table information (primary and extended partitions) is written in the
master boot record or mbr. You can use dd to copy the mbr to a file.

This example copies the master boot record from the first SCSI hard disk.
dd if=/dev/sda of=/SCSIdisk.mbr bs=512 count=1

The same tool can also be used to wipe out all information about partitions on a disk.
This example writes zeroes over the master boot record.

37
 Disk management

dd if=/dev/zero of=/dev/sda bs=512 count=1

Or to wipe out the whole partition or disk.

dd if=/dev/zero of=/dev/sda

partprobe
Don't forget that after restoring a master boot record with dd, that you need to force
the kernel to reread the partition table with partprobe. After running partprobe, the
partitions can be used again.

[root@RHEL5 ~]# partprobe

[root@RHEL5 ~]#

logical drives
The partition table does not contain information about logical drives. So the dd
backup of the mbr only works for primary and extended partitions. To backup the
partition table including the logical drives, you can use sfdisk.

This example shows how to backup all partition and logical drive information to a file.
sfdisk -d /dev/sda < parttable.sda.sfdisk

The following example copies the mbr and all logical drive info from /dev/sda to /
dev/sdb.
sfdisk -d /dev/sda | sfdisk /dev/sdb

38
 Disk management

2.5. practice: partitions

1. Use fdisk -l to display existing partitions and sizes.

2. Use df -h to display existing partitions and sizes.

3. Compare the output of fdisk and df.

4. Create a 200MB primary partition on a small disk.

5. Create a 400MB primary partition and two 300MB logical drives on a big disk.

6. Use df -h and fdisk -l to verify your work.

7. Compare the output again of fdisk and df. Do both commands display the new
partitions ?

8. Create a backup with dd of the mbr that contains your 200MB primary partition.

9. Take a backup of the partition table containing your 400MB primary and 300MB
logical drives. Make sure the logical drives are in the backup.

10. (optional) Remove all your partitions with fdisk. Then restore your backups.

39
 Disk management

2.6. solution: partitions

1. Use fdisk -l to display existing partitions and sizes.
as root: # fdisk -l

2. Use df -h to display existing partitions and sizes.

df -h

3. Compare the output of fdisk and df.

Some partitions will be listed in both outputs (maybe /dev/sda1 or /dev/hda1).

4. Create a 200MB primary partition on a small disk.

Choose one of the disks you added (this example uses /dev/sdc).
root@rhel53 ~# fdisk /dev/sdc
...
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 1
First cylinder (1-261, default 1): 1
Last cylinder or +size or +sizeM or +sizeK (1-261, default 261): +200m
Command (m for help): w
The partition table has been altered!
Calling ioctl() to re-read partition table.
Syncing disks.

5. Create a 400MB primary partition and two 300MB logical drives on a big disk.
Choose one of the disks you added (this example uses /dev/sdb)

fdisk /dev/sdb

inside fdisk : n p 1 +400m enter --- n e 2 enter enter --- n l +300m (twice)

6. Use df -h and fdisk -l to verify your work.

fdisk -l ; df -h

7. Compare the output again of fdisk and df. Do both commands display the new
partitions ?
The newly created partitions are visible with fdisk.

But they are not displayed by df.

8. Create a backup with dd of the mbr that contains your 200MB primary partition.
dd if=/dev/sdc of=bootsector.sdc.dd count=1 bs=512

9. Take a backup of the partition table containing your 400MB primary and 300MB
logical drives. Make sure the logical drives are in the backup.
sfdisk -d /dev/sdb > parttable.sdb.sfdisk

40
 Disk management

2.7. file systems

2.7.1. about file systems

After you are finished partitioning the hard disk, you can put a file system on
each partition. A file system is a way of organizing files on your partition. Besides
file-based storage, file systems usually include directories and access control, and
contain meta information about files like access times, modification times and file
ownership.

The properties (length, character set, ...) of filenames are determined by the file
system you choose. Directories are usually implemented as files, you will have to
learn how this is implemented! Access control in file systems is tracked by user
ownership (and group owner- and membership) in combination with one or more
access control lists.

The manual page about filesystems(5) is usually accessed by typing man fs. You can
also look at /proc/filesystems for currently loaded file system drivers.

root@rhel53 ~# cat /proc/filesystems | grep -v nodev

ext2
iso9660
ext3

2.7.2. common file systems

ext2 and ext3

Once the most common Linux file systems is the ext2 (the second extended) file
system. A disadvantage is that file system checks on ext2 can take a long time.
You will see that ext2 is being replaced by ext3 on most Linux machines. They are
essentially the same, except for the journaling which is only present in ext3.

Journaling means that changes are first written to a journal on the disk. The journal
is flushed regularly, writing the changes in the file system. Journaling keeps the file
system in a consistent state, so you don't need a file system check after an unclean
shutdown or power failure.

You can create these file systems with the /sbin/mkfs or /sbin/mke2fs commands.
Use mke2fs -j to create an ext3 file system. You can convert an ext2 to ext3
with tune2fs -j. You can mount an ext3 file system as ext2, but then you lose the
journaling. Do not forget to run mkinitrd if you are booting from this device.

41
 Disk management

ext4
Since 2009 the newest incarnation of the ext file system is ext4 is available in the
Linux kernel. ext4 support larger files (up to 16 terabyte) and larger file systems than
ext3 (and many more features).

vfat
The vfat file system exists in a couple of forms : fat12 for floppy disks, fat16 on ms-
dos, and fat32 for larger disks. The Linux vfat implementation supports all of these,
but vfat lacks a lot of features like security and links. fat disks can be read by every
operating system, and are used a lot for digital cameras, usb sticks and to exchange
data between different OS'ses on a home user's computer.

iso 9660
iso 9660 is the standard format for cdroms. Chances are you will encounter this
file system also on your hard disk in the form of images of cdroms (often with
the .iso extension). The iso 9660 standard limits filenames to the 8.3 format. The Unix
world didn't like this, and thus added the rock ridge extensions, which allows for
filenames up to 255 characters and Unix-style file-modes, ownership and symbolic
links. Another extensions to iso 9660 is joliet, which adds 64 unicode characters to
the filename. The el torito standard extends iso 9660 to be able to boot from CD-
ROM's.

udf
Most optical media today (including cd's and dvd's) use udf, the Universal Disk
Format.

swap
All things considered, swap is not a file system. But to use a partition as a swap
partition it must be formatted and mounted as swap space.

others...
You might encounter reiserfs on older Linux systems. Maybe you will see Sun's zfs,
or one of the dozen other file systems available.

2.7.3. putting a file system on a partition

We now have a fresh partition. The system binaries to make file systems can be found
with ls.

42
 Disk management

[root@RHEL4b ~]# ls -lS /sbin/mk*

-rwxr-xr-x 3 root root 34832 Apr 24 2006 /sbin/mke2fs
-rwxr-xr-x 3 root root 34832 Apr 24 2006 /sbin/mkfs.ext2
-rwxr-xr-x 3 root root 34832 Apr 24 2006 /sbin/mkfs.ext3
-rwxr-xr-x 3 root root 28484 Oct 13 2004 /sbin/mkdosfs
-rwxr-xr-x 3 root root 28484 Oct 13 2004 /sbin/mkfs.msdos
-rwxr-xr-x 3 root root 28484 Oct 13 2004 /sbin/mkfs.vfat
-rwxr-xr-x 1 root root 20313 Apr 10 2006 /sbin/mkinitrd
-rwxr-x--- 1 root root 15444 Oct 5 2004 /sbin/mkzonedb
-rwxr-xr-x 1 root root 15300 May 24 2006 /sbin/mkfs.cramfs
-rwxr-xr-x 1 root root 13036 May 24 2006 /sbin/mkswap
-rwxr-xr-x 1 root root 6912 May 24 2006 /sbin/mkfs
-rwxr-xr-x 1 root root 5905 Aug 3 2004 /sbin/mkbootdisk
[root@RHEL4b ~]#

It is time for you to read the manual pages of mkfs and mke2fs. In the example below,
you see the creation of an ext2 file system on /dev/sdb1. In real life, you might want
to use options like -m0 and -j.

root@RHELv4u2:~# mke2fs /dev/sdb1

mke2fs 1.35 (28-Feb-2004)
Filesystem label=
OS type: Linux
Block size=1024 (log=0)
Fragment size=1024 (log=0)
28112 inodes, 112420 blocks
5621 blocks (5.00%) reserved for the super user
First data block=1
Maximum filesystem blocks=67371008
14 block groups
8192 blocks per group, 8192 fragments per group
2008 inodes per group
Superblock backups stored on blocks:
8193, 24577, 40961, 57345, 73729

Writing inode tables: done

Writing superblocks and filesystem accounting information: done

This filesystem will be automatically checked every 37 mounts or

180 days, whichever comes first. Use tune2fs -c or -i to override.

2.7.4. tuning a file system

You can use tune2fs to list and set file system settings. The first screenshot lists the
reserved space for root (which is set at five percent).

[root@rhel4 ~]# tune2fs -l /dev/sda1 | grep -i "block count"

Block count: 104388
Reserved block count: 5219
[root@rhel4 ~]#

This example changes this value to ten percent. You can use tune2fs while the file
system is active, even if it is the root file system (as in this example).

43
 Disk management

[root@rhel4 ~]# tune2fs -m10 /dev/sda1

tune2fs 1.35 (28-Feb-2004)
Setting reserved blocks percentage to 10 (10430 blocks)
[root@rhel4 ~]# tune2fs -l /dev/sda1 | grep -i "block count"
Block count: 104388
Reserved block count: 10430
[root@rhel4 ~]#

2.7.5. checking a file system

The fsck command is a front end tool used to check a file system for errors.

[root@RHEL4b ~]# ls /sbin/*fsck*

/sbin/dosfsck /sbin/fsck /sbin/fsck.ext2 /sbin/fsck.msdos
/sbin/e2fsck /sbin/fsck.cramfs /sbin/fsck.ext3 /sbin/fsck.vfat
[root@RHEL4b ~]#

The last column in /etc/fstab is used to determine whether a file system should be
checked at boot-up.

[paul@RHEL4b ~]$ grep ext /etc/fstab

/dev/VolGroup00/LogVol00 / ext3 defaults 1 1
LABEL=/boot /boot ext3 defaults 1 2
[paul@RHEL4b ~]$

Manually checking a mounted file system results in a warning from fsck.

[root@RHEL4b ~]# fsck /boot

fsck 1.35 (28-Feb-2004)
e2fsck 1.35 (28-Feb-2004)
/dev/sda1 is mounted.

WARNING!!! Running e2fsck on a mounted filesystem may cause

SEVERE filesystem damage.

Do you really want to continue (y/n)? no

check aborted.

But after unmounting fsck and e2fsck can be used to check an ext2 file system.

[root@RHEL4b ~]# fsck /boot

fsck 1.35 (28-Feb-2004)
e2fsck 1.35 (28-Feb-2004)
/boot: clean, 44/26104 files, 17598/104388 blocks
[root@RHEL4b ~]# fsck -p /boot
fsck 1.35 (28-Feb-2004)
/boot: clean, 44/26104 files, 17598/104388 blocks
[root@RHEL4b ~]# e2fsck -p /dev/sda1
/boot: clean, 44/26104 files, 17598/104388 blocks

44
 Disk management

2.8. practice: file systems

1. List the filesystems that are known by your system.

2. Create an ext2 filesystem on the 200MB partition.

3. Create an ext3 filesystem on the 400MB partition and one of the 300MB logical
drives.

4. Set the reserved space for root on the logical drive to 0 percent.

5. Verify your work with fdisk and df.

45
 Disk management

2.9. solution: file systems

1. List the filesystems that are known by your system.
man fs

cat /proc/filesystems

cat /etc/filesystems (not on all Linux distributions)

2. Create an ext2 filesystem on the 200MB partition.

mke2fs /dev/sdc1 (replace sdc1 with the correct partition)

3. Create an ext3 filesystem on the 400MB partition and one of the 300MB logical
drives.
mke2fs -j /dev/sdb1 (replace sdb1 with the correct partition)

mke2fs -j /dev/sdb5 (replace sdb5 with the correct partition)

4. Set the reserved space for root on the logical drive to 0 percent.
tune2fs -m 0 /dev/sdb5

5. Verify your work with fdisk and df.

mkfs (mke2fs) makes no difference in the output of these commands

The big change is in the next topic: mounting

46
 Disk management

2.10. mounting
Once you've put a file system on a partition, you can mount it. Mounting a file system
makes it available for use, usually as a directory. We say mounting a file system
instead of mounting a partition because we will see later that we can also mount file
systems that do not exists on partitions.

2.10.1. mounting local file systems

On all Unix systems, every file and every directory is part of one big file tree. To
access a file, you need to know the full path starting from the root directory. When
adding a file system to your computer, you need to make it available somewhere in
the file tree. The directory where you make a file system available is called a mount
point.

/bin/mkdir
This example shows how to create a new mount point with mkdir.

root@RHELv4u2:~# mkdir /home/project55

/bin/mount
When the mount point is created, and a file system is present on the partition, then
mount can mount the file system on the mount point directory.

root@RHELv4u2:~# mount -t ext2 /dev/sdb1 /home/project55/

Once mounted, the new file system is accessible to users.

/etc/filesystems
Actually the explicit -t ext2 option to set the file system is not always necessary. The
mount command is able to automatically detect a lot of file systems.

When mounting a file system without specifying explicitly the file system, then
mount will first probe /etc/filesystems. Mount will skip lines with the nodev
directive.

paul@RHELv4u4:~$ cat /etc/filesystems

ext3
ext2
nodev proc
nodev devpts

47
 Disk management

iso9660
vfat
hfs
paul@RHELv4u4:~$

/proc/filesystems
When /etc/filesystems does not exist, or ends with a single * on the last line, then
mount will read /proc/filesystems.

[root@RHEL52 ~]# cat /proc/filesystems | grep -v ^nodev

ext2
iso9660
ext3

2.10.2. displaying mounted file systems

To display all mounted file systems, issue the mount command. Or look at the files
/proc/mounts and /etc/mtab.

/bin/mount
The simplest and most common way to view all mounts is by issuing the mount
command without any arguments.

root@RHELv4u2:~# mount | grep /dev/sdb

/dev/sdb1 on /home/project55 type ext2 (rw)

/proc/mounts
The kernel provides the info in /proc/mounts in file form, but /proc/mounts does not
exist as a file on any hard disk. Looking at /proc/mounts is looking at information
that comes directly from the kernel.

root@RHELv4u2:~# cat /proc/mounts | grep /dev/sdb

/dev/sdb1 /home/project55 ext2 rw 0 0

/etc/mtab
The /etc/mtab file is not updated by the kernel, but is maintained by the mount
command. Do not edit /etc/mtab manually.

root@RHELv4u2:~# cat /etc/mtab | grep /dev/sdb

48
 Disk management

/dev/sdb1 /home/project55 ext2 rw 0 0

/bin/df
A more user friendly way to look at mounted file systems is df. The df (diskfree)
command has the added benefit of showing you the free space on each mounted disk.
Like a lot of Linux commands, df supports the -h switch to make the output more
human readable.

root@RHELv4u2:~# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/VolGroup00-LogVol00
11707972 6366996 4746240 58% /
/dev/sda1 101086 9300 86567 10% /boot
none 127988 0 127988 0% /dev/shm
/dev/sdb1 108865 1550 101694 2% /home/project55
root@RHELv4u2:~# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/VolGroup00-LogVol00
12G 6.1G 4.6G 58% /
/dev/sda1 99M 9.1M 85M 10% /boot
none 125M 0 125M 0% /dev/shm
/dev/sdb1 107M 1.6M 100M 2% /home/project55

In the df -h example below you can see the size, free space, used gigabytes and
percentage and mount point of a partition.

root@laika:~# df -h | egrep -e "(sdb2|File)"

Filesystem Size Used Avail Use% Mounted on
/dev/sdb2 92G 83G 8.6G 91% /media/sdb2
root@laika:~#

/bin/du
The du command can summarize disk usage for files and directories. Preventing du
to go into subdirectories with the -s option will give you a total for that directory.
This option is often used together with -h, so du -sh on a mount point gives the total
amount used in that partition.

root@pasha:~# du -sh /home/reet

881G /home/reet

2.10.3. permanent mounts

Until now, we performed all mounts manually. This works nice, until the next reboot.
Luckily there is a way to tell your computer to automatically mount certain file
systems during boot.

49
 Disk management

/etc/fstab
This is done using the file system table located in the /etc/fstab file. Below is a sample
/etc/fstab file.

root@RHELv4u2:~# cat /etc/fstab

/dev/VolGroup00/LogVol00 / ext3 defaults 1 1
LABEL=/boot /boot ext3 defaults 1 2
none /dev/pts devpts gid=5,mode=620 0 0
none /dev/shm tmpfs defaults 0 0
none /proc proc defaults 0 0
none /sys sysfs defaults 0 0
/dev/VolGroup00/LogVol01 swap swap defaults 0 0

By adding the following line, we can automate the mounting of a file system.

/dev/sdb1 /home/project55 ext2 defaults 0 0

mount /mountpoint
Adding an entry to /etc/fstab has the added advantage that you can simplify the
mount command. The command in the screenshot below forces mount to look for
the partition info in /etc/fstab.
# mount /home/project55

2.10.4. securing mounts

File systems can be secured with several mount options. Here are some examples.

ro
The ro option will mount a file system as read only, preventing anyone from writing.
root@rhel53 ~# mount -t ext2 -o ro /dev/hdb1 /home/project42
root@rhel53 ~# touch /home/project42/testwrite
touch: cannot touch `/home/project42/testwrite': Read-only file system

noexec
The noexec option will prevent the execution of binaries and scripts on the mounted
file system.
root@rhel53 ~# mount -t ext2 -o noexec /dev/hdb1 /home/project42
root@rhel53 ~# cp /bin/cat /home/project42
root@rhel53 ~# /home/project42/cat /etc/hosts
-bash: /home/project42/cat: Permission denied
root@rhel53 ~# echo echo hello > /home/project42/helloscript

50
 Disk management

root@rhel53 ~# chmod +x /home/project42/helloscript

root@rhel53 ~# /home/project42/helloscript
-bash: /home/project42/helloscript: Permission denied

nosuid
The nosuid option will ignore setuid bit set binaries on the mounted file system.

Note that you can still set the setuid bit on files.
root@rhel53 ~# mount -o nosuid /dev/hdb1 /home/project42
root@rhel53 ~# cp /bin/sleep /home/project42/
root@rhel53 ~# chmod 4555 /home/project42/sleep
root@rhel53 ~# ls -l /home/project42/sleep
-r-sr-xr-x 1 root root 19564 Jun 24 17:57 /home/project42/sleep

But users cannot exploit the setuid feature.

root@rhel53 ~# su - paul
[paul@rhel53 ~]$ /home/project42/sleep 500 &
[1] 2876
[paul@rhel53 ~]$ ps -f 2876
UID PID PPID C STIME TTY STAT TIME CMD
paul 2876 2853 0 17:58 pts/0 S 0:00 /home/project42/sleep 500
[paul@rhel53 ~]$

noacl
To prevent cluttering permissions with acl's, use the noacl option.
root@rhel53 ~# mount -o noacl /dev/hdb1 /home/project42

More mount options can be found in the manual page of mount.

51
 Disk management

2.11. practice: mounting file systems

1. Mount the small 200MB partition on /home/project22.

2. Mount the big 400MB primary partition on /mnt, the copy some files to it
(everything in /etc). Then umount, and mount the file system as read only on /srv/
nfs/salesnumbers. Where are the files you copied ?

3. Verify your work with fdisk, df and mount. Also look in /etc/mtab and /proc/
mounts.

4. Make both mounts permanent, test that it works.

5. What happens when you mount a file system on a directory that contains some
files ?

6. What happens when you mount two file systems on the same mount point ?

7. (optional) Describe the difference between these file searching commands: find,
locate, updatedb, whereis, apropos and which.

8. (optional) Perform a file system check on the partition mounted at /srv/nfs/

salesnumbers.

52
 Disk management

2.12. solution: mounting file systems

1. Mount the small 200MB partition on /home/project22.
mkdir /home/project22
mount /dev/sdc1 /home/project22

2. Mount the big 400MB primary partition on /mnt, the copy some files to it
(everything in /etc). Then umount, and mount the file system as read only on /srv/
nfs/salesnumbers. Where are the files you copied ?
mount /dev/sdb1 /mnt
cp -r /etc /mnt
ls -l /mnt

umount /mnt
ls -l /mnt

mkdir -p /srv/nfs/salesnumbers
mount /dev/sdb1 /srv/nfs/salesnumbers

You see the files in /srv/nfs/salenumbers now...

But physically they are on ext3 on partition /dev/sdb1

3. Verify your work with fdisk, df and mount. Also look in /etc/mtab and /proc/
mounts.
fdisk -l
df -h
mount

All three the above commands should show your mounted partitions.

grep project22 /etc/mtab

grep project22 /proc/mounts

4. Make both mounts permanent, test that it works.

add the following lines to /etc/fstab

/dev/sdc1 /home/project22 auto defaults 0 0

/dev/sdb1 /srv/nfs/salesnumbers auto defaults 0 0

5. What happens when you mount a file system on a directory that contains some
files ?
The files are hidden until umount.

6. What happens when you mount two file systems on the same mount point ?
Only the last mounted fs is visible.

7. (optional) Describe the difference between these file searching commands: find,
locate, updatedb, whereis, apropos and which.
man is your friend

8. (optional) Perform a file system check on the partition mounted at /srv/nfs/

salesnumbers.

53
 Disk management

better to unmount first before

# fsck /dev/sdb1

54
 Disk management

2.13. uuid and filesystems

2.13.1. about unique objects

A uuid or universally unique identifier is used to uniquely identify objects. This
128bit standard allows anyone to create a unique uuid.

/sbin/vol_id
Below we use the vol_id utility to display the uuid of an ext3 file system.

root@laika:~# vol_id --uuid /dev/sda1

825d4b79-ec40-4390-8a71-9261df8d4c82

/lib/udev/vol_id
Red Hat Enterprise Linux 5 puts vol_id in /lib/udev/vol_id, which is not in the
$PATH. The syntax is also a bit different from Debian/Ubuntu.

root@rhel53 ~# /lib/udev/vol_id -u /dev/hda1

48a6a316-9ca9-4214-b5c6-e7b33a77e860

/sbin/tune2fs
We can also use tune2fs to find the uuid of a file system.

[root@RHEL5 ~]# tune2fs -l /dev/sda1 | grep UUID

Filesystem UUID: 11cfc8bc-07c0-4c3f-9f64-78422ef1dd5c
[root@RHEL5 ~]# /lib/udev/vol_id -u /dev/sda1
11cfc8bc-07c0-4c3f-9f64-78422ef1dd5c

2.13.2. uuid in /etc/fstab

You can use the uuid to make sure that a volume is universally uniquely identified
in /etc/fstab. The device name can change depending on the disk devices that are
present at boot time, but a uuid never changes.

First we use tune2fs to find the uuid.

[root@RHEL5 ~]# tune2fs -l /dev/sdc1 | grep UUID

Filesystem UUID: 7626d73a-2bb6-4937-90ca-e451025d64e8

55
 Disk management

Then we check that it is properly added to /etc/fstab, the uuid replaces the variable
devicename /dev/sdc1.

[root@RHEL5 ~]# grep UUID /etc/fstab

UUID=7626d73a-2bb6-4937-90ca-e451025d64e8 /home/pro42 ext3 defaults 0 0

Now we can mount the volume using the mount point defined in /etc/fstab.

[root@RHEL5 ~]# mount /home/pro42

[root@RHEL5 ~]# df -h | grep 42
/dev/sdc1 397M 11M 366M 3% /home/pro42

The real test now, is to remove /dev/sdb from the system, reboot the machine and
see what happens. After the reboot, the disk previously known as /dev/sdc is now /
dev/sdb.

[root@RHEL5 ~]# tune2fs -l /dev/sdb1 | grep UUID

Filesystem UUID: 7626d73a-2bb6-4937-90ca-e451025d64e8

And thanks to the uuid in /etc/fstab, the mountpoint is mounted on the same disk
as before.

[root@RHEL5 ~]# df -h | grep sdb

/dev/sdb1 397M 11M 366M 3% /home/pro42

2.13.3. uuid in menu.lst

Recent incarnations of the Ubuntu distribution will use a uuid to identify the root file
system. This example shows how a root=/dev/sda1 is replaced with a uuid.

title Ubuntu 9.10, kernel 2.6.31-19-generic

uuid f001ba5d-9077-422a-9634-8d23d57e782a
kernel /boot/vmlinuz-2.6.31-19-generic \
root=UUID=f001ba5d-9077-422a-9634-8d23d57e782a ro quiet splash
initrd /boot/initrd.img-2.6.31-19-generic

The screenshot above contains only four lines. The line starting with root= is the
continuation of the kernel line.

56
 Disk management

2.14. practice: uuid and filesystems

1. Find the uuid of one of your ext3 partitions with tune2fs and vol_id.

2. Use this uuid in /etc/fstab and test that it works with a simple mount.

3. (optional) Test it also by removing a disk (so the device name is changed). You
can edit settings in vmware/Virtualbox to remove a hard disk.

4. Display the root= directive in /boot/grub/menu.lst. (We see later in the course
how to maintain this file.)

5. (optional) Replace the /dev/xxx in /boot/grub/menu.lst with a uuid (use an extra

stanza for this). Test that it works.

57
 Disk management

2.15. solution: uuid and filesystems

1. Find the uuid of one of your ext3 partitions with tune2fs and vol_id.
root@rhel55:~# /lib/udev/vol_id -u /dev/hda1
60926898-2c78-49b4-a71d-c1d6310c87cc

root@ubu1004:~# tune2fs -l /dev/sda2 | grep UUID

Filesystem UUID: 3007b743-1dce-2d62-9a59-cf25f85191b7

2. Use this uuid in /etc/fstab and test that it works with a simple mount.
tail -1 /etc/fstab
UUID=60926898-2c78-49b4-a71d-c1d6310c87cc /home/pro42 ext3 defaults 0 0

3. (optional) Test it also by removing a disk (so the device name is changed). You
can edit settings in vmware/Virtualbox to remove a hard disk.

4. Display the root= directive in /boot/grub/menu.lst. (We see later in the course
how to maintain this file.)
paul@deb503:~$ grep ^[^#] /boot/grub/menu.lst | grep root=
kernel /boot/vmlinuz-2.6.26-2-686 root=/dev/hda1 ro selinux=1 quiet
kernel /boot/vmlinuz-2.6.26-2-686 root=/dev/hda1 ro selinux=1 single

5. (optional) Replace the /dev/xxx in /boot/grub/menu.lst with a uuid (use an extra

stanza for this). Test that it works.

58
 Disk management

2.16. RAID

2.16.1. Hardware or software

Redundant Array of Independent Disks or RAID can be set up using hardware or
software. Hardware RAID is more expensive, but offers better performance. Software
RAID is cheaper and easier to manage, but it uses your CPU and your memory.

2.16.2. RAID levels

RAID 0
RAID 0 uses two or more disks, and is often called striping (or stripe set, or striped
volume). Data is divided in chunks, those chunks are evenly spread across every disk
in the array. The main advantage of RAID 0 is that you can create larger drives.
RAID 0 is the only RAID without redundancy.

JBOD
JBOD uses two or more disks, and is often called concatenating (spanning, spanned
set, or spanned volume). Data is written to the first disk, until it is full. Then data is
written to the second disk... The main advantage of JBOD (Just a Bunch of Disks) is
that you can create larger drives. JBOD offers no redundancy.

RAID 1
RAID 1 uses exactly two disks, and is often called mirroring (or mirror set, or
mirrored volume). All data written to the array is written on each disk. The main
advantage of RAID 1 is redundancy. The main disadvantage is that you lose at least
half of your available disk space (in other words, you at least double the cost).

RAID 2, 3 and 4 ?
RAID 2 uses bit level striping, RAID 3 byte level, and RAID 4 is the same as RAID 5,
but with a dedicated parity disk. This is actually slower than RAID 5, because every
write would have to write parity to this one (bottleneck) disk. It is unlikely that you
will ever see these RAID levels in production.

RAID 5
RAID 5 uses three or more disks, each divided into chunks. Every time chunks are
written to the array, one of the disks will receive a parity chunk. Unlike RAID 4,

59
 Disk management

the parity chunk will alternate between all disks. The main advantage of this is that
RAID 5 will allow for full data recovery in case of one hard disk failure.

RAID 6
RAID 6 is very similar to RAID 5, but uses two parity chunks. RAID 6 protects
against two hard disk failures.

RAID 0+1
RAID 0+1 is a mirror(1) of stripes(0). This means you first create two RAID 0 stripe
sets, and then you set them up as a mirror set. For example, when you have six 100GB
disks, then the stripe sets are each 300GB. Combined in a mirror, this makes 300GB
total. RAID 0+1 will survive one disk failure. It will only survive the second disk
failure if this disk is in the same stripe set as the previous failed disk.

RAID 1+0
RAID 1+0 is a stripe(0) of mirrors(1). For example, when you have six 100GB disks,
then you first create three mirrors of 100GB each. You then stripe them together into
a 300GB drive. In this example, as long as not all disks in the same mirror fail, it can
survive up to three hard disk failures.

RAID 50
RAID 5+0 is a stripe(0) of RAID 5 arrays. Suppose you have nine disks of 100GB,
then you can create three RAID 5 arrays of 200GB each. You can then combine them
into one large stripe set.

many others
There are many other nested RAID combinations, like RAID 30, 51, 60, 100, 150, ...

2.16.3. Building a software RAID array

You can do this during the installation with Disk Druid (easy), or afterwards using
the command line (not so easy).

First, you have to attach some disks to your computer. In this scenario, three brand
new disks of one gigabyte each are added. Check with fdisk -l that they are connected.

root@RHELv4u2:~# fdisk -l

Disk /dev/sda: 12.8 GB, 12884901888 bytes

60
 Disk management

255 heads, 63 sectors/track, 1566 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sda1 * 1 13 104391 83 Linux
/dev/sda2 14 1566 12474472+ 8e Linux LVM

Disk /dev/sdb: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Disk /dev/sdb doesn't contain a valid partition table

Disk /dev/sdc: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Disk /dev/sdc doesn't contain a valid partition table

Disk /dev/sdd: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Disk /dev/sdd doesn't contain a valid partition table

So far so good! Next step is to create a partition of type fd on every disk. The fd type
is to set the partition as Linux RAID auto. Like this screenshot shows.

root@RHELv4u2:~# fdisk /dev/sdc

Device contains neither a valid DOS partition table, nor Sun, SGI or \
OSF disklabel
Building a new DOS disklabel. Changes will remain in memory only,
until you decide to write them. After that, of course, the previous
content won't be recoverable.

Warning: invalid flag 0x0000 of partition table 4 will be corrected b\

y w(rite)

Command (m for help): n

Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 1
First cylinder (1-130, default 1):
Using default value 1
Last cylinder or +size or +sizeM or +sizeK (1-130, default 130):
Using default value 130

Command (m for help): t

Selected partition 1
Hex code (type L to list codes): fd
Changed system type of partition 1 to fd (Linux raid autodetect)

Command (m for help): p

Disk /dev/sdc: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sdc1 1 130 1044193+ fd Linux raid autodetect

61
 Disk management

Command (m for help): w

The partition table has been altered!

Calling ioctl() to re-read partition table.

Syncing disks.
root@RHELv4u2:~#

Now all three disks are ready for RAID, so we have to tell the system what to do
with these disks.

root@RHELv4u2:~# fdisk -l

Disk /dev/sda: 12.8 GB, 12884901888 bytes

255 heads, 63 sectors/track, 1566 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sda1 * 1 13 104391 83 Linux
/dev/sda2 14 1566 12474472+ 8e Linux LVM

Disk /dev/sdb: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sdb1 1 130 1044193+ fd Linux raid autodetect

Disk /dev/sdc: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sdc1 1 130 1044193+ fd Linux raid autodetect

Disk /dev/sdd: 1073 MB, 1073741824 bytes

255 heads, 63 sectors/track, 130 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System

/dev/sdd1 1 130 1044193+ fd Linux raid autodetect

The next step used to be create the RAID table in /etc/raidtab. Nowadays, you can
just issue the command mdadm with the correct parameters. The command below
is split on two lines to fit this print, but you should type it on one line, without the
backslash (\).

root@RHELv4u2:~# mdadm --create /dev/md0 --chunk=64 --level=5 --raid-d\

evices=3 /dev/sdb1 /dev/sdc1 /dev/sdd1
mdadm: array /dev/md0 started.

Below a partial screenshot how fdisk -l sees the RAID5

root@RHELv4u2:~# fdisk -l

<cut>

62
 Disk management

Disk /dev/md0: 2138 MB, 2138308608 bytes

2 heads, 4 sectors/track, 522048 cylinders
Units = cylinders of 8 * 512 = 4096 bytes

Disk /dev/md0 doesn't contain a valid partition table

We will use this software RAID 5 array in the next topic, LVM.

2.16.4. /proc/mdstat
The status of the raid devices can be seen in /proc/mdstat. This example shows a
RAID 5 in the process of rebuilding.

[root@RHEL5 ~]# cat /proc/mdstat

Personalities : [raid6] [raid5] [raid4]
md0 : active raid5 sdg1[3] sdf1[1] sde1[0]
1677056 blocks level 5, 64k chunk, algorithm 2 [3/2] [UU_]
[=================>...] recovery = 89.1% (747952/838528) finish\
=0.0min speed=25791K/sec

unused devices: >none<

This example shows an active software RAID 5.

[root@RHEL5 ~]# cat /proc/mdstat

Personalities : [raid6] [raid5] [raid4]
md0 : active raid5 sdg1[2] sdf1[1] sde1[0]
1677056 blocks level 5, 64k chunk, algorithm 2 [3/3] [UUU]

unused devices: >none<

When there is no software RAID present, the following is displayed.

paul@RHELv4u4:~$ cat /proc/mdstat

Personalities :
unused devices: <none>
paul@RHELv4u4:~$

2.16.5. Removing a software RAID

The software raid is visible in /proc/mdstat when active. To remove the raid
completely so you can use the disks for other purposes, you first have to stop (de-
activate) it with mdadm.
mdadm --stop /dev/md0

When stopped, you can remove the raid with mdadm.

mdadm --remove /dev/md0

63
 Disk management

The disks can now be repartitioned.

64
 Disk management

2.17. practice: RAID

1. Add three virtual disks of 200MB each to the virtual Red Hat machine.

2. Create a software RAID 5 on the three disks. (It is not necessary to put a filesystem
on it)

3. Verify with fdisk and in /proc that the RAID exists.

4. (optional) Stop and remove the RAID, unless you want to use it in the next chapter
LVM.

65
Chapter 3. Logical volume management

Table of Contents
3.1. introduction to lvm ........................................................................................ 67
3.2. lvm terminology ............................................................................................. 68
3.3. example: using lvm ........................................................................................ 69
3.4. example: extend a logical volume ................................................................. 71
3.5. example: resize a physical Volume ............................................................... 73
3.6. example: mirror a logical volume .................................................................. 75
3.7. example: snapshot a logical volume .............................................................. 76
3.8. verifying existing physical volumes .............................................................. 77
3.9. verifying existing volume groups .................................................................. 79
3.10. verifying existing logical volumes ............................................................... 81
3.11. manage physical volumes ............................................................................ 82
3.12. manage volume groups ................................................................................ 84
3.13. manage logical volumes ............................................................................... 86
3.14. practice : lvm ............................................................................................... 89

66
 Logical volume management

3.1. introduction to lvm

3.1.1. problems with standard partitions

There are some problems when working with hard disks and standard partitions.
Consider a system with a small and a large hard disk device, partitioned like this.
The first disk (/dev/sda) is partitioned in two, the second disk (/dev/sdb) has three
partitions.

Table 3.1. disk partitioning example

/dev/sda /dev/sdb
/dev/sda1 /dev/sda2 /dev/sdb1 /dev/sdb2 /dev/sdb3 unused
/boot / /var /home /project42
ext2 ext3 ext2 reiserfs ext3

In the example above, consider the options when you want to enlarge the space
available for /project42. What can you do ? The solution will always force you to
unmount the filesystem, take a backup of the data, remove and recreate partitions,
and then restore the data and remount the file system.

3.1.2. solution with lvm

Using lvm will create a virtual layer between the mounted file systems and the
hardware devices. This virtual layer will allow for an administrator to enlarge a
mounted file system in use. When lvm is properly used, then there is no need to
unmount the file system to enlarge it.

Table 3.2. LVM Example

/dev/sda /dev/sdb
Volume Group
/boot / /var /home /project42
ext2 ext3 ext2 reiserfs ext3

3.1.3. about lvm

Most lvm implementations support physical storage grouping, logical volume
resizing and data migration.

Physical storage grouping is a fancy name for grouping multiple physical devices
(hard disks) into a logical mass storage device. To enlarge this physical group, hard
disks or even single partitions can be added at a later time. The size of lvm volumes
on this physical group is independent of the individual size of the components. The
total size of the group is the limit.

67
 Logical volume management

One of the nicest features of lvm is the logical volume resizing. You can increase the
size of an lvm volume, sometimes even without any downtime. Additionally, you
can migrate data away from a failing hard disk device.

3.2. lvm terminology

3.2.1. physical volume (pv)

A physical volume is a disk, a partition or a (hardware or software) RAID device.
All these devices can become a member of a Volume Group.

3.2.2. volume group (vg)

A Volume Group is an abstraction layer between Physical Devices and Logical
Volumes.

3.2.3. logical volume (lv)

A Logical Volume is created in a Volume Group. Logical Volumes that contain
a file system can be mounted. The use of logical volumes is similar to the use
of partitions (both are standard block devices) and is accomplished with the same
standard commands (mkfs, mount, fsck, df, ...).

68
 Logical volume management

3.3. example: using lvm

This example shows how you can use a device (in this case /dev/sdc, but it could have
been /dev/sdb or any other disk or partition) with lvm, how to create a volume group
(vg) and how to create and use a logical volume (vg/lvol0).

First thing to do, is create physical volumes that can join the volume group with
pvcreate. This command makes a disk or partition available for use in Volume
Groups. The screenshot shows how to present the SCSI Disk device to LVM.

root@RHEL4:~# pvcreate /dev/sdc

Physical volume "/dev/sdc" successfully created

Note for home users: lvm will work fine when using the complete disk, but another
operating system on the same computer will not recognize lvm and will mark the disk
as being empty! You can avoid this by creating a partition that spans the whole disk,
then run pvcreate on the partition instead of the disk.

Then vgcreate creates a volume group using one device. Note that more devices
could be added to the volume group.

root@RHEL4:~# vgcreate vg /dev/sdc

Volume group "vg" successfully created

The last step lvcreate creates a logical volume.

root@RHEL4:~# lvcreate --size 500m vg

Logical volume "lvol0" created

The logical volume /dev/vg/lvol0 can now be formatted with ext2, and mounted for
normal use.

root@RHELv4u2:~# mke2fs -m0 -j /dev/vg/lvol0

mke2fs 1.35 (28-Feb-2004)
Filesystem label=
OS type: Linux
Block size=1024 (log=0)
Fragment size=1024 (log=0)
128016 inodes, 512000 blocks
0 blocks (0.00%) reserved for the super user
First data block=1
Maximum filesystem blocks=67633152
63 block groups
8192 blocks per group, 8192 fragments per group
2032 inodes per group
Superblock backups stored on blocks:
8193, 24577, 40961, 57345, 73729, 204801, 221185, 401409

Writing inode tables: done

Creating journal (8192 blocks): done
Writing superblocks and filesystem accounting information: done

69
 Logical volume management

This filesystem will be automatically checked every 37 mounts or

180 days, whichever comes first. Use tune2fs -c or -i to override.
root@RHELv4u2:~# mkdir /home/project10
root@RHELv4u2:~# mount /dev/vg/lvol0 /home/project10/
root@RHELv4u2:~# df -h | grep proj
/dev/mapper/vg-lvol0 485M 11M 474M 3% /home/project10

A logical volume is very similar to a partition, it can be formatted with a file system,
and can be mounted so users can access it.

70
 Logical volume management

3.4. example: extend a logical volume

A logical volume can be extended without unmounting the file system. Whether or
not a volume can be extended depends on the file system it uses. Volumes that are
mounted as vfat or ext2 cannot be extended, so in the example here we use the ext3
file system.

The fdisk command shows us newly added scsi-disks that will serve our lvm volume.
This volume will then be extended. First, take a look at these disks.

[root@RHEL5 ~]# fdisk -l | grep sd[bc]

Disk /dev/sdb doesn't contain a valid partition table
Disk /dev/sdc doesn't contain a valid partition table
Disk /dev/sdb: 1181 MB, 1181115904 bytes
Disk /dev/sdc: 429 MB, 429496320 bytes

You already know how to partition a disk, below the first disk is partitioned (in one
big primary partition), the second disk is left untouched.

[root@RHEL5 ~]# fdisk -l | grep sd[bc]

Disk /dev/sdc doesn't contain a valid partition table
Disk /dev/sdb: 1181 MB, 1181115904 bytes
/dev/sdb1 1 143 1148616 83 Linux
Disk /dev/sdc: 429 MB, 429496320 bytes

You also know how to prepare disks for lvm with pvcreate, and how to create a
volume group with vgcreate. This example adds both the partitioned disk and the
untouched disk to the volume group named vg2.

[root@RHEL5 ~]# pvcreate /dev/sdb1

Physical volume "/dev/sdb1" successfully created
[root@RHEL5 ~]# pvcreate /dev/sdc
Physical volume "/dev/sdc" successfully created
[root@RHEL5 ~]# vgcreate vg2 /dev/sdb1 /dev/sdc
Volume group "vg2" successfully created

You can use pvdisplay to verify that both the disk and the partition belong to the
volume group.

[root@RHEL5 ~]# pvdisplay | grep -B1 vg2

PV Name /dev/sdb1
VG Name vg2
--
PV Name /dev/sdc
VG Name vg2

And you are familiar both with the lvcreate command to create a small logical volume
and the mke2fs command to put ext2 on it.

71
 Logical volume management

[root@RHEL5 ~]# lvcreate --size 200m vg2

Logical volume "lvol0" created
[root@RHEL5 ~]# mke2fs -m20 -j /dev/vg2/lvol0
...

As you see, we end up with a mounted logical volume that according to df is almost
200 megabyte in size.

[root@RHEL5 ~]# mkdir /home/resizetest

[root@RHEL5 ~]# mount /dev/vg2/lvol0 /home/resizetest/
[root@RHEL5 ~]# df -h | grep resizetest
194M 5.6M 149M 4% /home/resizetest

Extending the volume is easy with lvextend.

[root@RHEL5 ~]# lvextend -L +100 /dev/vg2/lvol0

Extending logical volume lvol0 to 300.00 MB
Logical volume lvol0 successfully resized

But as you can see, there is a small problem: it appears that df is not able to display
the extended volume in its full size. This is because the filesystem is only set for the
size of the volume before the extension was added.

[root@RHEL5 ~]# df -h | grep resizetest

194M 5.6M 149M 4% /home/resizetest

With lvdisplay however we can see that the volume is indeed extended.

[root@RHEL5 ~]# lvdisplay /dev/vg2/lvol0 | grep Size

LV Size 300.00 MB

To finish the extension, you need resize2fs to span the filesystem over the full size
of the logical volume.

[root@RHEL5 ~]# resize2fs /dev/vg2/lvol0

resize2fs 1.39 (29-May-2006)
Filesystem at /dev/vg2/lvol0 is mounted on /home/resizetest; on-line re\
sizing required
Performing an on-line resize of /dev/vg2/lvol0 to 307200 (1k) blocks.
The filesystem on /dev/vg2/lvol0 is now 307200 blocks long.

Congratulations, you just successfully expanded a logical volume.

[root@RHEL5 ~]# df -h | grep resizetest

291M 6.1M 225M 3% /home/resizetest
[root@RHEL5 ~]#

72
 Logical volume management

3.5. example: resize a physical Volume

This is a humble demonstration of how to resize a physical Volume with lvm (after
you resize it with fdisk). The demonstration starts with a 100MB partition named /
dev/sde1. We used fdisk to create it, and to verify the size.

[root@RHEL5 ~]# fdisk -l 2>/dev/null | grep sde1

/dev/sde1 1 100 102384 83 Linux
[root@RHEL5 ~]#

Now we can use pvcreate to create the Physical Volume, followed by pvs to verify
the creation.

[root@RHEL5 ~]# pvcreate /dev/sde1

Physical volume "/dev/sde1" successfully created
[root@RHEL5 ~]# pvs | grep sde1
/dev/sde1 lvm2 -- 99.98M 99.98M
[root@RHEL5 ~]#

The next step is ti use fdisk to enlarge the partition (actually deleting it and then
recreating /dev/sde1 with more cylinders).

[root@RHEL5 ~]# fdisk /dev/sde

Command (m for help): p

Disk /dev/sde: 858 MB, 858993152 bytes

64 heads, 32 sectors/track, 819 cylinders
Units = cylinders of 2048 * 512 = 1048576 bytes

Device Boot Start End Blocks Id System

/dev/sde1 1 100 102384 83 Linux

Command (m for help): d

Selected partition 1

Command (m for help): n

Command action
e extended
p primary partition (1-4)
p
Partition number (1-4):
Value out of range.
Partition number (1-4): 1
First cylinder (1-819, default 1):
Using default value 1
Last cylinder or +size or +sizeM or +sizeK (1-819, default 819): 200

Command (m for help): w

The partition table has been altered!

Calling ioctl() to re-read partition table.

Syncing disks.
[root@RHEL5 ~]#

73
 Logical volume management

When we now use fdisk and pvs to verify the size of the partition and the Physical
Volume, then there is a size difference. LVM is still using the old size.

[root@RHEL5 ~]# fdisk -l 2>/dev/null | grep sde1

/dev/sde1 1 200 204784 83 Linux
[root@RHEL5 ~]# pvs | grep sde1
/dev/sde1 lvm2 -- 99.98M 99.98M
[root@RHEL5 ~]#

Executing pvresize on the Physical Volume will make lvm aware of the size change
of the partition. The correct size can be displayed with pvs.

[root@RHEL5 ~]# pvresize /dev/sde1

Physical volume "/dev/sde1" changed
1 physical volume(s) resized / 0 physical volume(s) not resized
[root@RHEL5 ~]# pvs | grep sde1
/dev/sde1 lvm2 -- 199.98M 199.98M
[root@RHEL5 ~]#

74
 Logical volume management

3.6. example: mirror a logical volume

We start by creating three physical volumes for lvm. Then we verify the creation and
the size with pvs. Three physical disks because lvm uses two disks for the mirror and
a third disk for the mirror log!

[root@RHEL5 ~]# pvcreate /dev/sdb /dev/sdc /dev/sdd

Physical volume "/dev/sdb" successfully created
Physical volume "/dev/sdc" successfully created
Physical volume "/dev/sdd" successfully created
[root@RHEL5 ~]# pvs
PV VG Fmt Attr PSize PFree
/dev/sdb lvm2 -- 409.60M 409.60M
/dev/sdc lvm2 -- 409.60M 409.60M
/dev/sdd lvm2 -- 409.60M 409.60M

Then we create the Volume Group and verify again with pvs. Notice how the three
physical volumes now belong to vg33, and how the size is rounded down (in steps
of the extent size, here 4MB).

[root@RHEL5 ~]# vgcreate vg33 /dev/sdb /dev/sdc /dev/sdd

Volume group "vg33" successfully created
[root@RHEL5 ~]# pvs
PV VG Fmt Attr PSize PFree
/dev/sda2 VolGroup00 lvm2 a- 15.88G 0
/dev/sdb vg33 lvm2 a- 408.00M 408.00M
/dev/sdc vg33 lvm2 a- 408.00M 408.00M
/dev/sdd vg33 lvm2 a- 408.00M 408.00M
[root@RHEL5 ~]#

The last step is to create the Logical Volume with lvcreate. Notice the -m 1 switch to
create one mirror. Notice also the change in free space in all three Physical Volumes!

[root@RHEL5 ~]# lvcreate --size 300m -n lvmir -m 1 vg33

Logical volume "lvmir" created
[root@RHEL5 ~]# pvs
PV VG Fmt Attr PSize PFree
/dev/sda2 VolGroup00 lvm2 a- 15.88G 0
/dev/sdb vg33 lvm2 a- 408.00M 108.00M
/dev/sdc vg33 lvm2 a- 408.00M 108.00M
/dev/sdd vg33 lvm2 a- 408.00M 404.00M

You can see the copy status of the mirror with lvs. It currently shows a 100 percent
copy.

[root@RHEL5 ~]# lvs vg33/lvmir

LV VG Attr LSize Origin Snap% Move Log Copy%
lvmir vg33 mwi-ao 300.00M lvmir_mlog 100.00

75
 Logical volume management

3.7. example: snapshot a logical volume

A snapshot is a virtual copy of all the data at a point in time on a volume. A snapshot
Logical Volume will retain a copy of all changed files of the snapshotted Logical
Volume.

The example below creates a snapshot of the bigLV Logical Volume.

[root@RHEL5 ~]# lvcreate -L100M -s -n snapLV vg42/bigLV

Logical volume "snapLV" created
[root@RHEL5 ~]#

You can see with lvs that the snapshot snapLV is indeed a snapshot of bigLV.
Moments after taking the snapshot, there are few changes to bigLV (0.02 percent).

[root@RHEL5 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy%
bigLV vg42 owi-a- 200.00M
snapLV vg42 swi-a- 100.00M bigLV 0.02
[root@RHEL5 ~]#

But after using bigLV for a while, more changes are done. This means the snapshot
volume has to keep more original data (10.22 percent).

[root@RHEL5 ~]# lvs | grep vg42

bigLV vg42 owi-ao 200.00M
snapLV vg42 swi-a- 100.00M bigLV 10.22
[root@RHEL5 ~]#

You can now use regular backup tools (dump, tar, cpio, ...) to take a backup of the
snapshot Logical Volume. This backup will contain all data as it existed on bigLV
at the time the snapshot was taken. When the backup is done, you can remove the
snapshot.

[root@RHEL5 ~]# lvremove vg42/snapLV

Do you really want to remove active logical volume "snapLV"? [y/n]: y
Logical volume "snapLV" successfully removed
[root@RHEL5 ~]#

76
 Logical volume management

3.8. verifying existing physical volumes

3.8.1. lvmdiskscan
To get a list of block devices that can be used with LVM, use lvmdiskscan. The
example below uses grep to limit the result to SCSI devices.

[root@RHEL5 ~]# lvmdiskscan | grep sd

/dev/sda1 [ 101.94 MB]
/dev/sda2 [ 15.90 GB] LVM physical volume
/dev/sdb [ 409.60 MB]
/dev/sdc [ 409.60 MB]
/dev/sdd [ 409.60 MB] LVM physical volume
/dev/sde1 [ 95.98 MB]
/dev/sde5 [ 191.98 MB]
/dev/sdf [ 819.20 MB] LVM physical volume
/dev/sdg1 [ 818.98 MB]
[root@RHEL5 ~]#

3.8.2. pvs
The easiest way to verify whether devices are known to lvm is with the pvs command.
The screenshot below shows that only /dev/sda2 is currently known for use with
LVM. It shows that /dev/sda2 is part of Volgroup00 and is almost 16GB in size. It
also shows /dev/sdc and /dev/sdd as part of vg33. The device /dev/sdb is knwon to
lvm, but not linked to any Volume Group.

[root@RHEL5 ~]# pvs

PV VG Fmt Attr PSize PFree
/dev/sda2 VolGroup00 lvm2 a- 15.88G 0
/dev/sdb lvm2 -- 409.60M 409.60M
/dev/sdc vg33 lvm2 a- 408.00M 408.00M
/dev/sdd vg33 lvm2 a- 408.00M 408.00M
[root@RHEL5 ~]#

3.8.3. pvscan
The pvscan command will scan all disks for existing Physical Volumes. The
information is similar to pvs, plus you get a line with total sizes.

[root@RHEL5 ~]# pvscan

PV /dev/sdc VG vg33 lvm2 [408.00 MB / 408.00 MB free]
PV /dev/sdd VG vg33 lvm2 [408.00 MB / 408.00 MB free]
PV /dev/sda2 VG VolGroup00 lvm2 [15.88 GB / 0 free]
PV /dev/sdb lvm2 [409.60 MB]
Total: 4 [17.07 GB] / in use: 3 [16.67 GB] / in no VG: 1 [409.60 MB]
[root@RHEL5 ~]#

77
 Logical volume management

3.8.4. pvdisplay
Use pvdisplay to get more information about physical volumes. You can also use
pvdisplay without an argument to display information about all physical (lvm)
volumes.

[root@RHEL5 ~]# pvdisplay /dev/sda2

--- Physical volume ---
PV Name /dev/sda2
VG Name VolGroup00
PV Size 15.90 GB / not usable 20.79 MB
Allocatable yes (but full)
PE Size (KByte) 32768
Total PE 508
Free PE 0
Allocated PE 508
PV UUID TobYfp-Ggg0-Rf8r-xtLd-5XgN-RSPc-8vkTHD

[root@RHEL5 ~]#

78
 Logical volume management

3.9. verifying existing volume groups

3.9.1. vgs
Similar to pvs is the use of vgs to display a quick overview of all volume groups.
There is only one volume group in the screenshot below, it is named VolGroup00
and is almost 16GB in size.

[root@RHEL5 ~]# vgs

VG #PV #LV #SN Attr VSize VFree
VolGroup00 1 2 0 wz--n- 15.88G 0
[root@RHEL5 ~]#

3.9.2. vgscan
The vgscan command will scan all disks for existing Volume Groups. It will also
update the /etc/lvm/.cache file. This file contains a list of all current lvm devices.

[root@RHEL5 ~]# vgscan

Reading all physical volumes. This may take a while...
Found volume group "VolGroup00" using metadata type lvm2
[root@RHEL5 ~]#

LVM will run the vgscan automatically at boot-up, so if you add hot swap devices,
then you will need to run vgscan to update /etc/lvm/.cache with the new devices.

3.9.3. vgdisplay
The vgdisplay command will give you more detailed information about a volume
group (or about all volume groups if you omit the argument).

[root@RHEL5 ~]# vgdisplay VolGroup00

--- Volume group ---
VG Name VolGroup00
System ID
Format lvm2
Metadata Areas 1
Metadata Sequence No 3
VG Access read/write
VG Status resizable
MAX LV 0
Cur LV 2
Open LV 2
Max PV 0
Cur PV 1
Act PV 1
VG Size 15.88 GB
PE Size 32.00 MB
Total PE 508

79
 Logical volume management

Alloc PE / Size 508 / 15.88 GB

Free PE / Size 0 / 0
VG UUID qsXvJb-71qV-9l7U-ishX-FobM-qptE-VXmKIg

[root@RHEL5 ~]#

80
 Logical volume management

3.10. verifying existing logical volumes

3.10.1. lvs
Use lvs for a quick look at all existing logical volumes. Below you can see two logical
volumes named LogVol00 and LogVol01.

[root@RHEL5 ~]# lvs

LV VG Attr LSize Origin Snap% Move Log Copy%
LogVol00 VolGroup00 -wi-ao 14.88G
LogVol01 VolGroup00 -wi-ao 1.00G
[root@RHEL5 ~]#

3.10.2. lvscan
The lvscan command will scan all disks for existing Logical Volumes.

[root@RHEL5 ~]# lvscan

ACTIVE '/dev/VolGroup00/LogVol00' [14.88 GB] inherit
ACTIVE '/dev/VolGroup00/LogVol01' [1.00 GB] inherit
[root@RHEL5 ~]#

3.10.3. lvdisplay
More detailed information about logical volumes is available through the
lvdisplay(1) command.

[root@RHEL5 ~]# lvdisplay VolGroup00/LogVol01

--- Logical volume ---
LV Name /dev/VolGroup00/LogVol01
VG Name VolGroup00
LV UUID RnTGK6-xWsi-t530-ksJx-7cax-co5c-A1KlDp
LV Write Access read/write
LV Status available
# open 1
LV Size 1.00 GB
Current LE 32
Segments 1
Allocation inherit
Read ahead sectors 0
Block device 253:1

[root@RHEL5 ~]#

81
 Logical volume management

3.11. manage physical volumes

3.11.1. pvcreate
Use the pvcreate command to add devices to lvm. This example shows how to add
a disk (or hardware RAID device) to lvm.

[root@RHEL5 ~]# pvcreate /dev/sdb

Physical volume "/dev/sdb" successfully created
[root@RHEL5 ~]#

This example shows how to add a partition to lvm.

[root@RHEL5 ~]# pvcreate /dev/sdc1

Physical volume "/dev/sdc1" successfully created
[root@RHEL5 ~]#

You can also add multiple disks or partitions as target to pvcreate. This example adds
three disks to lvm.

[root@RHEL5 ~]# pvcreate /dev/sde /dev/sdf /dev/sdg

Physical volume "/dev/sde" successfully created
Physical volume "/dev/sdf" successfully created
Physical volume "/dev/sdg" successfully created
[root@RHEL5 ~]#

3.11.2. pvremove
Use the pvremove command to remove physical volumes from lvm. The devices
may not be in use.

[root@RHEL5 ~]# pvremove /dev/sde /dev/sdf /dev/sdg

Labels on physical volume "/dev/sde" successfully wiped
Labels on physical volume "/dev/sdf" successfully wiped
Labels on physical volume "/dev/sdg" successfully wiped
[root@RHEL5 ~]#

3.11.3. pvresize
When you used fdisk to resize a partition on a disk, then you must use pvresize to
make lvm recognize the new size of the physical volume that represents this partition.

[root@RHEL5 ~]# pvresize /dev/sde1

Physical volume "/dev/sde1" changed
1 physical volume(s) resized / 0 physical volume(s) not resized

82
 Logical volume management

3.11.4. pvchange
With pvchange you can prevent the allocation of a Physical Volume in a new Volume
Group or Logical Volume. This can be useful if you plan to remove a Physical
Volume.

[root@RHEL5 ~]# pvchange -xn /dev/sdd

Physical volume "/dev/sdd" changed
1 physical volume changed / 0 physical volumes not changed
[root@RHEL5 ~]#

To revert your previous decision, this example shows you how te re-enable the
Physical Volume to allow allocation.

[root@RHEL5 ~]# pvchange -xy /dev/sdd

Physical volume "/dev/sdd" changed
1 physical volume changed / 0 physical volumes not changed
[root@RHEL5 ~]#

3.11.5. pvmove
With pvmove you can move Logical Volumes from within a Volume Group to
another Physical Volume. This must be done before removing a Physical Volume.

[root@RHEL5 ~]# pvs | grep vg1

/dev/sdf vg1 lvm2 a- 816.00M 0
/dev/sdg vg1 lvm2 a- 816.00M 816.00M
[root@RHEL5 ~]# pvmove /dev/sdf
/dev/sdf: Moved: 70.1%
/dev/sdf: Moved: 100.0%
[root@RHEL5 ~]# pvs | grep vg1
/dev/sdf vg1 lvm2 a- 816.00M 816.00M
/dev/sdg vg1 lvm2 a- 816.00M 0

83
 Logical volume management

3.12. manage volume groups

3.12.1. vgcreate
Use the vgcreate command to create a volume group. You can immediately name all
the physical volumes that span the volume group.

[root@RHEL5 ~]# vgcreate vg42 /dev/sde /dev/sdf

Volume group "vg42" successfully created
[root@RHEL5 ~]#

3.12.2. vgextend
Use the vgextend command to extend an existing volume group with a physical
volume.

[root@RHEL5 ~]# vgextend vg42 /dev/sdg

Volume group "vg42" successfully extended
[root@RHEL5 ~]#

3.12.3. vgremove
Use the vgremove command to remove volume groups from lvm. The volume groups
may not be in use.

[root@RHEL5 ~]# vgremove vg42

Volume group "vg42" successfully removed
[root@RHEL5 ~]#

3.12.4. vgreduce
Use the vgreduce command to remove a Physical Volume from the Volume Group.

The following example adds Physical Volume /dev/sdg to the vg1 Volume Group
using vgextend. And then removes it again using vgreduce.

[root@RHEL5 ~]# pvs | grep sdg

/dev/sdg lvm2 -- 819.20M 819.20M
[root@RHEL5 ~]# vgextend vg1 /dev/sdg
Volume group "vg1" successfully extended
[root@RHEL5 ~]# pvs | grep sdg
/dev/sdg vg1 lvm2 a- 816.00M 816.00M
[root@RHEL5 ~]# vgreduce vg1 /dev/sdg
Removed "/dev/sdg" from volume group "vg1"
[root@RHEL5 ~]# pvs | grep sdg

84
 Logical volume management

/dev/sdg lvm2 -- 819.20M 819.20M

3.12.5. vgchange
Use the vgchange command to change parameters of a Volume Group.

This example shows how to prevent Physical Volumes from being added or removed
to the Volume Group vg1.

[root@RHEL5 ~]# vgchange -xn vg1

Volume group "vg1" successfully changed
[root@RHEL5 ~]# vgextend vg1 /dev/sdg
Volume group vg1 is not resizable.

You can also use vgchange to change most other properties of a Volume Group. This
example changes the maximum number of Logical Volumes and maximum number
of Physical Volumes that vg1 can serve.

[root@RHEL5 ~]# vgdisplay vg1 | grep -i max

MAX LV 0
Max PV 0
[root@RHEL5 ~]# vgchange -l16 vg1
Volume group "vg1" successfully changed
[root@RHEL5 ~]# vgchange -p8 vg1
Volume group "vg1" successfully changed
[root@RHEL5 ~]# vgdisplay vg1 | grep -i max
MAX LV 16
Max PV 8

3.12.6. vgmerge
Merging two Volume Groups into one is done with vgmerge. The following example
merges vg2 into vg1, keeping all the properties of vg1.

[root@RHEL5 ~]# vgmerge vg1 vg2

Volume group "vg2" successfully merged into "vg1"
[root@RHEL5 ~]#

85
 Logical volume management

3.13. manage logical volumes

3.13.1. lvcreate
Use the lvcreate command to create Logical Volumes in a Volume Group. This
example creates an 8GB Logical Volume in Volume Group vg42.

[root@RHEL5 ~]# lvcreate -L5G vg42

Logical volume "lvol0" created
[root@RHEL5 ~]#

As you can see, lvm automatically names the Logical Volume lvol0. The next
example creates a 200MB Logical Volume named MyLV in Volume Group vg42.

[root@RHEL5 ~]# lvcreate -L200M -nMyLV vg42

Logical volume "MyLV" created
[root@RHEL5 ~]#

The next example does the same thing, but with different syntax.

[root@RHEL5 ~]# lvcreate --size 200M -n MyLV vg42

Logical volume "MyLV" created
[root@RHEL5 ~]#

This example creates a Logical Volume that occupies 10 percent of the Volume
Group.

[root@RHEL5 ~]# lvcreate -l 10%VG -n MyLV2 vg42

Logical volume "MyLV2" created
[root@RHEL5 ~]#

This example creates a Logical Volume that occupies 30 percent of the remaining
free space in the Volume Group.

[root@RHEL5 ~]# lvcreate -l 30%FREE -n MyLV3 vg42

Logical volume "MyLV3" created
[root@RHEL5 ~]#

3.13.2. lvremove
Use the lvremove command to remove Logical Volumes from a Volume Group.
Removing a Logical Volume requires the name of the Volume Group.

[root@RHEL5 ~]# lvremove vg42/MyLV

86
 Logical volume management

Do you really want to remove active logical volume "MyLV"? [y/n]: y

Logical volume "MyLV" successfully removed
[root@RHEL5 ~]#

Removing multiple Logical Volumes will request confirmation for each individual
volume.

[root@RHEL5 ~]# lvremove vg42/MyLV vg42/MyLV2 vg42/MyLV3

Do you really want to remove active logical volume "MyLV"? [y/n]: y
Logical volume "MyLV" successfully removed
Do you really want to remove active logical volume "MyLV2"? [y/n]: y
Logical volume "MyLV2" successfully removed
Do you really want to remove active logical volume "MyLV3"? [y/n]: y
Logical volume "MyLV3" successfully removed
[root@RHEL5 ~]#

3.13.3. lvextend
Extending the volume is easy with lvextend. This example extends a 200MB Logical
Volume with 100 MB.

[root@RHEL5 ~]# lvdisplay /dev/vg2/lvol0 | grep Size

LV Size 200.00 MB
[root@RHEL5 ~]# lvextend -L +100 /dev/vg2/lvol0
Extending logical volume lvol0 to 300.00 MB
Logical volume lvol0 successfully resized
[root@RHEL5 ~]# lvdisplay /dev/vg2/lvol0 | grep Size
LV Size 300.00 MB

The next example creates a 100MB Logical Volume, and then extends it to 500MB.

[root@RHEL5 ~]# lvcreate --size 100M -n extLV vg42

Logical volume "extLV" created
[root@RHEL5 ~]# lvextend -L 500M vg42/extLV
Extending logical volume extLV to 500.00 MB
Logical volume extLV successfully resized
[root@RHEL5 ~]#

This example doubles the size of a Logical Volume.

[root@RHEL5 ~]# lvextend -l+100%LV vg42/extLV

Extending logical volume extLV to 1000.00 MB
Logical volume extLV successfully resized
[root@RHEL5 ~]#

3.13.4. lvrename
Renaming a Logical Volume is done with lvrename. This example renames extLV
to bigLV in the vg42 Volume Group.

87
 Logical volume management

[root@RHEL5 ~]# lvrename vg42/extLV vg42/bigLV

Renamed "extLV" to "bigLV" in volume group "vg42"
[root@RHEL5 ~]#

88
 Logical volume management

3.14. practice : lvm

1. Create a volume group that contains a complete disk and a partition on another disk.

2. Create two logical volumes (a small one and a bigger one) in this volumegroup.
Format them wih ext3, mount them and copy some files to them.

3. Verify usage with fdisk, mount, pvs, vgs, lvs, pvdisplay, vgdisplay, lvdisplay and
df. Does fdisk give you any information about lvm?

4. Enlarge the small logical volume by 50 percent, and verify your work!

5. Take a look at other commands that start with vg* , pv* or lv*.

6. Create a mirror and a striped Logical Volume.

7. Convert a linear logical volume to a mirror.

8. Convert a mirror logical volume to a linear.

9. Create a snapshot of a Logical Volume, take a backup of the snapshot. Then delete
some files on the Logical Volume, then restore your backup.

10. Move your volume group to another disk (keep the Logical Volumes mounted).

11. If time permits, split a Volume Group with vgsplit, then merge it again with
vgmerge.

89
Chapter 4. Booting the system

Table of Contents
4.1. boot terminology ............................................................................................ 90
4.2. grub ................................................................................................................ 93
4.3. lilo .................................................................................................................. 98
4.4. practice : bootloader ....................................................................................... 99
4.5. solution : bootloader .................................................................................... 100

4.1. boot terminology

The exact order of things that happen when starting a computer system, depends on
the hardware architecture (Intel x86 is different from Sun Sparc etc), on the boot
loader (grub is different from lilo) and on the operating system (Linux, Solaris, BSD
etc). Most of this chapter is focused on booting Linux on Intel x86 with grub.

4.1.1. post
A computer starts booting the moment you turn on the power (no kidding). This first
process is called post or power on self test. If all goes well then this leads to the bios.
If all goes not so well, then you might hear nothing, or hear beeping, or see an error
message on the screen, or maybe see smoke coming out of the computer (burning
hardware smells bad!).

4.1.2. bios
All Intel x86 computers will have a basic input/output system or bios to detect,
identify and initialize hardware. The bios then goes looking for a boot device. This
can be a floppy, hard disk, cdrom, network card or usb drive.

During the bios you can see a message on the screen telling you which key (often
Del or F2) to press to enter the bios setup.

90
 Booting the system

4.1.3. openboot
Sun sparc systems start with openboot to test the hardware and to boot the operating
system. Bill Callkins explains openboot in his Solaris System Administration books.
The details of openboot are not the focus of this course.

4.1.4. boot password

The bios allows you to set a password. Do not forget this password, or you will have
to open up the hardware to reset it. You can sometimes set a password to boot the
system, and another password to protect the bios from being modified.

4.1.5. boot device

The bios will look for a boot device in the order configured in the bios setup. Usually
an operating system on a production server boots of a hard disk.

91
 Booting the system

4.1.6. master boot record

The master boot record or mbr is the first sector of a hard disk. The partitioning of
a disk in primary partitions, and the active partition are defined in the mbr.

The mbr is 512 bytes long and can be copied with dd.

dd if=/dev/sda of=bootsect.mbr count=1 bs=512

4.1.7. bootloader
The mbr is executed by the bios and contains either (a small) bootloader or code
to load a bootloader.

Looking at the mbr with od can reveal information about the bootloader.

paul@laika:~$ sudo dd if=/dev/sda count=1 bs=16 skip=24 2>/dev/null|od -c

0000000 376 G R U B \0 G e o m \0 H a r d
0000020

There are a variety of bootloaders available, most common on Intel architecture

is grub, which is replacing lilo in many places. When installing Linux on sparc
architecture, you can choose silo, Itanium systems can use elilo, IBM S/390 and
zSeries use z/IPL, Alpha uses milo and PowerPC architectures use yaboot (yet
another boot loader).

92
 Booting the system

Bootable cd's and dvd's often use syslinux.

4.1.8. kernel
The goal of all this is to load an operating system, or rather the kernel of an operating
system. A typical bootloader like grub will copy a kernel from hard disk to memory,
and will then hand control of the computer to the kernel (execute the kernel).

Once the Linux kernel is loaded, the bootloader turns control over to it. From that
moment on, the kernel is in control of the system. After discussing bootloaders, we
continue with the init system that starts all the daemons.

4.2. grub

4.2.1. about grub

The most common bootloader on linux systems today is grub. On almost all Intel
based systems grub is replacing lilo (the Linux loader). Even Solaris switched to
grub on x86 architecture.

One of the big advantages of grub over lilo is the capability to change the
configuration during boot (by pressing e to edit the boot command line).

4.2.2. /boot/grub/menu.lst
grub's configuration file is called menu.lst and is located in /boot/grub. The
screenshot below show the location and size of menu.lst on Debian.

root@barry:~# ls -l /boot/grub/menu.lst
-rw-r--r-- 1 root root 5155 2009-03-31 18:20 /boot/grub/menu.lst

4.2.3. /boot/grub/grub.conf
Some distributions like Red Hat Enterprise Linux 5 use grub.conf and provide
a symbolic link to menu.lst. This is the same file, only the name changed from
grub.conf to menu.lst. Notice also in this screenshot that this file is a lot smaller
on Red Hat.

[root@RHEL52 grub]# ls -l grub.conf menu.lst

-rw------- 1 root root 1346 Jan 21 04:20 grub.conf
lrwxrwxrwx 1 root root 11 Oct 11 2008 menu.lst -> ./grub.conf

93
 Booting the system

4.2.4. menu commands

The menu commands always have to be at the top of grub's configuration file.

default
The default command sets a default entry to start. The first entry has number 0.

default 0

fallback
In case the default does not boot, use the fallback entry instead.

fallback 1

timeout
The timeout will wait a number of seconds before booting the default entry.

timeout 5

hiddenmenu
The hiddenmenu will hide the grub menu unless the user presses Esc before the
timeout expires.

hiddenmenu

title
With title we can start a new entry or stanza.

title Debian Lenny

password
You can add a password to prevent interactive selection of a boot environment while
grub is running.

94
 Booting the system

password --md5 $1$Ec.id/$T2C2ahI/EG3WRRsmmu/HN/

Use the grub interactive shell to create the password hash.

grub> md5crypt

Password: ********
Encrypted: $1$Ec.id/$T2C2ahI/EG3WRRsmmu/HN/

4.2.5. stanza commands

Every operating system or kernel that you want to boot with grub will have a
stanza aka an entry of a couple of lines. Listed here are some of the common stanza
commands.

boot
Technically the boot command is only mandatory when running the grub command
line. This command does not have any parameters and can only be set as the last
command of a stanza.

boot

kernel
The kernel command points to the location of the kernel. To boot Linux this means
booting a gzip compressed zImage or bzip2 compressed bzImage.

This screenshot shows a typical kernel command used to load a Debian kernel.

kernel /boot/vmlinuz-2.6.17-2-686 root=/dev/hda1 ro

And this is how Red Hat uses the kernel command.

kernel /vmlinuz-2.6.18-128.el5 ro root=/dev/VolGroup00/LogVol00 rhgb quiet

initrd
Many Linux installations will need an initial ramdisk at boot time. This can be set
in grub with the initrd command.

Here a screenshot of Debian 4.0

95
 Booting the system

initrd /boot/initrd.img-2.6.17-2-686

And the same for Red Hat Enterprise Linux 5.3

initrd /initrd-2.6.18-128.el5.img

root
The root command accepts the root device as a parameter.

The root command will point to the hard disk and partition to use, with hd0 as the
first hard disk device and hd1 as the second hard disk device. The same numbering
is used for partitions, so hd0,0 is the first partition on the first disk and hd0,1 is the
second partition on that disk.

root (hd0,0)

savedefault
The savedefault command can be used together with default saved as a menu
command. This combination will set the currently booted stanza as the next default
stanza to boot.

default saved
timeout 10

title Linux
root (hd0,0)
kernel /boot/vmlinuz
savedefault

title DOS
root (hd0,1)
makeactive
chainloader +1
savedefault

4.2.6. chainloading
With grub booting, there are two choices: loading an operating system or
chainloading another bootloader. The chainloading feature of grub loads the
bootsector of a partition (that contains an operating system).

Some older operating systems require a primary partition that is set as active. Only
one partition can be set active so grub can do this on the fly just before chainloading.

This screenshot shows how to set the first primary partition active with grub.

96
 Booting the system

root (hd0,0)
makeactive

Chainloading refers to grub loading another operating system's bootloader. The

chainloader switch receives one option: the number of sectors to read and boot. For
DOS and OS/2 one sector is enough. Note that DOS requires the boot/root partition
to be active!

Here is a complete example to chainload an old operating system.

title MS-DOS 6.22

root (hd0,1)
makeactive
chainloader +1

4.2.7. stanza examples

This is a screenshot of a typical Debian 4.0 stanza.

title Debian GNU/Linux, kernel 2.6.17-2-686

root (hd0,0)
kernel /boot/vmlinuz-2.6.17-2-686 root=/dev/hda1 ro
initrd /boot/initrd.img-2.6.17-2-686

Here a screenshot of a typical Red Hat Enterprise Linux stanza.

title Red Hat Enterprise Linux Server (2.6.18-128.el5)

root (hd0,0)
kernel /vmlinuz-2.6.18-98.el5 ro root=/dev/VolGroup00/LogVol00 rhgb quiet
initrd /initrd-2.6.18-98.el5.img

4.2.8. editing grub at boot time

At boot time, when the grub menu is displayed, you can type e to edit the current
stanza. This enables you to add parameters to the kernel.

One such parameter, useful when you lost the root password, is single. This will boot
the kernel in single user mode (although some distributions will still require you to
type the root password.

kernel /boot/vmlinuz-2.6.17-2-686 root=/dev/hda1 ro single

Another option to reset a root password is to use an init=/bin/bash parameter.

kernel /boot/vmlinuz-2.6.17-2-686 root=/dev/hda1 ro init=/bin/bash

97
 Booting the system

4.2.9. installing grub

Run the grub-install command to install grub. The command requires a destination
for overwriting the boot sector or mbr.

# grub-install /dev/hda

4.3. lilo

4.3.1. Linux loader

lilo used to be the most used Linux bootloader, but is steadily being replaced in x86
with grub.

4.3.2. lilo.conf
Here is an example of a typical lilo.conf file. The delay switch receives a number in
tenths of a second. So the delay below is three seconds, not thirty!

boot = /dev/hda
delay = 30

image = /boot/vmlinuz
root = /dev/hda1
label = Red Hat 5.2

image = /boot/vmlinuz
root = /dev/hda2
label = S.U.S.E. 8.0

other = /dev/hda4
table = /dev/hda
label = MS-DOS 6.22

The configration file shows three example stanzas. The first one boots Red Hat from
the first partition on the first disk (hda1). The second stanza boots Suse 8.0 from the
next partition. The last one loads MS-DOS.

98
 Booting the system

4.4. practice : bootloader

1. Make a copy of the kernel, initrd and System.map files in /boot. Put the copies also
in /boot but replace 2.6.x with 3.0 (just imagine that Linux 3.0 is out.).

2. Add a stanza in grub for the 3.0 files. Make sure the title is different.

3. Set the boot menu timeout to 30 seconds.

4. Reboot and test the new stanza.

99
 Booting the system

4.5. solution : bootloader

1. Make a copy of the kernel, initrd and System.map files in /boot. Put the copies also
in /boot but replace 2.6.x with 3.0 (just imagine that Linux 3.0 is out.).

cd /boot
cp vmlinuz-2.6.18-8.e15 vmlinuz-3.0
cp initrd-2.6.18-8.e15.img initrd-3.0.img
cp System.map-2.6.18-8.e15 System.map-3.0

Do not forget the initrd file ends in .img .

2. Add a stanza in grub for the 3.0 files. Make sure the title is different.

[root@RHEL5 ~]# grep 3.0 /boot/grub/menu.lst

title Red Hat Enterprise Linux Server (3.0)
kernel /vmlinuz-3.0 ro root=/dev/VolGroup00/LogVol00 rhgb quiet
initrd /initrd-3.0.img

3. Set the boot menu timeout to 30 seconds.

[root@RHEL5 ~]# grep time /boot/grub/menu.lst

timeout=30

4. Reboot and test the new stanza.

100
Chapter 5. init
Table of Contents
5.1. about sysv init .............................................................................................. 101
5.2. system init(ialization) ................................................................................... 101
5.3. daemon or demon ? ...................................................................................... 106
5.4. starting and stopping daemons ..................................................................... 106
5.5. chkconfig ...................................................................................................... 106
5.6. update-rc.d .................................................................................................... 108
5.7. bum ............................................................................................................... 109
5.8. runlevels ....................................................................................................... 110
5.9. practice: init ................................................................................................. 113
5.10. solution : init .............................................................................................. 114

5.1. about sysv init

Many Linux distributions use init scripts to start daemons in the same way that Unix
System V did. This chapter will explain in detail how that works.

Init starts daemons by using scripts, where each script starts one daemon, and where
each script waits for the previous script to finish. This serial process of starting
daemons is slow, and although slow booting is not a problem on servers where
uptime is measured in years, the recent uptake of Linux on the desktop results in user
complaints.

To improve Linux startup speed, Canonical has developed upstart, which was
first used in Ubuntu. Solaris also used init up to Solaris 9, for Solaris 10 Sun has
developed Service Management Facility. Both systems start daemons in parallel
and can replace the SysV init scripts. There is also an ongoing effort to create initng
(init next generation).

5.2. system init(ialization)

5.2.1. process id 1
The kernel receives system control from the bootloader. After a while the kernel starts
the init daemon. The init daemon (/sbin/init) is the first daemon that is started and
receives process id 1 (PID 1). Init never dies.

5.2.2. configuration in /etc/inittab

When /sbin/init is started, it will first read its configuration file /etc/inittab. In that
file, it will look for the value of initdefault (3 in the screenshot below).

101
 init

[paul@rhel4 ~]$ grep ^id /etc/inittab

id:3:initdefault:

5.2.3. initdefault
The value found in initdefault indicates the default runlevel. Some Linux
distributions have a brief description of runlevels in /etc/inittab, like here on Red Hat
Enterprise Linux 4.

# Default runlevel. The runlevels used by RHS are:

# 0 - halt (Do NOT set initdefault to this)
# 1 - Single user mode
# 2 - Multiuser, without NFS (The same as 3, if you don't have network)
# 3 - Full multiuser mode
# 4 - unused
# 5 - X11
# 6 - reboot (Do NOT set initdefault to this)

Runlevel 0 means the system is shutting down. Runlevel 1 is used for

troubleshooting, only the root user can log on, and only at the console. Runlevel 3
is typical for servers, whereas runlevel 5 is typical for desktops (graphical logon).
Besides runlevels 0, 1 and 6, the use may vary depending on the distribution. Debian
and derived Linux systems have full network and GUI logon on runlevels 2 to 5. So
always verify the proper meaning of runlevels on your system.

5.2.4. sysinit script

/etc/rc.d/rc.sysinit
The next line in /etc/inittab in Red Hat and derivatives is the following.

si::sysinit:/etc/rc.d/rc.sysinit

This means that independent of the selected runlevel, init will run the /etc/rc.d/
rc.sysinit script. This script initializes hardware, sets some basic environment,
populates /etc/mtab while mounting file systems, starts swap and more.

[paul@rhel ~]$ egrep -e"^# Ini" -e"^# Sta" -e"^# Che" /etc/rc.d/rc.sysinit
# Check SELinux status
# Initialize hardware
# Start the graphical boot, if necessary; /usr may not be mounted yet...
# Initialiaze ACPI bits
# Check filesystems
# Start the graphical boot, if necessary and not done yet.
# Check to see if SELinux requires a relabel
# Initialize pseudo-random number generator

102
 init

# Start up swapping.
# Initialize the serial ports.

That egrep command could also have been written with grep like this :

grep "^# \(Ini\|Sta\|Che\)".

/etc/init.d/rcS
Debian has the following line after initdefault.

si::sysinit:/etc/init.d/rcS

The /etc/init.d/rcS script will always run on Debian (independent of the selected
runlevel). The script is actually running all scripts in the /etc/rcS.d/ directory in
alphabetical order.

root@barry:~# cat /etc/init.d/rcS

#! /bin/sh
#
# rcS
#
# Call all S??* scripts in /etc/rcS.d/ in numerical/alphabetical order
#

exec /etc/init.d/rc S

5.2.5. rc scripts
Init will continue to read /etc/inittab and meets this section on Debian Linux.

l0:0:wait:/etc/init.d/rc 0
l1:1:wait:/etc/init.d/rc 1
l2:2:wait:/etc/init.d/rc 2
l3:3:wait:/etc/init.d/rc 3
l4:4:wait:/etc/init.d/rc 4
l5:5:wait:/etc/init.d/rc 5
l6:6:wait:/etc/init.d/rc 6

On Red Hat Enterprise Linux it is identical except init.d is rc.d.

l0:0:wait:/etc/rc.d/rc 0
l1:1:wait:/etc/rc.d/rc 1
l2:2:wait:/etc/rc.d/rc 2
l3:3:wait:/etc/rc.d/rc 3
l4:4:wait:/etc/rc.d/rc 4
l5:5:wait:/etc/rc.d/rc 5
l6:6:wait:/etc/rc.d/rc 6

103
 init

In both cases, this means that init will start the rc script with the runlevel as the
only parameter. Actually /etc/inittab has fields seperated by colons. The second field
determines the runlevel in which this line should be executed. So in both cases, only
one line of the seven will be executed, depending on the runlevel set by initdefault.

5.2.6. rc directories
When you take a look any of the /etc/rcX.d/ directories, then you will see a lot of
(links to) scripts who's name start with either uppercase K or uppercase S.

[root@RHEL52 rc3.d]# ls -l | tail -4

lrwxrwxrwx 1 root root 19 Oct 11 2008 S98haldaemon -> ../init.d/haldaemon
lrwxrwxrwx 1 root root 19 Oct 11 2008 S99firstboot -> ../init.d/firstboot
lrwxrwxrwx 1 root root 11 Jan 21 04:16 S99local -> ../rc.local
lrwxrwxrwx 1 root root 16 Jan 21 04:17 S99smartd -> ../init.d/smartd

The /etc/rcX.d/ directories only contain links to scripts in /etc/init.d/. Links allow
for the script to have a different name. When entering a runlevel, all scripts that start
with uppercase K or uppercase S will be started in alphabetical order. Those that start
with K will be started first, with stop as the only parameter. The remaining scripts
with S will be started with start as the only parameter.

All this is done by the /etc/rc.d/rc script on Red Hat and by the /etc/init.d/rc script
on Debian.

5.2.7. mingetty

mingetty in /etc/inittab
Almost at the end of /etc/inittab there is a section to start and respawn several
mingetty daemons.

[root@RHEL4b ~]# grep getty /etc/inittab

# Run gettys in standard runlevels
1:2345:respawn:/sbin/mingetty tty1
2:2345:respawn:/sbin/mingetty tty2
3:2345:respawn:/sbin/mingetty tty3
4:2345:respawn:/sbin/mingetty tty4
5:2345:respawn:/sbin/mingetty tty5
6:2345:respawn:/sbin/mingetty tty6

mingetty and /bin/login

This /sbin/mingetty will display a message on a virtual console and allow you to
type a userid. Then it executes the /bin/login command with that userid. The /bin/

104
 init

login program will verify whether that user exists in /etc/passwd and prompt for (and
verify) a password. If the password is correct, /bin/login passes control to the shell
listed in /etc/passwd.

respawning mingetty
The mingetty daemons are started by init and watched until they die (user exits the
shell and is logged out). When this happens, the init daemon will respawn a new
mingetty. So even if you kill a mingetty daemon, it will be restarted automatically.

This example shows that init respawns mingetty daemons. Look at the PID's of the
last two mingetty processes.

[root@RHEL52 ~]# ps -C mingetty

PID TTY TIME CMD
2407 tty1 00:00:00 mingetty
2408 tty2 00:00:00 mingetty
2409 tty3 00:00:00 mingetty
2410 tty4 00:00:00 mingetty
2411 tty5 00:00:00 mingetty
2412 tty6 00:00:00 mingetty

When we kill the last two mingettys, then init will notice this and start them again
(with a different PID).

[root@RHEL52 ~]# kill 2411 2412

[root@RHEL52 ~]# ps -C mingetty
PID TTY TIME CMD
2407 tty1 00:00:00 mingetty
2408 tty2 00:00:00 mingetty
2409 tty3 00:00:00 mingetty
2410 tty4 00:00:00 mingetty
2821 tty5 00:00:00 mingetty
2824 tty6 00:00:00 mingetty

disabling a mingetty
You can disable a mingetty for a certain tty by removing the runlevel from the
second field in its line in /etc/inittab. Don't forget to tell init about the change of its
configuration file with kill -1 1.

The example below shows how to disable mingetty on tty3 to tty6 in runlevels 4 and 5.

[root@RHEL52 ~]# grep getty /etc/inittab

# Run gettys in standard runlevels
1:2345:respawn:/sbin/mingetty tty1
2:2345:respawn:/sbin/mingetty tty2
3:23:respawn:/sbin/mingetty tty3
4:23:respawn:/sbin/mingetty tty4
5:23:respawn:/sbin/mingetty tty5

105
 init

6:23:respawn:/sbin/mingetty tty6

5.3. daemon or demon ?

A daemon is a process that runs in background, without a link to a GUI or terminal.
Daemons are usually started at system boot, and stay alive until the system shuts
down. In more recent technical writings, daemons are often refered to as services.

Unix daemons are not to be confused with demons. Evi Nemeth, co-author of the
UNIX System Administration Handbook has the following to say about daemons:

Many people equate the word "daemon" with the word "demon", implying some
kind of satanic connection between UNIX and the underworld. This is an egregious
misunderstanding. "Daemon" is actually a much older form of "demon"; daemons
have no particular bias towards good or evil, but rather serve to help define a person's
character or personality. The ancient Greeks' concept of a "personal daemon" was
similar to the modern concept of a "guardian angel" ....

5.4. starting and stopping daemons

The K and S scripts are links to the real scripts in /etc/init.d/. These can also be used
when the system is running to start and stop daemons (or services). Most of them
accept the following parameters: start, stop, restart, status.

For example in this screenshot we restart the samba daemon.

root@laika:~# /etc/init.d/samba restart

* Stopping Samba daemons... [ OK ]
* Starting Samba daemons... [ OK ]

You can achieve the same result on RHEL/Fedora with the service command.

[root@RHEL4b ~]# service smb restart

Shutting down SMB services: [ OK ]
Shutting down NMB services: [ OK ]
Starting SMB services: [ OK ]
Starting NMB services: [ OK ]

You might also want to take a look at chkconfig, update-rc.d.

5.5. chkconfig
The purpose of chkconfig is to relieve system administrators of manually managing
all the links and scripts in /etc/init.d and /etc/rcX.d/.

106
 init

5.5.1. chkconfig --list

Here we use chkconfig to list the status of a service in the different runlevels. You
can see that the crond daemon (or service) is only activated in runlevels 2 to 5.

[root@RHEL52 ~]# chkconfig --list crond

crond 0:off 1:off 2:on 3:on 4:on 5:on 6:off

When you compare the screenshot above with the one below, you can see that off
equals to a K link to the script, whereas on equals to an S link.

[root@RHEL52 etc]# find ./rc?.d/ -name \*crond -exec ls -l {} \;|cut -b40-

./rc0.d/K60crond -> ../init.d/crond
./rc1.d/K60crond -> ../init.d/crond
./rc2.d/S90crond -> ../init.d/crond
./rc3.d/S90crond -> ../init.d/crond
./rc4.d/S90crond -> ../init.d/crond
./rc5.d/S90crond -> ../init.d/crond
./rc6.d/K60crond -> ../init.d/crond

5.5.2. runlevel configuration

Here you see how to use chkconfig to disable (or enable) a service in a certain
runlevel.

This screenshot shows how to disable crond in runlevel 3.

[root@RHEL52 ~]# chkconfig --level 3 crond off

[root@RHEL52 ~]# chkconfig --list crond
crond 0:off 1:off 2:on 3:off 4:on 5:on 6:off

This screenshot shows how to enable crond in runlevels 3 and 4.

[root@RHEL52 ~]# chkconfig --level 34 crond on

[root@RHEL52 ~]# chkconfig --list crond
crond 0:off 1:off 2:on 3:on 4:on 5:on 6:off

5.5.3. chkconfig configuration

Every script in /etc/init.d/ can have (comment) lines to tell chkconfig what to do with
the service. The line with # chkconfig: contains the runlevels in which the service
should be started (2345), followed by the priority for start (90) and stop (60).

[root@RHEL52 ~]# head -9 /etc/init.d/crond | tail -5

# chkconfig: 2345 90 60
# description: cron is a standard UNIX program that runs user-specified
# programs at periodic scheduled times. vixie cron adds a

107
 init

# number of features to the basic UNIX cron, including better

# security and more powerful configuration options.

5.5.4. enable and disable services

Services can be enabled or disabled in all runlevels with one command. Runlevels
0, 1 and 6 are always stopping services (or calling the scripts with stop) even when
their name starts with uppercase S.

[root@RHEL52 ~]# chkconfig crond off

[root@RHEL52 ~]# chkconfig --list crond
crond 0:off 1:off 2:off 3:off 4:off 5:off 6:off
[root@RHEL52 ~]# chkconfig crond on
[root@RHEL52 ~]# chkconfig --list crond
crond 0:off 1:off 2:on 3:on 4:on 5:on 6:off

5.6. update-rc.d

5.6.1. about update-rc.d

The Debian equivalent of chkconfig is called update-rc.d. This tool is designed for
use in scripts, if you prefer a graphical tool then look at bum.

When there are existing links in /etc/rcX.d/ then update-rc.d does not do anything.
This is to avoid that post installation scripts using update-rc.d are overwriting
changes made by a system administrator.

root@barry:~# update-rc.d cron remove

update-rc.d: /etc/init.d/cron exists during rc.d purge (use -f to force)

As you can see in the next screenshot, nothing changed for the cron daemon.

root@barry:~# find /etc/rc?.d/ -name '*cron' -exec ls -l {} \;|cut -b44-

/etc/rc0.d/K11cron -> ../init.d/cron
/etc/rc1.d/K11cron -> ../init.d/cron
/etc/rc2.d/S89cron -> ../init.d/cron
/etc/rc3.d/S89cron -> ../init.d/cron
/etc/rc4.d/S89cron -> ../init.d/cron
/etc/rc5.d/S89cron -> ../init.d/cron
/etc/rc6.d/K11cron -> ../init.d/cron

5.6.2. removing a service

Here we remove cron from all runlevels. Remember that the proper way to disable
a service is to put K scripts oin all runlevels!

108
 init

root@barry:~# update-rc.d -f cron remove

Removing any system startup links for /etc/init.d/cron ...
/etc/rc0.d/K11cron
/etc/rc1.d/K11cron
/etc/rc2.d/S89cron
/etc/rc3.d/S89cron
/etc/rc4.d/S89cron
/etc/rc5.d/S89cron
/etc/rc6.d/K11cron
root@barry:~# find /etc/rc?.d/ -name '*cron' -exec ls -l {} \;|cut -b44-
root@barry:~#

5.6.3. enable a service

This screenshot shows how to use update-rc.d to enable a service in runlevels 2, 3,
4 and 5 and disable the service in runlevels 0, 1 and 6.

root@barry:~# update-rc.d cron defaults

Adding system startup for /etc/init.d/cron ...
/etc/rc0.d/K20cron -> ../init.d/cron
/etc/rc1.d/K20cron -> ../init.d/cron
/etc/rc6.d/K20cron -> ../init.d/cron
/etc/rc2.d/S20cron -> ../init.d/cron
/etc/rc3.d/S20cron -> ../init.d/cron
/etc/rc4.d/S20cron -> ../init.d/cron
/etc/rc5.d/S20cron -> ../init.d/cron

5.6.4. customize a service

And here is an example on how to set your custom configuration for the cron daemon.

root@barry:~# update-rc.d -n cron start 11 2 3 4 5 . stop 89 0 1 6 .

Adding system startup for /etc/init.d/cron ...
/etc/rc0.d/K89cron -> ../init.d/cron
/etc/rc1.d/K89cron -> ../init.d/cron
/etc/rc6.d/K89cron -> ../init.d/cron
/etc/rc2.d/S11cron -> ../init.d/cron
/etc/rc3.d/S11cron -> ../init.d/cron
/etc/rc4.d/S11cron -> ../init.d/cron
/etc/rc5.d/S11cron -> ../init.d/cron

5.7. bum
This screenshot shows bum in advanced mode.

109
 init

5.8. runlevels

5.8.1. display the runlevel

You can see your current runlevel with the runlevel or who -r commands.

The runlevel command is typical Linux and will output the previous and the current
runlevel. If there was no previous runlevel, then it will mark it with the letter N.

[root@RHEL4b ~]# runlevel

N 3

The history of who -r dates back to Seventies Unix, it still works on Linux.

[root@RHEL4b ~]# who -r

run-level 3 Jul 28 09:15 last=S

5.8.2. changing the runlevel

You can switch to another runlevel with the telinit command. On Linux /sbin/telinit
is usually a (hard) link to /sbin/init.

This screenshot shows how to switch from runlevel 2 to runlevel 3 without reboot.

root@barry:~# runlevel
N 2
root@barry:~# init 3

110
 init

root@barry:~# runlevel
2 3

5.8.3. /sbin/shutdown
The shutdown command is used to properly shut down a system.

Common switches used with shutdown are -a, -t, -h and -r.

The -a switch forces /sbin/shutdown to use /etc/shutdown.allow. The -t switch is

used to define the number of seconds between the sending of the TERM signal and
the KILL signal. The -h switch halts the system instead of changing to runlevel 1.
The -r switch tells /sbin/shutdown to reboot after shutting down.

This screenshot shows how to use shutdown with five seconds between TERM and
KILL signals.

root@barry:~# shutdown -t5 -h now

The now is the time argument. This can be +m for the number of minutes to wait
before shutting down (with now as an alias for +0. The command will also accept
hh:mm instead of +m.

5.8.4. halt, reboot and poweroff

The binary /sbin/reboot is the same as /sbin/halt and /sbin/poweroff. Depending on
the name we use to call the command, it can behave differently.

When in runlevel 0 or 6 halt, reboot and poweroff will tell the kernel to halt, reboot
or poweroff the system.

When not in runlevel 0 or 6, typing reboot as root actually calls the shutdown
command with the -r switch and typing poweroff will switch off the power when
halting the system.

5.8.5. /var/log/wtmp
halt, reboot and poweroff all write to /var/log/wtmp. To look at /var/log/wtmp, we
need to use th last.

[root@RHEL52 ~]# last | grep reboot

reboot system boot 2.6.18-128.el5 Fri May 29 11:44 (192+05:01)
reboot system boot 2.6.18-128.el5 Wed May 27 12:10 (06:49)
reboot system boot 2.6.18-128.el5 Mon May 25 19:34 (1+15:59)
reboot system boot 2.6.18-128.el5 Mon Feb 9 13:20 (106+21:13)

111
 init

5.8.6. Ctrl-Alt-Del
When rc is finished starting all those scripts, init will continue to read /etc/inittab.
The next line is about what to do when the user hits Ctrl-Alt-Delete on the keyboard.

Here is what Debian 4.0 does.

root@barry:~# grep -i ctrl /etc/inittab

# What to do when CTRL-ALT-DEL is pressed.
ca:12345:ctrlaltdel:/sbin/shutdown -t1 -a -r now

Which is very similar to the default Red Hat Enterprise Linux 5.2 action.

[root@RHEL52 ~]# grep -i ctrl /etc/inittab

# Trap CTRL-ALT-DELETE
ca::ctrlaltdel:/sbin/shutdown -t3 -r now

One noticable difference is that Debian forces shutdown to use /etc/shutdown.allow,

where Red Hat allows everyone to invoke shutdown pressing Ctrl-Alt-Delete.

5.8.7. UPS and loss of power

[root@RHEL52 ~]# grep ^p /etc/inittab
pf::powerfail:/sbin/shutdown -f -h +2 "Power Failure; System Shutting Down"
pr:12345:powerokwait:/sbin/shutdown -c "Power Restored; Shutdown Cancelled"

It will read commands on what to execute in case of powerfailure, powerok and

Ctrl-Alt-Delete. The init process never stops keeping an eye on power failures and
that triple key combo.

root@barry:~# grep ^p /etc/inittab

pf::powerwait:/etc/init.d/powerfail start
pn::powerfailnow:/etc/init.d/powerfail now
po::powerokwait:/etc/init.d/powerfail stop

112
 init

5.9. practice: init

1. Change /etc/inittab so that only two mingetty's are respawned. Kill the other
mingetty's and verify that they don't come back.

2. Use the Red Hat Enterprise Linux virtual machine. Go to runlevel 5, display the
current and previous runlevel, then go back to runlevel 3.

3. Is the sysinit script on your computers setting or changing the PATH environment
variable ?

4. List all init.d scripts that are started in runlevel 2.

5. Write a script that acts like a daemon script in /etc/init.d/. It should have a case
statement to act on start/stop/restart and status. Test the script!

6. Use chkconfig to setup your script to start in runlevels 3,4 and 5, and to stop in
any other runlevel.

113
 init

5.10. solution : init

1. Change /etc/inittab so that only two mingetty's are respawned. Kill the other
mingetty's and verify that they don't come back.

Killing the mingetty's will result in init respawning them. You can edit /etc/inittab
so it looks like the screenshot below. Don't forget to also run kill -1 1.

[root@RHEL5 ~]# grep tty /etc/inittab

# Run gettys in standard runlevels
1:2345:respawn:/sbin/mingetty tty1
2:2345:respawn:/sbin/mingetty tty2
3:2:respawn:/sbin/mingetty tty3
4:2:respawn:/sbin/mingetty tty4
5:2:respawn:/sbin/mingetty tty5
6:2:respawn:/sbin/mingetty tty6
[root@RHEL5 ~]#

2. Use the Red Hat Enterprise Linux virtual machine. Go to runlevel 5, display the
current and previous runlevel, then go back to runlevel 3.

init 5 (watch the console for the change taking place)

runlevel
init 3 (again you can follow this on the console)

3. Is the sysinit script on your computers setting or changing the PATH environment
variable ?

On Red Hat, grep for PATH in /etc/rc.sysinit, on Debian/Ubuntu check /etc/rc.local

and /etc/ini.t/rc.local. The answer is probably no, but on RHEL5 the rc.sysinit script
does set the HOSTNAME variable.

[root@RHEL5 etc]# grep HOSTNAME rc.sysinit

4. List all init.d scripts that are started in runlevel 2.

root@RHEL5 ~# chkconfig --list | grep '2:on'

5. Write a script that acts like a daemon script in /etc/init.d/. It should have a case
statement to act on start/stop/restart and status. Test the script!

The script could look something like this.

#!/bin/bash
#
# chkconfig: 345 99 01
# description: pold demo script
#
# /etc/init.d/pold

114
 init

case "$1" in
start)
echo -n "Starting pold..."
sleep 1;
touch /var/lock/subsys/pold
echo "done."
echo pold started >> /var/log/messages
;;
stop)
echo -n "Stopping pold..."
sleep 1;
rm -rf /var/lock/subsys/pold
echo "done."
echo pold stopped >> /var/log/messages
;;
*)
echo "Usage: /etc/init.d/pold {start|stop}"
exit 1
;;
esac
exit 0

The touch /var/lock/subsys/pold is mandatory and must be the same filename as the
script name, if you want the stop sequence (the K01pold link) to be run.

6. Use chkconfig to setup your script to start in runlevels 3,4 and 5, and to stop in
any other runlevel.
chkconfig --add pold

The command above will only work when the # chkconfig: and # description: lines
in the pold script are there.

115
Chapter 6. Linux Kernel

Table of Contents
6.1. about the Linux kernel ................................................................................. 116
6.2. Linux kernel source ..................................................................................... 118
6.3. kernel boot files ........................................................................................... 122
6.4. Linux kernel modules .................................................................................. 123
6.5. compiling a kernel ....................................................................................... 127
6.6. compiling one module ................................................................................. 130

6.1. about the Linux kernel

6.1.1. kernel versions

In 1991 Linux Torvalds wrote (the first version of) the Linux kernel. He put it online,
and other people started contributing code. Over 4000 individuals contributed source
code to the latest kernel release (version 2.6.27 in November 2008).

Major Linux kernel versions used to come in even and odd numbers. Versions 2.0,
2.2, 2.4 and 2.6 are considered stable kernel versions. Whereas 2.1, 2.3 and 2.5 were
unstable (read development) versions. Since the release of 2.6.0 in January 2004, all
development has been done in the 2.6 tree. There is currently no v2.7.x and according
to Linus the even/stable vs odd/development scheme is abandoned forever.

6.1.2. uname -r
To see your current Linux kernel version, issue the uname -r command as shown
below.

This first example shows Linux major version 2.6 and minor version 24. The rest -22-
generic is specific to the distribution (Ubuntu in this case).

paul@laika:~$ uname -r
2.6.24-22-generic

The same command on Red Hat Enterprise Linux shows an older kernel (2.6.18) with
-92.1.17.el5 being specific to the distribution.

[paul@RHEL52 ~]$ uname -r

2.6.18-92.1.17.el5

116
 Linux Kernel

6.1.3. /proc/cmdline
The parameters that were passed to the kernel at boot time are in /proc/cmdline.

paul@RHELv4u4:~$ cat /proc/cmdline

ro root=/dev/VolGroup00/LogVol00 rhgb quiet

6.1.4. single user mode

When booting the kernel with the single parameter, it starts in single user mode.
Linux can start in a bash shell with the root user logged on (without password).

Some distributions prevent the use of this feature (at kernel compile time).

6.1.5. init=/bin/bash
Normally the kernel invokes init as the first daemon process. Adding init=/bin/bash
to the kernel parameters will instead invoke bash (again with root logged on without
providing a password).

6.1.6. /var/log/messages
The kernel reports during boot to syslog which writes a lot of kernel actions in /var/
log/messages. Looking at this file reveals when the kernel was started, including all
the devices that were detected at boot time.

[root@RHEL53 ~]# grep -A16 "syslogd 1.4.1:" /var/log/messages|cut -b24-

syslogd 1.4.1: restart.
kernel: klogd 1.4.1, log source = /proc/kmsg started.
kernel: Linux version 2.6.18-128.el5 ([email protected]...
kernel: BIOS-provided physical RAM map:
kernel: BIOS-e820: 0000000000000000 - 000000000009f800 (usable)
kernel: BIOS-e820: 000000000009f800 - 00000000000a0000 (reserved)
kernel: BIOS-e820: 00000000000ca000 - 00000000000cc000 (reserved)
kernel: BIOS-e820: 00000000000dc000 - 0000000000100000 (reserved)
kernel: BIOS-e820: 0000000000100000 - 000000001fef0000 (usable)
kernel: BIOS-e820: 000000001fef0000 - 000000001feff000 (ACPI data)
kernel: BIOS-e820: 000000001feff000 - 000000001ff00000 (ACPI NVS)
kernel: BIOS-e820: 000000001ff00000 - 0000000020000000 (usable)
kernel: BIOS-e820: 00000000fec00000 - 00000000fec10000 (reserved)
kernel: BIOS-e820: 00000000fee00000 - 00000000fee01000 (reserved)
kernel: BIOS-e820: 00000000fffe0000 - 0000000100000000 (reserved)
kernel: 0MB HIGHMEM available.
kernel: 512MB LOWMEM available.

This example shows how to use /var/log/messages to see kernel information about
/dev/sda.

[root@RHEL53 ~]# grep sda /var/log/messages | cut -b24-

kernel: SCSI device sda: 41943040 512-byte hdwr sectors (21475 MB)
kernel: sda: Write Protect is off

117
 Linux Kernel

kernel: sda: cache data unavailable

kernel: sda: assuming drive cache: write through
kernel: SCSI device sda: 41943040 512-byte hdwr sectors (21475 MB)
kernel: sda: Write Protect is off
kernel: sda: cache data unavailable
kernel: sda: assuming drive cache: write through
kernel: sda: sda1 sda2
kernel: sd 0:0:0:0: Attached scsi disk sda
kernel: EXT3 FS on sda1, internal journal

6.1.7. dmesg
The dmesg command prints out all the kernel bootup messages (from the last boot).

[root@RHEL53 ~]# dmesg | head

Linux version 2.6.18-128.el5 ([email protected])
BIOS-provided physical RAM map:
BIOS-e820: 0000000000000000 - 000000000009f800 (usable)
BIOS-e820: 000000000009f800 - 00000000000a0000 (reserved)
BIOS-e820: 00000000000ca000 - 00000000000cc000 (reserved)
BIOS-e820: 00000000000dc000 - 0000000000100000 (reserved)
BIOS-e820: 0000000000100000 - 000000001fef0000 (usable)
BIOS-e820: 000000001fef0000 - 000000001feff000 (ACPI data)
BIOS-e820: 000000001feff000 - 000000001ff00000 (ACPI NVS)
BIOS-e820: 000000001ff00000 - 0000000020000000 (usable)

Thus to find information about /dev/sda, using dmesg will yield only kernel messages
from the last boot.

[root@RHEL53 ~]# dmesg | grep sda

SCSI device sda: 41943040 512-byte hdwr sectors (21475 MB)
sda: Write Protect is off
sda: Mode Sense: 5d 00 00 00
sda: cache data unavailable
sda: assuming drive cache: write through
SCSI device sda: 41943040 512-byte hdwr sectors (21475 MB)
sda: Write Protect is off
sda: Mode Sense: 5d 00 00 00
sda: cache data unavailable
sda: assuming drive cache: write through
sda: sda1 sda2
sd 0:0:0:0: Attached scsi disk sda
EXT3 FS on sda1, internal journal

6.2. Linux kernel source

6.2.1. ftp.kernel.org
The home of the Linux kernel source is ftp.kernel.org. It contains all official releases
of the Linux kernel source code from 1991. It provides free downloads over http, ftp
and rsync of all these releases, as well as changelogs and patches. More information
can be otained on the website www.kernel.org.

118
 Linux Kernel

Anyone can anonymously use an ftp client to access ftp.kernel.org

paul@laika:~$ ftp ftp.kernel.org

Connected to pub3.kernel.org.
220 Welcome to ftp.kernel.org.
Name (ftp.kernel.org:paul): anonymous
331 Please specify the password.
Password:
230- Welcome to the
230-
230- LINUX KERNEL ARCHIVES
230- ftp.kernel.org

All the Linux kernel versions are located in the pub/linux/kernel/ directory.

ftp> ls pub/linux/kernel/v*
200 PORT command successful. Consider using PASV.
150 Here comes the directory listing.
drwxrwsr-x 2 536 536 4096 Mar 20 2003 v1.0
drwxrwsr-x 2 536 536 20480 Mar 20 2003 v1.1
drwxrwsr-x 2 536 536 8192 Mar 20 2003 v1.2
drwxrwsr-x 2 536 536 40960 Mar 20 2003 v1.3
drwxrwsr-x 3 536 536 16384 Feb 08 2004 v2.0
drwxrwsr-x 2 536 536 53248 Mar 20 2003 v2.1
drwxrwsr-x 3 536 536 12288 Mar 24 2004 v2.2
drwxrwsr-x 2 536 536 24576 Mar 20 2003 v2.3
drwxrwsr-x 5 536 536 28672 Dec 02 08:14 v2.4
drwxrwsr-x 4 536 536 32768 Jul 14 2003 v2.5
drwxrwsr-x 7 536 536 110592 Dec 05 22:36 v2.6
226 Directory send OK.
ftp>

6.2.2. /usr/src
On your local computer, the kernel source is located in /usr/src. Note though that
the structure inside /usr/src might be different depending on the distribution that you
are using.

First let's take a look at /usr/src on Debian. There appear to be two versions of the
complete Linux source code there. Looking for a specific file (e1000_main.c) with
find reveals it's exact location.

paul@barry:~$ ls -l /usr/src/
drwxr-xr-x 20 root root 4096 2006-04-04 22:12 linux-source-2.6.15
drwxr-xr-x 19 root root 4096 2006-07-15 17:32 linux-source-2.6.16
paul@barry:~$ find /usr/src -name e1000_main.c
/usr/src/linux-source-2.6.15/drivers/net/e1000/e1000_main.c
/usr/src/linux-source-2.6.16/drivers/net/e1000/e1000_main.c

This is very similar to /usr/src on Ubuntu, except there is only one kernel here (and
it is newer).

paul@laika:~$ ls -l /usr/src/

119
 Linux Kernel

drwxr-xr-x 23 root root 4096 2008-11-24 23:28 linux-source-2.6.24

paul@laika:~$ find /usr/src -name "e1000_main.c"
/usr/src/linux-source-2.6.24/drivers/net/e1000/e1000_main.c

Now take a look at /usr/src on Red Hat Enterprise Linux.

[paul@RHEL52 ~]$ ls -l /usr/src/

drwxr-xr-x 5 root root 4096 Dec 5 19:23 kernels
drwxr-xr-x 7 root root 4096 Oct 11 13:22 redhat

We will have to dig a little deeper to find the kernel source on Red Hat!

[paul@RHEL52 ~]$ cd /usr/src/redhat/BUILD/

[paul@RHEL52 BUILD]$ find . -name "e1000_main.c"
./kernel-2.6.18/linux-2.6.18.i686/drivers/net/e1000/e1000_main.c

6.2.3. downloading the kernel source

Debian
Installing the kernel source on Debian is really simple with aptitude install linux-
source. You can do a search for all linux-source packeges first, like in this screenshot.

root@barry:~# aptitude search linux-source

v linux-source -
v linux-source-2.6 -
id linux-source-2.6.15 - Linux kernel source for version 2.6.15
i linux-source-2.6.16 - Linux kernel source for version 2.6.16
p linux-source-2.6.18 - Linux kernel source for version 2.6.18
p linux-source-2.6.24 - Linux kernel source for version 2.6.24

And then use aptitude install to download and install the Debian Linux kernel source
code.

root@barry:~# aptitude install linux-source-2.6.24

When the aptitude is finished, you will see a new file named /usr/src/linux-source-
<version>.tar.bz2

root@barry:/usr/src# ls -lh
drwxr-xr-x 20 root root 4.0K 2006-04-04 22:12 linux-source-2.6.15
drwxr-xr-x 19 root root 4.0K 2006-07-15 17:32 linux-source-2.6.16
-rw-r--r-- 1 root root 45M 2008-12-02 10:56 linux-source-2.6.24.tar.bz2

Ubuntu
Ubuntu is based on Debian and also uses aptitude, so the task is very similar.

120
 Linux Kernel

root@laika:~# aptitude search linux-source

i linux-source - Linux kernel source with Ubuntu patches
v linux-source-2.6 -
i A linux-source-2.6.24 - Linux kernel source for version 2.6.24
root@laika:~# aptitude install linux-source

And when aptitude finishes, we end up with a /usr/src/linux-source-

<version>.tar.bz file.

oot@laika:~# ll /usr/src
total 45M
-rw-r--r-- 1 root root 45M 2008-11-24 23:30 linux-source-2.6.24.tar.bz2

Red Hat Enterprise Linux

The Red Hat kernel source is located on the fourth source cdrom. The file is called
kernel-2.6.9-42.EL.src.rpm (example for RHELv4u4). It is also available online
at ftp://ftp.redhat.com/pub/redhat/linux/enterprise/5Server/en/os/SRPMS/ (example
for RHEL5).

To download the kernel source on RHEL, use this long wget command (on one line,
without the trailing \).

wget ftp://ftp.redhat.com/pub/redhat/linux/enterprise/5Server/en/os/\
SRPMS/kernel-`uname -r`.src.rpm

When the wget download is finished, you end up with a 60M .rpm file.

[root@RHEL52 src]# ll
total 60M
-rw-r--r-- 1 root root 60M Dec 5 20:54 kernel-2.6.18-92.1.17.el5.src.rpm
drwxr-xr-x 5 root root 4.0K Dec 5 19:23 kernels
drwxr-xr-x 7 root root 4.0K Oct 11 13:22 redhat

We will need to perform some more steps before this can be used as kernel source
code.

First, we issue the rpm -i kernel-2.6.9-42.EL.src.rpm command to install this Red

Hat package.

[root@RHEL52 src]# ll
total 60M
-rw-r--r-- 1 root root 60M Dec 5 20:54 kernel-2.6.18-92.1.17.el5.src.rpm
drwxr-xr-x 5 root root 4.0K Dec 5 19:23 kernels
drwxr-xr-x 7 root root 4.0K Oct 11 13:22 redhat
[root@RHEL52 src]# rpm -i kernel-2.6.18-92.1.17.el5.src.rpm

The we move to the SPECS directory and perform an rpmbuild.

121
 Linux Kernel

[root@RHEL52 ~]# cd /usr/src/redhat/SPECS

[root@RHEL52 SPECS]# rpmbuild -bp -vv --target=i686 kernel-2.6.spec

The rpmbuild command put the RHEL Linux kernel source code in /usr/src/redhat/
BUILD/kernel-<version>/.

[root@RHEL52 kernel-2.6.18]# pwd

/usr/src/redhat/BUILD/kernel-2.6.18
[root@RHEL52 kernel-2.6.18]# ll
total 20K
drwxr-xr-x 2 root root 4.0K Dec 6 2007 config
-rw-r--r-- 1 root root 3.1K Dec 5 20:58 Config.mk
drwxr-xr-x 20 root root 4.0K Dec 5 20:58 linux-2.6.18.i686
drwxr-xr-x 19 root root 4.0K Sep 20 2006 vanilla
drwxr-xr-x 8 root root 4.0K Dec 6 2007 xen

6.3. kernel boot files

6.3.1. vmlinuz
The vmlinuz file in /boot is the compressed kernel.

paul@barry:~$ ls -lh /boot | grep vmlinuz

-rw-r--r-- 1 root root 1.2M 2006-03-06 16:22 vmlinuz-2.6.15-1-486
-rw-r--r-- 1 root root 1.1M 2006-03-06 16:30 vmlinuz-2.6.15-1-686
-rw-r--r-- 1 root root 1.3M 2008-02-11 00:00 vmlinuz-2.6.18-6-686
paul@barry:~$

6.3.2. initrd
The kernel uses initrd (an initial RAM disk) at boot time. The initrd is mounted before
the kernel loads, and can contain additional drivers and modules. It is a compressed
cpio archive, so you can look at the contents in this way.

root@RHELv4u4:/boot# mkdir /mnt/initrd

root@RHELv4u4:/boot# cp initrd-2.6.9-42.0.3.EL.img TMPinitrd.gz
root@RHELv4u4:/boot# gunzip TMPinitrd.gz
root@RHELv4u4:/boot# file TMPinitrd
TMPinitrd: ASCII cpio archive (SVR4 with no CRC)
root@RHELv4u4:/boot# cd /mnt/initrd/
root@RHELv4u4:/mnt/initrd# cpio -i | /boot/TMPinitrd
4985 blocks
root@RHELv4u4:/mnt/initrd# ls -l
total 76
drwxr-xr-x 2 root root 4096 Feb 5 08:36 bin
drwxr-xr-x 2 root root 4096 Feb 5 08:36 dev
drwxr-xr-x 4 root root 4096 Feb 5 08:36 etc
-rwxr-xr-x 1 root root 1607 Feb 5 08:36 init
drwxr-xr-x 2 root root 4096 Feb 5 08:36 lib
drwxr-xr-x 2 root root 4096 Feb 5 08:36 loopfs
drwxr-xr-x 2 root root 4096 Feb 5 08:36 proc

122
 Linux Kernel

lrwxrwxrwx 1 root root 3 Feb 5 08:36 sbin -> bin

drwxr-xr-x 2 root root 4096 Feb 5 08:36 sys
drwxr-xr-x 2 root root 4096 Feb 5 08:36 sysroot
root@RHELv4u4:/mnt/initrd#

6.3.3. System.map
The System.map contains the symbol table and changes with every kernel compile.
The symbol table is also present in /proc/kallsyms (pre 2.6 kernels name this file /
proc/ksyms).

root@RHELv4u4:/boot# head System.map-`uname -r`

00000400 A __kernel_vsyscall
0000041a A SYSENTER_RETURN_OFFSET
00000420 A __kernel_sigreturn
00000440 A __kernel_rt_sigreturn
c0100000 A _text
c0100000 T startup_32
c01000c6 t checkCPUtype
c0100147 t is486
c010014e t is386
c010019f t L6
root@RHELv4u4:/boot# head /proc/kallsyms
c0100228 t _stext
c0100228 t calibrate_delay_direct
c0100228 t stext
c0100337 t calibrate_delay
c01004db t rest_init
c0100580 t do_pre_smp_initcalls
c0100585 t run_init_process
c01005ac t init
c0100789 t early_param_test
c01007ad t early_setup_test
root@RHELv4u4:/boot#

6.3.4. .config
The last file copied to the /boot directory is the kernel configuration used for
compilation. This file is not necessary in the /boot directory, but it is common practice
to put a copy there. It allows you to recompile a kernel, starting from the same
configuration as an existing working one.

6.4. Linux kernel modules

6.4.1. about kernel modules

The Linux kernel is a monolithic kernel with loadable modules. These modules
contain parts of the kernel used typically for device drivers, file systems and network
protocols. Most of the time the necessary kernel modules are loaded automatically
and dynamically without administrator interaction.

123
 Linux Kernel

6.4.2. /lib/modules
The modules are stored in the /lib/modules/<kernel-version> directory. There is a
separate directory for each kernel that was compiled for your system.

paul@laika:~$ ll /lib/modules/
total 12K
drwxr-xr-x 7 root root 4.0K 2008-11-10 14:32 2.6.24-16-generic
drwxr-xr-x 8 root root 4.0K 2008-12-06 15:39 2.6.24-21-generic
drwxr-xr-x 8 root root 4.0K 2008-12-05 12:58 2.6.24-22-generic

6.4.3. <module>.ko
The file containing the modules usually ends in .ko. This screenshot shows the
location of the isdn module files.

paul@laika:~$ find /lib/modules -name isdn.ko

/lib/modules/2.6.24-21-generic/kernel/drivers/isdn/i4l/isdn.ko
/lib/modules/2.6.24-22-generic/kernel/drivers/isdn/i4l/isdn.ko
/lib/modules/2.6.24-16-generic/kernel/drivers/isdn/i4l/isdn.ko

6.4.4. lsmod
To see a list of currently loaded modules, use lsmod. You see the name of each loaded
module, the size, the use count, and the names of other modules using this one.

[root@RHEL52 ~]# lsmod | head -5

Module Size Used by
autofs4 24517 2
hidp 23105 2
rfcomm 42457 0
l2cap 29505 10 hidp,rfcomm

6.4.5. /proc/modules
/proc/modules lists all modules loaded by the kernel. The output would be too long
to display here, so lets grep for the vm module.

We see that vmmon and vmnet are both loaded. You can display the same information
with lsmod. Actually lsmod only reads and reformats the output of /proc/modules.

paul@laika:~$ cat /proc/modules | grep vm

vmnet 36896 13 - Live 0xffffffff88b21000 (P)
vmmon 194540 0 - Live 0xffffffff88af0000 (P)
paul@laika:~$ lsmod | grep vm
vmnet 36896 13
vmmon 194540 0
paul@laika:~$

124
 Linux Kernel

6.4.6. module dependencies

Some modules depend on others. In the following example, you can see that the nfsd
module is used by exportfs, lockd and sunrpc.

paul@laika:~$ cat /proc/modules | grep nfsd

nfsd 267432 17 - Live 0xffffffff88a40000
exportfs 7808 1 nfsd, Live 0xffffffff88a3d000
lockd 73520 3 nfs,nfsd, Live 0xffffffff88a2a000
sunrpc 185032 12 nfs,nfsd,lockd, Live 0xffffffff889fb000
paul@laika:~$ lsmod | grep nfsd
nfsd 267432 17
exportfs 7808 1 nfsd
lockd 73520 3 nfs,nfsd
sunrpc 185032 12 nfs,nfsd,lockd
paul@laika:~$

6.4.7. insmod
Kernel modules can be manually loaded with the insmod command. This is a very
simple (and obsolete) way of loading modules. The screenshot shows insmod loading
the fat module (for fat file system support).

root@barry:/lib/modules/2.6.17-2-686# pwd
/lib/modules/2.6.17-2-686
root@barry:/lib/modules/2.6.17-2-686# lsmod | grep fat
root@barry:/lib/modules/2.6.17-2-686# insmod kernel/fs/fat/fat.ko
root@barry:/lib/modules/2.6.17-2-686# lsmod | grep fat
fat 46588 0

insmod is not detecting dependencies, so it fails to load the isdn module (because the
isdn module depends on the slhc module).

[root@RHEL52 drivers]# pwd

/lib/modules/2.6.18-92.1.18.el5/kernel/drivers
[root@RHEL52 kernel]# insmod isdn/i4l/isdn.ko
insmod: error inserting 'isdn/i4l/isdn.ko': -1 Unknown symbol in module

6.4.8. modinfo
As you can see in the screenshot of modinfo below, the isdn module depends in the
slhc module.

[root@RHEL52 drivers]# modinfo isdn/i4l/isdn.ko | head -6

filename: isdn/i4l/isdn.ko
license: GPL
author: Fritz Elfert

125
 Linux Kernel

description: ISDN4Linux: link layer

srcversion: 99650346E708173496F6739
depends: slhc

6.4.9. modprobe
The big advantage of modprobe over insmod is that modprobe will load all
necessary modules, whereas insmod requires manual loading of dependencies.
Another advantage is that you don't need to point to the filename with full path.

This screenshot shows how modprobe loads the isdn module, automatically loading
slhc in background.

[root@RHEL52 kernel]# lsmod | grep isdn

[root@RHEL52 kernel]# modprobe isdn
[root@RHEL52 kernel]# lsmod | grep isdn
isdn 122433 0
slhc 10561 1 isdn
[root@RHEL52 kernel]#

6.4.10. /lib/modules/<kernel>/modules.dep
Module dependencies are stored in modules.dep.

[root@RHEL52 2.6.18-92.1.18.el5]# pwd

/lib/modules/2.6.18-92.1.18.el5
[root@RHEL52 2.6.18-92.1.18.el5]# head -3 modules.dep
/lib/modules/2.6.18-92.1.18.el5/kernel/drivers/net/tokenring/3c359.ko:
/lib/modules/2.6.18-92.1.18.el5/kernel/drivers/net/pcmcia/3c574_cs.ko:
/lib/modules/2.6.18-92.1.18.el5/kernel/drivers/net/pcmcia/3c589_cs.ko:

6.4.11. depmod
The modules.dep file can be updated (recreated) with the depmod command. In this
screenshot no modules were added, so depmod generates the same file.

root@barry:/lib/modules/2.6.17-2-686# ls -l modules.dep
-rw-r--r-- 1 root root 310676 2008-03-01 16:32 modules.dep
root@barry:/lib/modules/2.6.17-2-686# depmod
root@barry:/lib/modules/2.6.17-2-686# ls -l modules.dep
-rw-r--r-- 1 root root 310676 2008-12-07 13:54 modules.dep

6.4.12. rmmod
Similar to insmod, the rmmod command is rarely used anymore.

126
 Linux Kernel

[root@RHELv4u3 ~]# modprobe isdn

[root@RHELv4u3 ~]# rmmod slhc
ERROR: Module slhc is in use by isdn
[root@RHELv4u3 ~]# rmmod isdn
[root@RHELv4u3 ~]# rmmod slhc
[root@RHELv4u3 ~]# lsmod | grep isdn
[root@RHELv4u3 ~]#

6.4.13. modprobe -r
Contrary to rmmod, modprobe will automatically remove unneeded modules.

[root@RHELv4u3 ~]# modprobe isdn

[root@RHELv4u3 ~]# lsmod | grep isdn
isdn 133537 0
slhc 7233 1 isdn
[root@RHELv4u3 ~]# modprobe -r isdn
[root@RHELv4u3 ~]# lsmod | grep isdn
[root@RHELv4u3 ~]# lsmod | grep slhc
[root@RHELv4u3 ~]#

6.4.14. /etc/modprobe.conf
The /etc/modprobe.conf file and the /etc/modprobe.d directory can contain aliases
(used by humans) and options (for dependent modules) for modprobe.

[root@RHEL52 ~]# cat /etc/modprobe.conf

alias scsi_hostadapter mptbase
alias scsi_hostadapter1 mptspi
alias scsi_hostadapter2 ata_piix
alias eth0 pcnet32
alias eth2 pcnet32
alias eth1 pcnet32

6.5. compiling a kernel

6.5.1. extraversion
Enter into /usr/src/redhat/BUILD/kernel-2.6.9/linux-2.6.9/ and change the
extraversion in the Makefile.

[root@RHEL52 linux-2.6.18.i686]# pwd

/usr/src/redhat/BUILD/kernel-2.6.18/linux-2.6.18.i686
[root@RHEL52 linux-2.6.18.i686]# vi Makefile
[root@RHEL52 linux-2.6.18.i686]# head -4 Makefile
VERSION = 2
PATCHLEVEL = 6
SUBLEVEL = 18
EXTRAVERSION = -paul2008

127
 Linux Kernel

6.5.2. make mrproper

Now clean up the source from any previous installs with make mrproper. If this is
your first after downloading the source code, then this is not needed.

[root@RHEL52 linux-2.6.18.i686]# make mrproper

CLEAN scripts/basic
CLEAN scripts/kconfig
CLEAN include/config
CLEAN .config .config.old

6.5.3. .config
Now copy a working .config from /boot to our kernel directory. This file contains the
configuration that was used for your current working kernel. It determines whether
modules are included in compilation or not.

[root@RHEL52 linux-2.6.18.i686]# cp /boot/config-2.6.18-92.1.18.el5 .config

6.5.4. make menuconfig

Now run make menuconfig (or the graphical make xconfig). This tool allows you
to select whether to compile stuff as a module (m), as part of the kernel (*), or not
at all (smaller kernel size). If you remove too much, your kernel will not work. The
configuration will be stored in the hidden .config file.

[root@RHEL52 linux-2.6.18.i686]# make menuconfig

6.5.5. make clean

Issue a make clean to prepare the kernel for compile. make clean will remove most
generated files, but keeps your kernel configuration. Running a make mrproper at
this point would destroy the .config file that you built with make menuconfig.

[root@RHEL52 linux-2.6.18.i686]# make clean

6.5.6. make bzImage

And then run make bzImage, sit back and relax while the kernel compiles. You can
use time make bzImage to know how long it takes to compile, so next time you can
go for a short walk.

128
 Linux Kernel

[root@RHEL52 linux-2.6.18.i686]# time make bzImage

HOSTCC scripts/basic/fixdep
HOSTCC scripts/basic/docproc
HOSTCC scripts/kconfig/conf.o
HOSTCC scripts/kconfig/kxgettext.o
...

This command will end with telling you the location of the bzImage file (and with
time info if you also specified the time command.

Kernel: arch/i386/boot/bzImage is ready (#1)

real 13m59.573s
user 1m22.631s
sys 11m51.034s
[root@RHEL52 linux-2.6.18.i686]#

You can already copy this image to /boot with cp arch/i386/boot/bzImage /boot/
vmlinuz-<kernel-version>.

6.5.7. make modules

Now run make modules. It can take 20 to 50 minutes to compile all the modules.

[root@RHEL52 linux-2.6.18.i686]# time make modules

CHK include/linux/version.h
CHK include/linux/utsrelease.h
CC [M] arch/i386/kernel/msr.o
CC [M] arch/i386/kernel/cpuid.o
CC [M] arch/i386/kernel/microcode.o

6.5.8. make modules_install

To copy all the compiled modules to /lib/modules just run make modules_install
(takes about 20 seconds). Here's a screenshot from before the command.

[root@RHEL52 linux-2.6.18.i686]# ls -l /lib/modules/

total 20
drwxr-xr-x 6 root root 4096 Oct 15 13:09 2.6.18-92.1.13.el5
drwxr-xr-x 6 root root 4096 Nov 11 08:51 2.6.18-92.1.17.el5
drwxr-xr-x 6 root root 4096 Dec 6 07:11 2.6.18-92.1.18.el5
[root@RHEL52 linux-2.6.18.i686]# make modules_install

And here is the same directory after. Notice that make modules_install created a
new directory for the new kernel.

[root@RHEL52 linux-2.6.18.i686]# ls -l /lib/modules/

total 24

129
 Linux Kernel

drwxr-xr-x 6 root root 4096 Oct 15 13:09 2.6.18-92.1.13.el5

drwxr-xr-x 6 root root 4096 Nov 11 08:51 2.6.18-92.1.17.el5
drwxr-xr-x 6 root root 4096 Dec 6 07:11 2.6.18-92.1.18.el5
drwxr-xr-x 3 root root 4096 Dec 6 08:50 2.6.18-paul2008

6.5.9. /boot
We still need to copy the kernel, the System.map and our configuration file to /boot.
Strictly speaking the .config file is not obligatory, but it might help you in future
compilations of the kernel.

[root@RHEL52 ]# pwd
/usr/src/redhat/BUILD/kernel-2.6.18/linux-2.6.18.i686
[root@RHEL52 ]# cp System.map /boot/System.map-2.6.18-paul2008
[root@RHEL52 ]# cp .config /boot/config-2.6.18-paul2008
[root@RHEL52 ]# cp arch/i386/boot/bzImage /boot/vmlinuz-2.6.18-paul2008

6.5.10. mkinitrd
The kernel often uses an initrd file at bootup. We can use mkinitrd to generate this
file. Make sure you use the correct kernel name!

[root@RHEL52 ]# pwd
/usr/src/redhat/BUILD/kernel-2.6.18/linux-2.6.18.i686
[root@RHEL52 ]# mkinitrd /boot/initrd-2.6.18-paul2008 2.6.18-paul2008

6.5.11. bootloader
Compilation is now finished, don't forget to create an additional stanza in grub or lilo.

6.6. compiling one module

6.6.1. hello.c
A little C program that will be our module.

[root@rhel4a kernel_module]# cat hello.c

#include <linux/module.h>
#include <section>

int init_module(void)
{
printk(KERN_INFO "Start Hello World...\n");
return 0;
}

130
 Linux Kernel

void cleanup_module(void)
{
printk(KERN_INFO "End Hello World... \n");
}

6.6.2. Makefile
The make file for this module.

[root@rhel4a kernel_module]# cat Makefile

obj-m += hello.o
all:
make -C /lib/modules/$(shell uname -r)/build M=$(PWD) modules
clean:
make -C /lib/modules/$(shell uname -r)/build M=$(PWD) clean

These are the only two files needed.

[root@rhel4a kernel_module]# ll
total 16
-rw-rw-r-- 1 paul paul 250 Feb 15 19:14 hello.c
-rw-rw-r-- 1 paul paul 153 Feb 15 19:15 Makefile

6.6.3. make
The running of the make command.

[root@rhel4a kernel_module]# make

make -C /lib/modules/2.6.9-paul-2/build M=~/kernel_module modules
make[1]: Entering dir... `/usr/src/redhat/BUILD/kernel-2.6.9/linux-2.6.9'
CC [M] /home/paul/kernel_module/hello.o
Building modules, stage 2.
MODPOST
CC /home/paul/kernel_module/hello.mod.o
LD [M] /home/paul/kernel_module/hello.ko
make[1]: Leaving dir... `/usr/src/redhat/BUILD/kernel-2.6.9/linux-2.6.9'
[root@rhel4a kernel_module]#

Now we have more files.

[root@rhel4a kernel_module]# ll
total 172
-rw-rw-r-- 1 paul paul 250 Feb 15 19:14 hello.c
-rw-r--r-- 1 root root 64475 Feb 15 19:15 hello.ko
-rw-r--r-- 1 root root 632 Feb 15 19:15 hello.mod.c
-rw-r--r-- 1 root root 37036 Feb 15 19:15 hello.mod.o
-rw-r--r-- 1 root root 28396 Feb 15 19:15 hello.o
-rw-rw-r-- 1 paul paul 153 Feb 15 19:15 Makefile
[root@rhel4a kernel_module]#

131
 Linux Kernel

6.6.4. hello.ko
Use modinfo to verify that it is really a module.

[root@rhel4a kernel_module]# modinfo hello.ko

filename: hello.ko
vermagic: 2.6.9-paul-2 SMP 686 REGPARM 4KSTACKS gcc-3.4
depends:
[root@rhel4a kernel_module]#

Good, so now we can load our hello module.

[root@rhel4a kernel_module]# lsmod | grep hello

[root@rhel4a kernel_module]# insmod ./hello.ko
[root@rhel4a kernel_module]# lsmod | grep hello
hello 5504 0
[root@rhel4a kernel_module]# tail -1 /var/log/messages
Feb 15 19:16:07 rhel4a kernel: Start Hello World...
[root@rhel4a kernel_module]# rmmod hello
[root@rhel4a kernel_module]#

Finally /var/log/messages has a little surprise.

[root@rhel4a kernel_module]# tail -2 /var/log/messages

Feb 15 19:16:07 rhel4a kernel: Start Hello World...
Feb 15 19:16:35 rhel4a kernel: End Hello World...
[root@rhel4a kernel_module]#

132
Chapter 7. Introduction to network sniffing

Table of Contents
7.1. about sniffing ............................................................................................... 133
7.2. wireshark ...................................................................................................... 133
7.3. tcpdump ........................................................................................................ 135
7.4. practice: network sniffing ............................................................................ 136
7.5. solution: network sniffing ............................................................................ 137

7.1. about sniffing

A good network administrator should be able to use a sniffer like wireshark or
tcpdump to troubleshoot network problems.

A good student will often use a sniffer to learn about networking.

7.2. wireshark

7.2.1. installing wireshark

This example shows how to install wireshark on .deb based distributions like Ubuntu
and Debian.
aptitude install wireshark

On .rpm based distributions you can use yum to install wireshark.

yum install wireshark

7.2.2. selecting interface

When you first fire up wireshark, you will need to select an interface to sniff. You will
see a dialog box that looks similar to this. Choose the interface that you want to sniff.

133
 Introduction to network sniffing

On some distributions only root is allowed to sniff the network. You might need to
use sudo wireshark.

7.2.3. start sniffing

In this example here, we sniffed a ping between two computers. The top pane
shows that wireshark recognizes the icmp protocol, and captured all the ping packets
between the two computers.

7.2.4. looking inside packets

The middle can be expanded. When selecting a line in this panel, you can see the
corresponding bytes in the frame in the bottom panel.

7.2.5. use filters

You might get lost in too many packets. A quick solution to see only the packets that
are of interest to you is to apply filters. When you type arp and click apply, you will
only see arp packets displayed.

You can combine two protocols with a logical or between them. The example below
shows how to filter only arp and bootp (or dhcp) packets.

This example shows how to filter for dns traffic containing a certain ip address.

134
 Introduction to network sniffing

7.3. tcpdump
Sniffing on the command line can be done with tcpdump. Here are some examples.

Using the tcpdump host $ip command displays all traffic with one host
(192.168.1.38 in this example).

root@ubuntu910:~# tcpdump host 192.168.1.38

tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes

Capturing only ssh (tcp port 22) traffic can be done with tcpdump tcp port $port.
This screenshot is cropped to 76 characters for readability in the pdf.

root@deb503:~# tcpdump tcp port 22

tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth1, link-type EN10MB (Ethernet), capture size 96 bytes
14:22:20.716313 IP deb503.local.37973 > rhel53.local.ssh: P 666050963:66605
14:22:20.719936 IP rhel53.local.ssh > deb503.local.37973: P 1:49(48) ack 48
14:22:20.720922 IP rhel53.local.ssh > deb503.local.37973: P 49:113(64) ack
14:22:20.721321 IP rhel53.local.ssh > deb503.local.37973: P 113:161(48) ack
14:22:20.721820 IP deb503.local.37973 > rhel53.local.ssh: . ack 161 win 200
14:22:20.722492 IP rhel53.local.ssh > deb503.local.37973: P 161:225(64) ack
14:22:20.760602 IP deb503.local.37973 > rhel53.local.ssh: . ack 225 win 200
14:22:23.108106 IP deb503.local.54424 > ubuntu910.local.ssh: P 467252637:46
14:22:23.116804 IP ubuntu910.local.ssh > deb503.local.54424: P 1:81(80) ack
14:22:23.116844 IP deb503.local.54424 > ubuntu910.local.ssh: . ack 81 win 2
^C
10 packets captured
10 packets received by filter
0 packets dropped by kernel

Same as above, but write the output to a file with the tcpdump -w $filename
command.

root@ubuntu910:~# tcpdump -w sshdump.tcpdump tcp port 22

tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes
^C
17 packets captured
17 packets received by filter
0 packets dropped by kernel

With tcpdump -r $filename the file created above can be displayed.

root@ubuntu910:~# tcpdump -r sshdump.tcpdump

Many more examples can be found in the manual page of tcpdump.

135
 Introduction to network sniffing

7.4. practice: network sniffing

1. Install wireshark on your computer (not inside a virtual machine).

2. Start a ping between your computer and another computer.

3. Start sniffing the network.

4. Display only the ping echo's in the top pane using a filter.

5. Now ping to a name (like www.linux-training.be) and try to sniff the DNS query
and response. Which DNS server was used ? Was it a tcp or udp query and response ?

136
 Introduction to network sniffing

7.5. solution: network sniffing

1. Install wireshark on your computer (not inside a virtual machine).
Debian/Ubuntu: aptitude install wireshark

Red Hat/Mandriva/Fedora: yum install wireshark

2. Start a ping between your computer and another computer.

ping $ip_address

3. Start sniffing the network.

(sudo) wireshark

select an interface (probably eth0)

4. Display only the ping echo's in the top pane using a filter.
type 'icmp' (without quotes) in the filter box, and then click 'apply'

5. Now ping to a name (like www.linux-training.be) and try to sniff the DNS query
and response. Which DNS server was used ? Was it a tcp or udp query and response ?
First start the sniffer.

Enter 'dns' in the filter box and click apply.

root@ubuntu910:~# ping www.linux-training.be

PING www.linux-training.be (88.151.243.8) 56(84) bytes of data.
64 bytes from fosfor.openminds.be (88.151.243.8): icmp_seq=1 ttl=58 time=14.9 ms
64 bytes from fosfor.openminds.be (88.151.243.8): icmp_seq=2 ttl=58 time=16.0 ms
^C
--- www.linux-training.be ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 14.984/15.539/16.095/0.569 ms

The wireshark screen should look something like this.

The details in wireshark will say the DNS query was inside a udp packet.

137
Chapter 8. Introduction to networking

Table of Contents
8.1. introduction to computer networks .............................................................. 139
8.2. about tcp/ip .................................................................................................. 144
8.3. practice : about tcp/ip ................................................................................... 146
8.4. solution : about tcp/ip .................................................................................. 147
8.5. using tcp/ip ................................................................................................... 148
8.6. practice : using tcp/ip ................................................................................... 154
8.7. solution : using tcp/ip ................................................................................... 155
8.8. multiple ip-addresses .................................................................................... 157
8.9. practice : multiple ip-addresses .................................................................... 157
8.10. solution : multiple ip-addresses .................................................................. 158
8.11. multihomed hosts ....................................................................................... 159
8.12. practice : multihomed hosts ....................................................................... 161
8.13. solution : multihomed hosts ....................................................................... 162
8.14. introduction to iptables .............................................................................. 164
8.15. practice : iptables ....................................................................................... 165
8.16. solution : iptables ....................................................................................... 166
8.17. xinetd and inetd ......................................................................................... 167
8.18. practice : inetd and xinetd .......................................................................... 169
8.19. openssh ....................................................................................................... 170
8.20. practice: ssh ................................................................................................ 174
8.21. network file system .................................................................................... 175
8.22. practice : network file system .................................................................... 177

138
 Introduction to networking

8.1. introduction to computer networks

8.1.1. theory about network layers

seven OSI layers

When talking about protocol layers, people usually mention the seven layers of the
osi protocol (Application, Presentation, Session, Transport, Network, Data Link and
Physical). We will discuss layers 2 and 3 in depth, and focus less on the other layers.
The reason is that these layers are important for understanding networks. You will
hear administrators use words like "this is a layer 2 device" or "this is a layer 3
broadcast", and you should be able to understand what they are talking about.

four DoD layers

The DoD (or tcp/ip) model has only four layers, roughly mapping its network access
layer to OSI layers 1 and 2 (Physical and Datalink), its internet (IP) layer to the
OSI network layer, its host-to-host (tcp, udp) layer to OSI layer 4 (transport) and
its application layer to OSI layers 5, 6 and 7.

Below an attempt to put OSI and DoD layers next to some protocols and devices.

139
 Introduction to networking

short introduction to the physical layer

The physical layer, or layer 1, is all about voltage, electrical signals and mechanical
connections. Some networks might still use coax cables, but most will have migrated
to utp (cat 5 or better) with rj45 connectors.

Devices like repeaters and hubs are part of this layer. You cannot use software to
'see' a repeater or hub on the network. The only thing these devices are doing is
amplifying electrical signals on cables. Passive hubs are multiport amplifiers that
amplify an incoming electrical signal on all other connections. Active hubs do this
by reading and retransmitting bits, without interpreting any meaning in those bits.

Network technologies like csma/cd and token ring are defined on this layer.

This is all we have to say about layer 1 in this book.

short introduction to the data link layer

The data link layer, or layer 2 is about frames. A frame has a crc (cyclic redundancy
check). In the case of ethernet (802.3), each network card is identifiable by a unique
48-bit mac address (media access control address).

On this layer we find devices like bridges and switches. A bridge is more intelligent
than a hub because a bridge can make decisions based on the mac address of
computers. A switch also understands mac addresses.

In this book we will discuss commands like arp and ifconfig to explore this layer.

short introduction to the network layer

Layer 3 is about ip packets. This layer gives every host a unique 32-bit ip address.
But ip is not the only protocol on this layer, there is also icmp, igmp, ipv6 and more.
A complete list can be found in the /etc/protocols file.

On this layer we find devices like routers and layer 3 switches, devices that know
(and have) an ip address.

In tcp/ip this layer is commonly referred to as the internet layer.

short introduction to the transport layer

We will discuss the tcp and udp protocols in the context of layer 4. The DoD model
calls this the host-to-host layer.

short introduction to layers 5, 6 and 7

The tcp/ip application layer includes layers 5, 6 and 7. Details on the difference
between these layers our out of scope of this course.

140
 Introduction to networking

8.1.2. network layers in this book

Stacking of layers in this book is based on the Protocols in Frame explanation in
the wireshark sniffer. When sniffing a dhcp packet, we notice the following in the
sniffer.

[Protocols in Frame: eth:ip:udp:bootp]

Sniffing for ntp (Network Time Protocol) packets gives us this line, which makes us
conclude to put ntp next to bootp in the protocol chart below.

[Protocols in Frame: eth:ip:udp:ntp]

Sniffing an arp broadcast makes us put arp next to ip. All these protocols are
explained later in this chapter.

[Protocols in Frame: eth:arp]

Below is a protocol chart based on wireshark's knowledge. It contains some very

common protocols that are discussed in this book. The chart does not contain all
protocols.

8.1.3. tcp/ip
In the Sixties development of the tcp/ip protocol stack was started by the US
Department of Defense. In the Eighties a lot of commercial enterprises developed
their own protocol stack: IBM created sna, Novell had ipx/spx, Microsoft completed
netbeui and Apple worked with appletalk. All the efforts from the Eighties failed to
survive the Nineties. By the end of the Nineties, almost all computers in the world
were able to speak tcp/ip.

In my humble opinion, the main reason for the survival of tcp/ip over all the other
protocols is its openness. Everyone is free to develop and use the tcp/ip protocol suite.

141
 Introduction to networking

8.1.4. rfc (request for comment)

The protocols that are used on the internet are defined in rfc's. An rfc or request for
comment describes the inner working of all internet protocols. The IETF (Internet
Engineering Task Force) is the sole publisher of these protocols since 1986.

The official website for the rfc's is http://www.rfc-editor.org. This website contains
all rfc's in plain text, for example rfc2132 (which defines dhcp and bootp) is accessible
at http://www.rfc-editor.org/rfc/rfc2132.txt.

8.1.5. lan - man - wan

The term lan is used for local area networks, as opposed to a wan for wide area
networks. The difference between the two is determined by the distance between the
computers, and not by the number of computers in a network. Some protocols like
ATM are designed for use in a wan, others like ethernet are designed for use in a lan.

A man is defined as a metropolitan area network (usually connecting buildings that

are close to each other. A MAN often uses fddi or even ethernet for connectivity.

8.1.6. unicast - multicast - broadcast

A unicast communication originates from one computer and is destined for one
computer. A broadcast is meant for all, and a multicast is for a group of computers.

Careful, a layer 2 broadcast is very different from a layer 3 broadcast. A layer two
broadcast is received by all network cards on the same segment (it does not pass any
router), whereas a layer 3 broadcast is received by all hosts in the same ip subnet.

8.1.7. internet - intranet - extranet

The internet is a global network. It connects many networks using the tcp/ip protocol.

The origin of the internet is the arpanet. The arpanet was created in 1969, that year
only four computers were connected in the network. In 1971 e-mail was invented,
taking 75 percent of all arpanet traffic in 1973. 1973 was the year ftp was introduced,
and also saw the connection of the first European countries (Norway and UK). In
2009 the internet is available to 25 percent of the world population.

An intranet is a private internet. An intranet uses the same protocols as the internet,
but is only accessible to people from within one organization.

An extranet is similar to an intranet, but some trusted organizations (partners/clients/

suppliers/...) also get access.

142
 Introduction to networking

8.1.8. vpn (virtual private network)

A vpn is a network that uses another (usually bigger) network for connectivity.
Typically a vpn is an encrypted tunnel over the internet.

143
 Introduction to networking

8.2. about tcp/ip

8.2.1. overview of tcp/ip v4

The unicast Internet Protocol is one of the oldest network protocols, commonly
used today for lan networks and for the internet. Every host gets a unique 32-bit
ip-address, this is either static or received from a DHCP server. Internet networks
contain several subnets. Those subnets used to be classful (A,B,C,D or E), but
this wasted a lot of address space. Today we work with cidr notation to determine
network id and host id.

In a couple of years we will all be using ipv6! At least, that is what people say since
1999...

8.2.2. internet and routers

The internet is a collection of routers that act as gateways between different
segments. Routers use their routing table to determine the route of tcp/ip packets.
Routers are layer 3 devices, layer 2 contains bridges and switches, layer 1 is cabling
with repeaters and hubs. Layer 2 devices know your 48-bit unique in the world mac
address.

8.2.3. many protocols

For reliable connections, you use tcp, whereas udp is connectionless but faster. The
icmp error messages are used by ping, multicast groups are managed by igmp.

These protocols are visible in the protocol field of the ip header, and are listed in the
/etc/protocols file.

paul@laika:~$ grep tcp /etc/protocols

tcp 6 TCP # transmission control protocol
paul@laika:~$

8.2.4. arp
The ip to mac resolution is handled by the layer two broadcast protocol arp. The
arp table can be displayed with the arp tool.
root@barry:~# arp -a
? (192.168.1.191) at 00:0C:29:3B:15:80 [ether] on eth1
agapi (192.168.1.73) at 00:03:BA:09:7F:D2 [ether] on eth1
anya (192.168.1.1) at 00:12:01:E2:87:FB [ether] on eth1
faith (192.168.1.41) at 00:0E:7F:41:0D:EB [ether] on eth1
kiss (192.168.1.49) at 00:D0:E0:91:79:95 [ether] on eth1
laika (192.168.1.40) at 00:90:F5:4E:AE:17 [ether] on eth1

144
 Introduction to networking

pasha (192.168.1.71) at 00:03:BA:02:C3:82 [ether] on eth1

shaka (192.168.1.72) at 00:03:BA:09:7C:F9 [ether] on eth1
root@barry:~#

Anya is a Cisco Firewall, Faith is an HP Color printer, Kiss is a Kiss DP600, laika is
a Clevo laptop and Agapi, Shaka and Pasha are SPARC servers. The question mark
is a Red Hat Enterprise Linux server running in vmware.

8.2.5. hostname
Every host receives a hostname, usually placed in a DNS name space forming the
fqdn or Fully Qualified Domain Name.

root@rhel6 ~# hostname
rhel6
root@rhel6 ~# hostname --fqdn
rhel6.classroom.local

8.2.6. ip services
Common application level protocols like SMTP, HTTP, SSH, telnet and FTP have
fixed port numbers. To find a port number, look in /etc/services.

paul@laika:~$ grep tftp /etc/services

tftp 69/udp
paul@laika:~$

145
 Introduction to networking

8.3. practice : about tcp/ip

1. Which ports are used by http, pop3, ssh, telnet, nntp and ftp ?

2. Explain why e-mail and websites are sent over tcp and not udp.

3. Display the hostname of your computer.

4. Does your computer have an fqdn ?

5. Which ip-addresses did your computer recently have contact with ?

146
 Introduction to networking

8.4. solution : about tcp/ip

1. Which ports are used by http, pop3, ssh, telnet, nntp and ftp ?
root@rhel6 ~# grep ^'http ' /etc/services
http 80/tcp www www-http # WorldWideWeb HTTP
http 80/udp www www-http # HyperText Transfer Protocol
root@rhel6 ~# grep ^'smtp ' /etc/services
smtp 25/tcp mail
smtp 25/udp mail
root@rhel6 ~# grep ^'ssh ' /etc/services
ssh 22/tcp # The Secure Shell (SSH) Protocol
ssh 22/udp # The Secure Shell (SSH) Protocol
root@rhel6 ~# grep ^'telnet ' /etc/services
telnet 23/tcp
telnet 23/udp
root@rhel6 ~# grep ^'nntp ' /etc/services
nntp 119/tcp readnews untp # USENET News Transfer Protocol
nntp 119/udp readnews untp # USENET News Transfer Protocol
root@rhel6 ~# grep ^'ftp ' /etc/services
ftp 21/tcp
ftp 21/udp fsp fspd

Note that sctp was ommitted from the screenshot.

2. Explain why e-mail and websites are sent over tcp and not udp.
Because tcp is reliable and udp is not.

3. Display the hostname of your computer.

paul@deb503:~$ hostname
deb503

4. Does your computer have an fqdn ?

paul@deb503:~$ hostname --fqdn
deb503

5. Which ip-addresses did your computer recently have contact with ?

root@rhel6 ~# arp -a
? (192.168.1.1) at 00:02:cf:aa:68:f0 [ether] on eth2
? (192.168.1.30) at 00:26:bb:12:7a:5e [ether] on eth2
? (192.168.1.31) at 08:00:27:8e:8a:a8 [ether] on eth2

147
 Introduction to networking

8.5. using tcp/ip

8.5.1. command line or graphical tools ?

If you can, setup your tcp/ip configuration at install time, otherwise use the graphical
tool from your distribution. In the case of RHEL4 and RHEL5, this is the Network
Administration Tool, in RHEL6 it is Network Manager, Novell and OpenSUSE
users can use YaST (Yet another Setup Tool). Avoid mixed use of the GUI tool with
command line or direct editing of network configuration files. You should choose
only one method to manage these files, because some graphical tools will override
your manually edited settings.

Now that we settled this, let's take a look at the files, commands and scripts that
configure your network.

8.5.2. /sbin/ifconfig
You can use the ifconfig command to see the tcp/ip configuration of a network
interface. The first ethernet network card on linux is eth0.
[root@RHEL4b ~]# ifconfig eth0
eth0 Link encap:Ethernet HWaddr 00:0C:29:3B:15:80
inet addr:192.168.1.191 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::20c:29ff:fe3b:1580/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:84 errors:0 dropped:0 overruns:0 frame:0
TX packets:80 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:9216 (9.0 KiB) TX bytes:8895 (8.6 KiB)
Interrupt:185 Base address:0x1400

You can also disable a network interface with ifconfig eth0 down, or enable it with
ifconfig eth0 up. Using these commands does not change the configuration of this
network card.

Every user has access to /sbin/ifconfig, providing the path is set. Normal users cannot
use it to disable or enable interfaces, or set the ip address.

[root@RHEL4b ~]# ifconfig eth0 192.168.1.199

[root@RHEL4b ~]#

The ip address change will be valid until the next change, or until reboot. You can
also supply the subnet mask with ifconfig.

root@laika:~# ifconfig eth0 192.168.1.40 netmask 255.255.255.0

root@laika:~#

Careful, if you try this via an ssh connection, then you might lose your ssh connection.

148
 Introduction to networking

8.5.3. /etc/init.d/network(ing)
If you have a problem with network interfaces, you can try to restart the network init
script, as shown here on Ubuntu 7.04. The script stops and starts the interfaces, and
renews an ip configuration with the DHCP server.

root@laika:~# /etc/init.d/networking restart

* Reconfiguring network interfaces...
There is already a pid file /var/run/dhclient.eth0.pid with pid 14570
killed old client process, removed PID file
Internet Systems Consortium DHCP Client V3.0.4
Copyright 2004-2006 Internet Systems Consortium.
All rights reserved.
For info, please visit http://www.isc.org/sw/dhcp/

Listening on LPF/eth0/00:90:f5:4e:ae:17
Sending on LPF/eth0/00:90:f5:4e:ae:17
Sending on Socket/fallback
DHCPRELEASE on eth0 to 192.168.1.1 port 67
There is already a pid file /var/run/dhclient.eth0.pid with pid 134993416
Internet Systems Consortium DHCP Client V3.0.4
Copyright 2004-2006 Internet Systems Consortium.
All rights reserved.
For info, please visit http://www.isc.org/sw/dhcp/

Listening on LPF/eth0/00:90:f5:4e:ae:17
Sending on LPF/eth0/00:90:f5:4e:ae:17
Sending on Socket/fallback
DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 5
DHCPOFFER from 192.168.1.1
DHCPREQUEST on eth0 to 255.255.255.255 port 67
DHCPACK from 192.168.1.1
bound to 192.168.1.40 -- renewal in 249143 seconds.
root@laika:~#

8.5.4. /etc/sysconfig
Red Hat derived Linux systems store their network configuration files in the /etc/
sysconfig/ directory. Debian derived systems do not have this directory.

/etc/sysconfig/network
Routing and host information for all network interfaces is specified in the /etc/
sysconfig/network file. Below an example, setting 192.168.1.1 as the router (default
gateway), and leaving the default hostname of localhost.localdomain. Common
options not shown in this screenshot are GATEWAYDEV to set one of your network
cards as the gateway device, and NISDOMAIN to specify the NIS domain name.
root@rhel6 ~# cat /etc/sysconfig/network
NETWORKING=yes
HOSTNAME=rhel6

Another screenshot of the same file on RHEL4.

[paul@RHEL4b ~]$ cat /etc/sysconfig/network

149
 Introduction to networking

NETWORKING=yes
HOSTNAME=RHEL4b
GATEWAY=192.168.1.1

/etc/sysconfig/network-scripts
For every network card in your computer, you should have an interface configuration
file named /etc/sysconfig/network-scripts/ifcfg-$IFNAME. Be careful when
editing these files, your edits will work, until you start the system-config-network
(might soon be renamed to redhat-config-network) tool. This tool can and will
overwrite your manual edits.

The first ethernet NIC will get ifcfg-eth0, the next one ifcfg-eth1 and so on. Below
is an example.

paul@RHELv4u2:~$ cat /etc/sysconfig/network-scripts/ifcfg-eth0

DEVICE=eth0
BOOTPROTO=static
BROADCAST=192.168.1.255
HWADDR=00:0C:29:5A:86:D7
IPADDR=192.168.1.222
NETMASK=255.255.255.0
NETWORK=192.168.1.0
ONBOOT=yes
TYPE=Ethernet

When the second nic is configured for dhcp, then this is the ifcfg-eth1.

paul@RHELv4u2:~$ cat /etc/sysconfig/network-scripts/ifcfg-eth1

DEVICE=eth1
BOOTPROTO=dhcp
HWADDR=00:0C:29:6A:34:D8
ONBOOT=yes
TYPE=Ethernet

Besides dhcp and bootp the BOOTPROTO variable can be static or none, both
meaning there should be no protocol used at boot time to set the interface values. The
BROADCAST variable is no longer needed, it will be calculated.

The HWADDR can be used to make sure that the nic's get the correct name when
multiple nic's are present in the computer. It can not be used to set the MAC address of
a nic. For this, you need to specify the MACADDR variable. Do not use HWADDR
and MACADDR in the same ifcfg file.

8.5.5. /sbin/ifup and /sbin/ifdown

The ifup and ifdown commands take an interface as argument and bring it up or
down. The screenshot below deactivates the eth0 network interface.

root@laika:~# ifdown eth0

150
 Introduction to networking

There is already a pid file /var/run/dhclient.eth0.pid with pid 14925

killed old client process, removed PID file
Internet Systems Consortium DHCP Client V3.0.4
Copyright 2004-2006 Internet Systems Consortium.
All rights reserved.
For info, please visit http://www.isc.org/sw/dhcp/

Listening on LPF/eth0/00:90:f5:4e:ae:17
Sending on LPF/eth0/00:90:f5:4e:ae:17
Sending on Socket/fallback
DHCPRELEASE on eth0 to 192.168.1.1 port 67

On debian derived systems, these commands will look at /etc/network/interfaces,

whereas on Red Hat derived systems they will look at /etc/sysconfig/network-scripts/
ifcfg- files. In the screenshot below ifup is used to bring up the eth0 interface. Because
the /etc/network/interfaces file says eth0 uses DHCP, the ifup tool will (try to) start
the dhclient daemon.

root@laika:~# ifup eth0

There is already a pid file /var/run/dhclient.eth0.pid with pid 134993416
Internet Systems Consortium DHCP Client V3.0.4
Copyright 2004-2006 Internet Systems Consortium.
All rights reserved.
For info, please visit http://www.isc.org/sw/dhcp/

Listening on LPF/eth0/00:90:f5:4e:ae:17
Sending on LPF/eth0/00:90:f5:4e:ae:17
Sending on Socket/fallback
DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 8
DHCPOFFER from 192.168.1.1
DHCPREQUEST on eth0 to 255.255.255.255 port 67
DHCPACK from 192.168.1.1
bound to 192.168.1.40 -- renewal in 231552 seconds.
root@laika:~#

8.5.6. /sbin/dhclient
Home and client Linux desktops often have dhclient running. This is a daemon that
enables a network interface to lease an ip configuration from a DHCP server. When
your adapter is configured for DHCP or BOOTP, then /sbin/ifup will start the dhclient
daemon.

8.5.7. /sbin/route
You can see the computer's local routing table with the route command (and also
with netstat -r ).

root@RHEL4b ~]# netstat -r

Kernel IP routing table
Destination Gateway Genmask Flags MSS Window irtt Iface
192.168.1.0 * 255.255.255.0 U 0 0 0 eth0
[root@RHEL4b ~]# route
Kernel IP routing table

151
 Introduction to networking

Destination Gateway Genmask Flags Metric Ref Use Iface

192.168.1.0 * 255.255.255.0 U 0 0 0 eth0
[root@RHEL4b ~]#

It appears this computer does not have a gateway configured, so we use route add
default gw to add a default gateway.

[root@RHEL4b ~]# route add default gw 192.168.1.1

[root@RHEL4b ~]# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.1.0 * 255.255.255.0 U 0 0 0 eth0
default 192.168.1.1 0.0.0.0 UG 0 0 0 eth0
[root@RHEL4b ~]#

8.5.8. ping
If you can ping to another host, then ip is configured.

[root@RHEL4b ~]# ping 192.168.1.5

PING 192.168.1.5 (192.168.1.5) 56(84) bytes of data.
64 bytes from 192.168.1.5: icmp_seq=0 ttl=64 time=1004 ms
64 bytes from 192.168.1.5: icmp_seq=1 ttl=64 time=1.19 ms
64 bytes from 192.168.1.5: icmp_seq=2 ttl=64 time=0.494 ms
64 bytes from 192.168.1.5: icmp_seq=3 ttl=64 time=0.419 ms

--- 192.168.1.5 ping statistics ---

4 packets transmitted, 4 received, 0% packet loss, time 3009ms
rtt min/avg/max/mdev = 0.419/251.574/1004.186/434.520 ms, pipe 2
[root@RHEL4b ~]#

8.5.9. Red Hat network settings backup

It is always a good idea to have a backup of current network settings. The system-
config-network-cmd can do this for you.

root ~# system-config-network-cmd -e > NetworkSettings20070208.txt

And system-config-network-cmd can also be used to restore these settings.

root ~# system-config-network-cmd -i -c < NetworkSettings20070208.txt

For other Linux Systems, take a backup of the relevant portions in /etc.

8.5.10. restarting the network

To stop, start or restart all network interfaces and services, use service network stop|
start|restart. Do not stop the network when connected through ssh.

152
 Introduction to networking

root@rhel55 ~# service network restart

Shutting down interface eth0: [ OK ]
Shutting down interface eth1: [ OK ]
Shutting down interface eth2: [ OK ]
Shutting down loopback interface: [ OK ]
Bringing up loopback interface: [ OK ]
Bringing up interface eth0:
Determining IP information for eth0... done.
[ OK ]
Bringing up interface eth1:
Determining IP information for eth1... done.
[ OK ]
Bringing up interface eth2:
Determining IP information for eth2... done.
[ OK ]

8.5.11. ethtool
To display or change network card settings, use ethtool. The results depend on the
capabilities of your network card. The example shows a network that auto-negotiates
it's bandwidth.

root@laika:~# ethtool eth0

Settings for eth0:
Supported ports: [ TP ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Supports auto-negotiation: Yes
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Advertised auto-negotiation: Yes
Speed: 1000Mb/s
Duplex: Full
Port: Twisted Pair
PHYAD: 0
Transceiver: internal
Auto-negotiation: on
Supports Wake-on: pumbg
Wake-on: g
Current message level: 0x00000033 (51)
Link detected: yes

This example shows how to use ethtool to switch the bandwidth from 1000Mbit to
100Mbit and back. Note that some time passes before the nic is back to 1000Mbit.

root@laika:~# ethtool eth0 | grep Speed

Speed: 1000Mb/s
root@laika:~# ethtool -s eth0 speed 100
root@laika:~# ethtool eth0 | grep Speed
Speed: 100Mb/s
root@laika:~# ethtool -s eth0 speed 1000
root@laika:~# ethtool eth0 | grep Speed
Speed: 1000Mb/s

153
 Introduction to networking

8.6. practice : using tcp/ip

1. Use ifconfig to list all your network interfaces and their ip-addresses. Write down
your ip-address and subnet mask.

2. Use the GUI tool of your distro to set a fix ip address (use the same address as the
one you got from dhcp). Verify with ifconfig and ping to a neighbour that it works.
Also look at the configuration files in /etc/network or /etc/sysconfig to see how the
GUI tool sets a fixed address.

3. Use the GUI tool to enable dhcp again (and verify the changes in the config files).

4. Use ifdown or ifconfig to disable and enable a network card. Verify the results
with ifconfig.

5. Is the dhclient daemon running ?

6. Verify that you have a default gateway with two commands.

7. Ping the default gateway, then look at the mac-address of the default gateway.

154
 Introduction to networking

8.7. solution : using tcp/ip

1. Use ifconfig to list all your network interfaces and their ip-addresses. Write down
your ip-address and subnet mask.
root@rhel6 ~# ifconfig | grep -B1 'inet '
eth1 Link encap:Ethernet HWaddr 08:00:27:81:23:AC
inet addr:192.168.1.33 Bcast:192.168.1.255 Mask:255.255.255.0
--
eth2 Link encap:Ethernet HWaddr 08:00:27:D5:D1:59
inet addr:192.168.1.90 Bcast:192.168.1.255 Mask:255.255.255.0
--
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0

2. Use the GUI tool of your distro to set a fix ip address (use the same address as the
one you got from dhcp). Verify with ifconfig and ping to a neighbour that it works.
Also look at the configuration files in /etc/network or /etc/sysconfig to see how the
GUI tool sets a fixed address.

Rightclick on network manager to open.

ifconfig ethx
ping ip-address neighbour
cat /etc/sysconfig/netwrok-scripts/ifcfg-'press tab-tab'

3. Use the GUI tool to enable dhcp again (and verify the changes in the config files).
root@rhel55 ~# cat /etc/sysconfig/network-scripts/ifcfg-eth2
# Advanced Micro Devices [AMD] 79c970 [PCnet32 LANCE]
DEVICE=eth2
BOOTPROTO=dhcp
ONBOOT=yes
HWADDR=08:00:27:a7:57:46
root@rhel55 ~# cat /etc/sysconfig/network-scripts/ifcfg-eth1
# Advanced Micro Devices [AMD] 79c970 [PCnet32 LANCE]
DEVICE=eth1
BOOTPROTO=dhcp
ONBOOT=yes
HWADDR=08:00:27:40:74:2f

4. Use ifdown or ifconfig to disable and enable a network card. Verify the results
with ifconfig.
root@rhel55 ~# ifconfig eth2 | grep inet
inet addr:192.168.1.35 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::a00:27ff:fea7:5746/64 Scope:Link
root@rhel55 ~# ifdown eth2
root@rhel55 ~# ifconfig eth2 | grep inet
root@rhel55 ~# ifup eth2

Determining IP information for eth2... done.

root@rhel55 ~# ifconfig eth2 | grep inet
inet addr:192.168.1.35 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::a00:27ff:fea7:5746/64 Scope:Link

root@rhel55 ~# ifconfig eth2 | grep -A2 'inet '

inet addr:192.168.1.35 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::a00:27ff:fea7:5746/64 Scope:Link

155
 Introduction to networking

UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1

root@rhel55 ~# ifconfig eth2 down
root@rhel55 ~# ifconfig eth2 | grep -A2 'inet '
inet addr:192.168.1.35 Bcast:192.168.1.255 Mask:255.255.255.0
BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:161 errors:0 dropped:0 overruns:0 frame:0
root@rhel55 ~# ifconfig eth2 up
root@rhel55 ~# ifconfig eth2 | grep -A2 'inet '
inet addr:192.168.1.35 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::a00:27ff:fea7:5746/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1

5. Is the dhclient daemon running ?

root@rhel55 ~# ps -C dhclient
PID TTY TIME CMD
1693 ? 00:00:00 dhclient
3257 ? 00:00:00 dhclient

6. Verify that you have a default gateway with two commands.

root@rhel55 ~# route
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.1.0 * 255.255.255.0 U 0 0 0 eth0
192.168.1.0 * 255.255.255.0 U 0 0 0 eth2
root@rhel55 ~# netstat -nr
Kernel IP routing table
Destination Gateway Genmask Flags MSS Window irtt Iface
192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth2

7. Ping the default gateway, then look at the mac-address of the default gateway.
[email protected]:~$ ping -c1 192.168.1.1
PING 192.168.1.1 (192.168.1.1) 56(84) bytes of data.
64 bytes from 192.168.1.1: icmp_seq=1 ttl=254 time=4.77 ms

--- 192.168.1.1 ping statistics ---

1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 4.770/4.770/4.770/0.000 ms
[email protected]:~$ arp -a | grep 192.168.1.1
illyria (192.168.1.1) at 00:02:cf:aa:68:f0 [ether] on eth1

156
 Introduction to networking

8.8. multiple ip-addresses

8.8.1. binding multiple ip-addresses

To bind more than one ip-address to the same interface, use ifcfg-eth0:0, where the
last zero can be anything else. Only two directives are required in the file.

root@RHELv4u2:/etc/sysconfig/network-scripts# cat ifcfg-eth0:0

DEVICE=eth0:0
IPADDR=192.168.1.232

8.8.2. enabling extra ip-addresses

To activate a virtual network interface, use ifup, to deactivate it, use ifdown.

root@RHELv4u2:~# ifdown eth0:0

root@RHELv4u2:~# ifup eth0:0

8.9. practice : multiple ip-addresses

1. Add an extra ip-address to one of your network cards. Test that it works (have
your neighbour ssh to it)!

2. Use ifdown to disable this extra ip-address.

157
 Introduction to networking

8.10. solution : multiple ip-addresses

1. Add an extra ip-address to one of your network cards. Test that it works (have
your neighbour ssh to it)!
root@rhel55 network-scripts# cat ifcfg-eth0:0
DEVICE=eth0:0
IPADDR=192.168.1.99
root@rhel55 network-scripts# ifup eth0:0
root@rhel55 network-scripts# ifconfig | head -15
eth0 Link encap:Ethernet HWaddr 08:00:27:80:22:66
inet addr:192.168.1.34 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::a00:27ff:fe80:2266/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:2529 errors:0 dropped:0 overruns:0 frame:0
TX packets:1755 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:241174 (235.5 KiB) TX bytes:236415 (230.8 KiB)
Interrupt:10 Base address:0xd020

eth0:0 Link encap:Ethernet HWaddr 08:00:27:80:22:66

inet addr:192.168.1.99 Bcast:192.168.1.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
Interrupt:10 Base address:0xd020

2. Use ifdown to disable this extra ip-address.

root@rhel55 network-scripts# ifdown eth0:0

158
 Introduction to networking

8.11. multihomed hosts

8.11.1. bonding
You can combine (bond) two physical network interfaces as one logical interface.
Having two network cards serve the same ip-address doubles the bandwidth, and
provides hardware redundancy. For bonding to work, you have to load the kernel
module for bonding. You can do this manually with modprobe.

root@RHELv4u2:~# modprobe bonding

root@RHELv4u2:~# lsmod | grep bon
bonding 58984 0

Or automatically, by adding the alias to /etc/modprobe.conf (used to be called /etc/

modules.conf).

root@RHELv4u2:~# echo alias bond0 bonding >> /etc/modprobe.conf

You need two network cards to enable bonding, and add the MASTER and SLAVE
variables. In this case we used eth0 and eth1, configured like this.

root@RHELv4u2:~# cat /etc/sysconfig/network-scripts/ifcfg-eth0

DEVICE=eth0
BROADCAST=192.168.1.255
HWADDR=00:0C:29:5A:86:D7
IPADDR=192.168.1.222
NETMASK=255.255.255.0
NETWORK=192.168.1.0
ONBOOT=yes
TYPE=Ethernet
GATEWAY=192.168.1.1
MASTER=bond0
SLAVE=yes
USERCTL=no
root@RHELv4u2:~# cat /etc/sysconfig/network-scripts/ifcfg-eth1
DEVICE=eth1
BROADCAST=192.168.1.255
HWADDR=00:0C:29:5A:86:E1
IPADDR=192.168.1.232
NETMASK=255.255.255.0
NETWORK=192.168.1.0
ONBOOT=yes
TYPE=Ethernet
GATEWAY=192.168.1.1
MASTER=bond0
SLAVE=yes
USERCTL=no
root@RHELv4u2:~#

And you need to set up a bonding interface. In this case, we call it bond0.

159
 Introduction to networking

root@RHELv4u2:~# cat /etc/sysconfig/network-scripts/ifcfg-bond0

DEVICE=bond0
BOOTPROTO=none
ONBOOT=no
NETWORK=192.168.1.0
NETMASK=255.255.255.0
IPADDR=192.168.1.229
USERCTL=no
root@RHELv4u2:~#

To bring up the interface, just use the ifup bond0 command.

root@RHELv4u2:/etc/sysconfig/network-scripts# ifup bond0

Enslaving eth0 to bond0
Enslaving eth1 to bond0
root@RHELv4u2:~#

The ifconfig command will show you all activated interfaces.

root@RHELv4u2:~# ifconfig
bond0 Link encap:Ethernet HWaddr 00:0C:29:5A:86:D7
inet addr:192.168.1.229 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::200:ff:fe00:0/64 Scope:Link
UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
RX packets:3835 errors:0 dropped:0 overruns:0 frame:0
TX packets:1001 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:469645 (458.6 KiB) TX bytes:139816 (136.5 KiB)

eth0 Link encap:Ethernet HWaddr 00:0C:29:5A:86:D7

inet6 addr: fe80::20c:29ff:fe5a:86d7/64 Scope:Link
UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
RX packets:3452 errors:0 dropped:0 overruns:0 frame:0
TX packets:837 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:412155 (402.4 KiB) TX bytes:117844 (115.0 KiB)
Interrupt:11 Base address:0x1400

eth1 Link encap:Ethernet HWaddr 00:0C:29:5A:86:D7

inet6 addr: fe80::20c:29ff:fe5a:86d7/64 Scope:Link
UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
RX packets:392 errors:0 dropped:0 overruns:0 frame:0
TX packets:177 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:58084 (56.7 KiB) TX bytes:24078 (23.5 KiB)
Interrupt:10 Base address:0x1480

8.11.2. /proc/net/bond*
You can verify the proper working of the bonding interfaces by looking at /proc/net/
bonding/. Below is a screenshot of a Red Hat Enterprise 5 server, with eth1 and eth2
in bonding.

[root@RHEL5 ~]# cat /proc/net/bonding/bond0

Ethernet Channel Bonding Driver: v3.1.2 (January 20, 2007)

160
 Introduction to networking

Bonding Mode: load balancing (round-robin)

MII Status: up
MII Polling Interval (ms): 0
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth1

MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:0c:29:a0:9d:e3

Slave Interface: eth2

MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:0c:29:a0:9d:ed
[root@RHEL5 ~]#

8.12. practice : multihomed hosts

1. Add two network cards to the virtual machine, and bond the two cards as one virtual
(double bandwidth and failover) card.

161
 Introduction to networking

8.13. solution : multihomed hosts

1. Add two network cards to the virtual machine, and bond the two cards as one virtual
(double bandwidth and failover) card.

Here is a succinct version of the config files to bond eth1 and eth2 on RHEL5.
root@rhel55 network-scripts# cat ifcfg-eth1
DEVICE=eth1
ONBOOT=yes
HWADDR=08:00:27:40:74:2f
MASTER=bond0
SLAVE=yes
USERCTL=no
root@rhel55 network-scripts# cat ifcfg-eth2
DEVICE=eth2
ONBOOT=yes
HWADDR=08:00:27:a7:57:46
MASTER=bond0
SLAVE=yes
USERCTL=no

ifcfg-bond0 can also be succinct.

root@rhel55 network-scripts# cat ifcfg-bond0
DEVICE=bond0
BOOTPROTO=none
ONBOOT=no
IPADDR=192.168.1.70

You might need modprobe.

root@rhel55 network-scripts# modprobe bonding

Then bring the interface up.

root@rhel55 network-scripts# ifup bond0
root@rhel55 network-scripts# ifconfig
bond0 Link encap:Ethernet HWaddr 08:00:27:40:74:2F
inet addr:192.168.1.70 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::a00:27ff:fe40:742f/64 Scope:Link
UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
RX packets:2856 errors:2 dropped:0 overruns:0 frame:0
TX packets:951 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:254917 (248.9 KiB) TX bytes:106473 (103.9 KiB)

eth0 Link encap:Ethernet HWaddr 08:00:27:80:22:66

inet addr:192.168.1.29 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::a00:27ff:fe80:2266/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:4763 errors:0 dropped:0 overruns:0 frame:0
TX packets:3287 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:451280 (440.7 KiB) TX bytes:476878 (465.7 KiB)
Interrupt:10 Base address:0xd020

eth1 Link encap:Ethernet HWaddr 08:00:27:40:74:2F

UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
RX packets:1142 errors:1 dropped:0 overruns:0 frame:0
TX packets:153 errors:0 dropped:0 overruns:0 carrier:0

162
 Introduction to networking

collisions:0 txqueuelen:1000
RX bytes:95248 (93.0 KiB) TX bytes:26754 (26.1 KiB)
Interrupt:9 Base address:0xd240

eth2 Link encap:Ethernet HWaddr 08:00:27:40:74:2F

UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
RX packets:1714 errors:1 dropped:0 overruns:0 frame:0
TX packets:798 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:159669 (155.9 KiB) TX bytes:79719 (77.8 KiB)
Interrupt:5 Base address:0xd260

lo Link encap:Local Loopback

inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:204 errors:0 dropped:0 overruns:0 frame:0
TX packets:204 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:21310 (20.8 KiB) TX bytes:21310 (20.8 KiB)

163
 Introduction to networking

8.14. introduction to iptables

8.14.1. iptables firewall

The Linux kernel has a built-in stateful firewall named iptables. To stop the iptables
firewall on Red Hat, use the service command.

root@RHELv4u4:~# service iptables stop

Flushing firewall rules: [ OK ]
Setting chains to policy ACCEPT: filter [ OK ]
Unloading iptables modules: [ OK ]
root@RHELv4u4:~#

The easy way to configure iptables, is to use a graphical tool like KDE's kmyfirewall
or Security Level Configuration Tool. You can find the latter in the graphical menu,
somewhere in System Tools - Security, or you can start it by typing system-config-
securitylevel in bash. These tools allow for some basic firewall configuration. You
can decide whether to enable or disable the firewall, and what typical standard ports
are allowed when the firewall is active. You can even add some custom ports. When
you are done, the configuration is written to /etc/sysconfig/iptables on Red Hat.

root@RHELv4u4:~# cat /etc/sysconfig/iptables

# Firewall configuration written by system-config-securitylevel
# Manual customization of this file is not recommended.
*filter
:INPUT ACCEPT [0:0]
:FORWARD ACCEPT [0:0]
:OUTPUT ACCEPT [0:0]
:RH-Firewall-1-INPUT - [0:0]
-A INPUT -j RH-Firewall-1-INPUT
-A FORWARD -j RH-Firewall-1-INPUT
-A RH-Firewall-1-INPUT -i lo -j ACCEPT
-A RH-Firewall-1-INPUT -p icmp --icmp-type any -j ACCEPT
-A RH-Firewall-1-INPUT -p 50 -j ACCEPT
-A RH-Firewall-1-INPUT -p 51 -j ACCEPT
-A RH-Firewall-1-INPUT -p udp --dport 5353 -d 224.0.0.251 -j ACCEPT
-A RH-Firewall-1-INPUT -p udp -m udp --dport 631 -j ACCEPT
-A RH-Firewall-1-INPUT -m state --state ESTABLISHED,RELATED -j ACCEPT
-A RH-F...NPUT -m state --state NEW -m tcp -p tcp --dport 22 -j ACCEPT
-A RH-F...NPUT -m state --state NEW -m tcp -p tcp --dport 80 -j ACCEPT
-A RH-F...NPUT -m state --state NEW -m tcp -p tcp --dport 21 -j ACCEPT
-A RH-F...NPUT -m state --state NEW -m tcp -p tcp --dport 25 -j ACCEPT
-A RH-Firewall-1-INPUT -j REJECT --reject-with icmp-host-prohibited
COMMIT
root@RHELv4u4:~#

To start the service, issue the service iptables start command. You can configure
iptables to start at boot time with chkconfig.

root@RHELv4u4:~# service iptables start

Applying iptables firewall rules: [ OK ]
root@RHELv4u4:~# chkconfig iptables on
root@RHELv4u4:~#

164
 Introduction to networking

One of the nice features of iptables is that it displays extensive status information
when queried with the service iptables status command.

root@RHELv4u4:~# service iptables status

Table: filter
Chain INPUT (policy ACCEPT)
target prot opt source destination
RH-Firewall-1-INPUT all -- 0.0.0.0/0 0.0.0.0/0

Chain FORWARD (policy ACCEPT)

target prot opt source destination
RH-Firewall-1-INPUT all -- 0.0.0.0/0 0.0.0.0/0

Chain OUTPUT (policy ACCEPT)

target prot opt source destination

Chain RH-Firewall-1-INPUT (2 references)

target prot opt source destination
ACCEPT all -- 0.0.0.0/0 0.0.0.0/0
ACCEPT icmp -- 0.0.0.0/0 0.0.0.0/0 icmp type 255
ACCEPT esp -- 0.0.0.0/0 0.0.0.0/0
ACCEPT ah -- 0.0.0.0/0 0.0.0.0/0
ACCEPT udp -- 0.0.0.0/0 224.0.0.251 udp dpt:5353
ACCEPT udp -- 0.0.0.0/0 0.0.0.0/0 udp dpt:631
ACCEPT all -- 0.0.0.0/0 0.0.0.0/0 state RELATED,ESTABLISHED
ACCEPT tcp -- 0.0.0.0/0 0.0.0.0/0 state NEW tcp dpt:22
ACCEPT tcp -- 0.0.0.0/0 0.0.0.0/0 state NEW tcp dpt:80
ACCEPT tcp -- 0.0.0.0/0 0.0.0.0/0 state NEW tcp dpt:21
ACCEPT tcp -- 0.0.0.0/0 0.0.0.0/0 state NEW tcp dpt:25
REJECT all -- 0.0.0.0/0 0.0.0.0/0 reject-with icmp-host-prohibited

root@RHELv4u4:~#

Mastering firewall configuration requires a decent knowledge of tcp/ip. Good

iptables tutorials can be found online here http://iptables-tutorial.frozentux.net/
iptables-tutorial.html and here http://tldp.org/HOWTO/IP-Masquerade-HOWTO/.

8.15. practice : iptables

1. Verify whether the firewall is running.

2. Stop the running firewall.

165
 Introduction to networking

8.16. solution : iptables

1. Verify whether the firewall is running.
root@rhel55 ~# service iptables status | head
Table: filter
Chain INPUT (policy ACCEPT)
num target prot opt source destination
1 RH-Firewall-1-INPUT all -- 0.0.0.0/0 0.0.0.0/0

Chain FORWARD (policy ACCEPT)

num target prot opt source destination
1 RH-Firewall-1-INPUT all -- 0.0.0.0/0 0.0.0.0/0

Chain OUTPUT (policy ACCEPT)

2. Stop the running firewall.

root@rhel55 ~# service iptables stop
Flushing firewall rules: [ OK ]
Setting chains to policy ACCEPT: filter [ OK ]
Unloading iptables modules: [ OK ]
root@rhel55 ~# service iptables status
Firewall is stopped.

166
 Introduction to networking

8.17. xinetd and inetd

8.17.1. the superdaemon

Back when resources like RAM memory were limited, a super-server was devised
to listen to all sockets and start the appropriate daemon only when needed. Services
like swat, telnet and ftp are typically served by such a super-server. The xinetd
superdaemon is more recent than inetd. We will discuss the configuration both
daemons.

Recent Linux distributions like RHEL5 and Ubuntu10.04 do not activate inetd or
xinetd by default, unless an application requires it.

8.17.2. inetd or xinetd

First verify whether your computer is running inetd or xinetd. This Debian 4.0 Etch
is running inetd.

root@barry:~# ps fax | grep inet

3870 ? Ss 0:00 /usr/sbin/inetd

This Red Hat Enterprise Linux 4 update 4 is running xinetd.

[root@RHEL4b ~]# ps fax | grep inet

3003 ? Ss 0:00 xinetd -stayalive -pidfile /var/run/xinetd.pid

Both daemons have the same functionality (listening to many ports, starting other
daemons when they are needed), but they have different configuration files.

8.17.3. xinetd superdaemon

The xinetd daemon is often called a superdaemon because it listens to a lot of
incoming connections, and starts other daemons when they are needed. When a
connection request is received, xinetd will first check TCP wrappers (/etc/hosts.allow
and /etc/hosts.deny) and then give control of the connection to the other daemon. This
superdaemon is configured through /etc/xinetd.conf and the files in the directory /
etc/xinetd.d. Let's first take a look at /etc/xinetd.conf.

paul@RHELv4u2:~$ cat /etc/xinetd.conf

#
# Simple configuration file for xinetd
#
# Some defaults, and include /etc/xinetd.d/

defaults
{

167
 Introduction to networking

instances = 60
log_type = SYSLOG authpriv
log_on_success = HOST PID
log_on_failure = HOST
cps = 25 30
}

includedir /etc/xinetd.d

paul@RHELv4u2:~$

According to the settings in this file, xinetd can handle 60 client requests at once. It
uses the authpriv facility to log the host ip-address and pid of successful daemon
spawns. When a service (aka protocol linked to daemon) gets more than 25 cps
(connections per second), it holds subsequent requests for 30 seconds.

The directory /etc/xinetd.d contains more specific configuration files. Let's also take
a look at one of them.

paul@RHELv4u2:~$ ls /etc/xinetd.d
amanda chargen-udp echo klogin rexec talk
amandaidx cups-lpd echo-udp krb5-telnet rlogin telnet
amidxtape daytime eklogin kshell rsh tftp
auth daytime-udp finger ktalk rsync time
chargen dbskkd-cdb gssftp ntalk swat time-udp
paul@RHELv4u2:~$ cat /etc/xinetd.d/swat
# default: off
# description: SWAT is the Samba Web Admin Tool. Use swat \
# to configure your Samba server. To use SWAT, \
# connect to port 901 with your favorite web browser.
service swat
{
port = 901
socket_type = stream
wait = no
only_from = 127.0.0.1
user = root
server = /usr/sbin/swat
log_on_failure += USERID
disable = yes
}
paul@RHELv4u2:~$

The services should be listed in the /etc/services file. Port determines the service port,
and must be the same as the port specified in /etc/services. The socket_type should be
set to stream for tcp services (and to dgram for udp). The log_on_failure += concats
the userid to the log message formatted in /etc/xinetd.conf. The last setting disable
can be set to yes or no. Setting this to no means the service is enabled!

Check the xinetd and xinetd.conf manual pages for many more configuration options.

8.17.4. inetd superdaemon

This superdaemon has only one configuration file /etc/inetd.conf. Every protocol or
daemon that it is listening for, gets one line in this file.

168
 Introduction to networking

root@barry:~# grep ftp /etc/inetd.conf

tftp dgram udp wait nobody /usr/sbin/tcpd /usr/sbin/in.tftpd /boot/tftp
root@barry:~#

You can disable a service in inetd.conf above by putting a # at the start of that line.
Here an example of the disabled vmware web interface (listening on tcp port 902).

paul@laika:~$ grep vmware /etc/inetd.conf

#902 stream tcp nowait root /usr/sbin/vmware-authd vmware-authd

8.18. practice : inetd and xinetd

1. Verify on all systems whether they are using xinetd or inetd.

2. Look at the configuration files.

3. (If telnet is installable, then replace swat in these questions with telnet) Is
swat installed ? If not, then install swat and look at the changes in the (x)inetd
configuration. Is swat enabled or disabled ?

4. Disable swat, test it. Enable swat, test it.

169
 Introduction to networking

8.19. openssh

8.19.1. secure shell

Avoid using telnet, rlogin and rsh to remotely connect to your servers. These
older protocols do not encrypt the login session, which means your user id and
password can be sniffed by tools like wireshark or tcpdump. To securely connect
to your servers, use ssh. An ssh connection always starts with a cryptographic
handshake, followed by encryption of the transport layer using a symmetric cypher.
Then authentication takes place (using user id/password or public/private keys) and
communication can take place over the encrypted connection. In other words, the
tunnel is encrypted before you start typing anything.

The openssh package is maintained by the OpenBSD people and is distributed with
a lot of operating systems (it may even be the most popular package in the world).
Below sample use of ssh to connect from one server (RHELv4u2) to another one
(RHELv4u4).

paul@RHELv4u2:~$ ssh 192.168.1.220

The authenticity of host '192.168.1.220' can't be established.
RSA key fingerprint is c4:3c:52:e6:d8:8b:ce:17:8b:c9:78:5a:f3:51:06:4f.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '192.168.1.220' (RSA) to the list of known...
[email protected]'s password:
Last login: Sun Jan 21 07:16:26 2007 from 192.168.1.40
paul@RHELv4u4:~$

The second time ssh remembers the connection. It added an entry to the ~/.ssh/
known_hosts file.

paul@RHELv4u2:~$ ssh 192.168.1.220

[email protected]'s password:
Last login: Sun Jan 21 08:49:19 2007 from 192.168.1.222
paul@RHELv4u4:~$

8.19.2. ssh protocol versions

The ssh protocol has two versions (1 and 2). Avoid using version 1 anywhere, since
it contains some known vulnerabilities. You can control the protocol version via /etc/
ssh/ssh_config for the client side and /etc/ssh/sshd_config for the openssh-server
daemon.

root@laika:/etc/ssh# grep Protocol ssh_config

# Protocol 2,1
root@laika:/etc/ssh# grep Protocol sshd_config
Protocol 2
root@laika:/etc/ssh#

170
 Introduction to networking

Configuration of ssh is done in the /etc/ssh directory and is pretty straightforward.

8.19.3. public and private keys

Imagine Alice and Bob, two people that like to communicate with each other.
Using public and private keys they can communicate with encryption and with
authentication.

When Alice wants to send an encrypted message to Bob, she uses the public key of
Bob. Bob shares his public key with Alice, but keeps his private key private! Since
Bob is the only one to have Bob's private key, Alice is sure that Bob is the only one
that can read the encrypted message.

When Bob wants to verify that the message came from Alice, Bob uses the public
key of Alice to verify that Alice signed the message with her private key. Since Alice
is the only one to have Alice's private key, Bob is sure the message came from Alice.

8.19.4. setting up passwordless ssh

To set up passwordless ssh authentication through public/private keys, use ssh-
keygen to generate a key pair without a passphrase, and then copy your public key
to the destination server. Let's do this step by step.

In the example that follows, we will set up ssh without password between Alice and
Bob. Alice has an account on a Red Hat Enterprise Linux server, Bob is using Ubuntu
on his laptop. Bob wants to give Alice access using ssh and the public and private
key system. This means that even if Bob changes his password on his laptop, Alice
will still have access.

ssh-keygen
The example below shows how Alice uses ssh-keygen to generate a key pair. Alice
does not enter a passphrase.

[alice@RHEL5 ~]$ ssh-keygen -t rsa

Generating public/private rsa key pair.
Enter file in which to save the key (/home/alice/.ssh/id_rsa):
Created directory '/home/alice/.ssh'.
Enter passphrase (empty for no passphrase):
Enter same passphrase again:
Your identification has been saved in /home/alice/.ssh/id_rsa.
Your public key has been saved in /home/alice/.ssh/id_rsa.pub.
The key fingerprint is:
9b:ac:ac:56:c2:98:e5:d9:18:c4:2a:51:72:bb:45:eb alice@RHEL5
[alice@RHEL5 ~]$

171
 Introduction to networking

~/.ssh
While ssh-keygen generates a public and a private key, it will also create a hidden
.ssh directory with proper permissions. If you create the .ssh directory manually, then
you need to chmod 700 it! Otherwise ssh will refuse to use the keys (world readable
private keys are not secure!).

As you can see, the .ssh directory is secure in Alice's home directory.

[alice@RHEL5 ~]$ ls -ld .ssh

drwx------ 2 alice alice 4096 May 1 07:38 .ssh
[alice@RHEL5 ~]$

Bob is using Ubuntu at home. He decides to manually create the .ssh directory, so
he needs to manually secure it.

bob@laika:~$ mkdir .ssh

bob@laika:~$ ls -ld .ssh
drwxr-xr-x 2 bob bob 4096 2008-05-14 16:53 .ssh
bob@laika:~$ chmod 700 .ssh/
bob@laika:~$

id_rsa and id_rsa.pub

The ssh-keygen command generate two keys in .ssh. The public key is named ~/.ssh/
id_rsa.pub. The private key is named ~/.ssh/id_rsa.

[alice@RHEL5 ~]$ ls -l .ssh/

total 16
-rw------- 1 alice alice 1671 May 1 07:38 id_rsa
-rw-r--r-- 1 alice alice 393 May 1 07:38 id_rsa.pub
[alice@RHEL5 ~]$

scp
To copy the public key from Alice's server tot Bob's laptop, Alice decides to use scp.

[alice@RHEL5 .ssh]$ scp id_rsa.pub [email protected]:~/.ssh/authorized_keys

[email protected]'s password:
id_rsa.pub 100% 393 0.4KB/s 00:00
[alice@RHEL5 .ssh]$

Be careful when copying a second key! Do not overwrite the first key, instead append
the key to the same ~/.ssh/authorized_keys file!

authorized_keys
In your ~/.ssh directory, you can create a file called authorized_keys. This file can
contain one or more public keys from people you trust. Those trusted people can

172
 Introduction to networking

use their private keys to prove their identity and gain access to your account via ssh
(without password). The example shows Bob's authorized_keys file containing the
public key of Alice.

bob@laika:~$ cat .ssh/authorized_keys

ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAQEApCQ9xzyLzJes1sR+hPyqW2vyzt1D4zTLqk\
MDWBR4mMFuUZD/O583I3Lg/Q+JIq0RSksNzaL/BNLDou1jMpBe2Dmf/u22u4KmqlJBfDhe\
yTmGSBzeNYCYRSMq78CT9l9a+y6x/shucwhaILsy8A2XfJ9VCggkVtu7XlWFDL2cum08/0\
mRFwVrfc/uPsAn5XkkTscl4g21mQbnp9wJC40pGSJXXMuFOk8MgCb5ieSnpKFniAKM+tEo\
/vjDGSi3F/bxu691jscrU0VUdIoOSo98HUfEf7jKBRikxGAC7I4HLa+/zX73OIvRFAb2hv\
tUhn6RHrBtUJUjbSGiYeFTLDfcTQ== alice@RHEL5
bob@laika:~$

passwordless ssh
Alice can now use ssh to connect passwordless to Bob's laptop. In combination with
ssh's capability to execute commands on the remote host, this can be useful in pipes
across different machines.

[alice@RHEL5 ~]$ ssh [email protected] "ls -l .ssh"

total 4
-rw-r--r-- 1 bob bob 393 2008-05-14 17:03 authorized_keys
[alice@RHEL5 ~]$

8.19.5. X forwarding via SSH

The ssh protocol will remember the servers it connected to (and warn you in case
something suspicious happened), and will use strong 128-bit encryption. Another
popular feature of ssh is called X11 forwarding and is implemented with ssh -X.

Below an example of X11 forwarding: user paul logs in as user greet on her computer
to start the graphical application mozilla-thunderbird. Although the application will
run on the remote computer from greet, it will be displayed on the screen attached
locally to paul's computer.

paul@laika:~/PDF$ ssh -X [email protected] -p 55555

Warning: Permanently added the RSA host key for IP address \
'81.240.174.161' to the list of known hosts.
Password:
Linux raika 2.6.8-2-686 #1 Tue Aug 16 13:22:48 UTC 2005 i686 GNU/Linux

Last login: Thu Jan 18 12:35:56 2007

greet@raika:~$ ps fax | grep thun
greet@raika:~$ mozilla-thunderbird &
[1] 30336

8.19.6. troubleshooting ssh

Use ssh -v to get debug information about the ssh connection attempt.

173
 Introduction to networking

paul@laika:~$ ssh -v [email protected]

OpenSSH_4.3p2 Debian-8ubuntu1, OpenSSL 0.9.8c 05 Sep 2006
debug1: Reading configuration data /home/paul/.ssh/config
debug1: Reading configuration data /etc/ssh/ssh_config
debug1: Applying options for *
debug1: Connecting to 192.168.1.192 [192.168.1.192] port 22.
debug1: Connection established.
debug1: identity file /home/paul/.ssh/identity type -1
debug1: identity file /home/paul/.ssh/id_rsa type 1
debug1: identity file /home/paul/.ssh/id_dsa type -1
debug1: Remote protocol version 1.99, remote software version OpenSSH_3
debug1: match: OpenSSH_3.9p1 pat OpenSSH_3.*
debug1: Enabling compatibility mode for protocol 2.0
...

8.20. practice: ssh

1. Create a user for your neighbour, then test ssh to your neighbour (by ip-address or
by hostname). (You might need to install the openssh-server with aptitude.)

2. Create a bookmark in Firefox, then close your firefox! Use ssh -X to run firefox
on your screen, but on your neighbour's computer. Do you see your neighbour's
bookmark ?

3. Verify in the ssh configuration files that only protocol version 2 is allowed.

4. Use ssh-keygen to create a key pair without passphrase. Setup passwordless ssh
between you and your neighbour. (or between the ubuntu and the Red Hat)

174
 Introduction to networking

8.21. network file system

8.21.1. protocol versions

The older nfs versions 2 and 3 are stateless (udp) by default, but they can use tcp.
Clients connect to the server using rpc (on Linux this is controlled by the portmap
daemon. Look at rpcinfo to verify that nfs and its related services are running.

root@RHELv4u2:~# /etc/init.d/portmap status

portmap (pid 1920) is running...
root@RHELv4u2:~# rpcinfo -p
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 32768 status
100024 1 tcp 32769 status
root@RHELv4u2:~# service nfs start
Starting NFS services: [ OK ]
Starting NFS quotas: [ OK ]
Starting NFS daemon: [ OK ]
Starting NFS mountd: [ OK ]

The same rpcinfo command when nfs is started.

root@RHELv4u2:~# rpcinfo -p
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 32768 status
100024 1 tcp 32769 status
100011 1 udp 985 rquotad
100011 2 udp 985 rquotad
100011 1 tcp 988 rquotad
100011 2 tcp 988 rquotad
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
100003 4 udp 2049 nfs
100003 2 tcp 2049 nfs
100003 3 tcp 2049 nfs
100003 4 tcp 2049 nfs
100021 1 udp 32770 nlockmgr
100021 3 udp 32770 nlockmgr
100021 4 udp 32770 nlockmgr
100021 1 tcp 32789 nlockmgr
100021 3 tcp 32789 nlockmgr
100021 4 tcp 32789 nlockmgr
100005 1 udp 1004 mountd
100005 1 tcp 1007 mountd
100005 2 udp 1004 mountd
100005 2 tcp 1007 mountd
100005 3 udp 1004 mountd
100005 3 tcp 1007 mountd
root@RHELv4u2:~#

175
 Introduction to networking

nfs version 4 requires tcp (port 2049) and supports Kerberos user authentication as
an option. nfs authentication only takes place when mounting the share. nfs versions
2 and 3 authenticate only the host.

8.21.2. server configuration

nfs is configured in /etc/exports. Here is a sample /etc/exports to explain the syntax.
You need some way (NIS domain or LDAP) to synchronize userid's across computers
when using nfs a lot. The rootsquash option will change UID 0 to the UID of the
nfsnobody user account. The sync option will write writes to disk before completing
the client request.

paul@laika:~$ cat /etc/exports

# Everyone can read this share
/mnt/data/iso *(ro)

# Only the computers barry and pasha can readwrite this one
/var/www pasha(rw) barry(rw)

# same, but without root squashing for barry

/var/ftp pasha(rw) barry(rw,no_root_squash)

# everyone from the netsec.lan domain gets access

/var/backup *.netsec.lan(rw)

# ro for one network, rw for the other

/var/upload 192.168.1.0/24(ro) 192.168.5.0/24(rw)

You don't need to restart the nfs server to start exporting your newly created exports.
You can use the exportfs -va command to do this. It will write the exported directories
to /var/lib/nfs/etab, where they are immediately applied.

8.21.3. client configuration

We have seen the mount command and the /etc/fstab file before.

root@RHELv4u2:~# mount -t nfs barry:/mnt/data/iso /home/project55/

root@RHELv4u2:~# cat /etc/fstab | grep nfs
barry:/mnt/data/iso /home/iso nfs defaults 0 0
root@RHELv4u2:~#

Here is another simple example. Suppose the project55 people tell you they only
need a couple of CD-ROM images, and you already have them available on an nfs
server. You could issue the following command to mount this storage on their /home/
project55 mount point.

root@RHELv4u2:~# mount -t nfs 192.168.1.40:/mnt/data/iso /home/project55/

root@RHELv4u2:~# ls -lh /home/project55/
total 3.6G
drwxr-xr-x 2 1000 1000 4.0K Jan 16 17:55 RHELv4u1

176
 Introduction to networking

drwxr-xr-x 2 1000 1000 4.0K Jan 16 14:14 RHELv4u2

drwxr-xr-x 2 1000 1000 4.0K Jan 16 14:54 RHELv4u3
drwxr-xr-x 2 1000 1000 4.0K Jan 16 11:09 RHELv4u4
-rw-r--r-- 1 root root 1.6G Oct 13 15:22 sled10-vmwarews5-vm.zip
root@RHELv4u2:~#

8.22. practice : network file system

1. Create two directories with some files. Use nfs to share one of them as read only,
the other must be writable. Have your neighbour connect to them to test.

2. Investigate the user owner of the files created by your neighbour.

3. Protect a share by ip-address or hostname, so only your neighbour can connect.

177
Chapter 9. Scheduling

Table of Contents
9.1. about scheduling .......................................................................................... 178
9.2. one time jobs with at ................................................................................... 178
9.3. cron ............................................................................................................... 180
9.4. practice : scheduling .................................................................................... 182
9.5. solution : scheduling .................................................................................... 183

9.1. about scheduling

Linux administrators use the at to schedule one time jobs. Recurring jobs are better
scheduled with cron. The next two sections will discuss both tools.

9.2. one time jobs with at

9.2.1. at
Simple scheduling can be done with the at command. This screenshot shows the
scheduling of the date command at 22:01 and the sleep command at 22:03.

root@laika:~# at 22:01
at> date
at> <EOT>
job 1 at Wed Aug 1 22:01:00 2007
root@laika:~# at 22:03
at> sleep 10
at> <EOT>
job 2 at Wed Aug 1 22:03:00 2007
root@laika:~#

In real life you will hopefully be scheduling more useful commands ;-)

9.2.2. atq
It is easy to check when jobs are scheduled with the atq or at -l commands.

root@laika:~# atq
1 Wed Aug 1 22:01:00 2007 a root
2 Wed Aug 1 22:03:00 2007 a root
root@laika:~# at -l
1 Wed Aug 1 22:01:00 2007 a root
2 Wed Aug 1 22:03:00 2007 a root
root@laika:~#

178
 Scheduling

The at command understands English words like tomorrow and teatime to schedule
commands the next day and at four in the afternoon.

root@laika:~# at 10:05 tomorrow

at> sleep 100
at> <EOT>
job 5 at Thu Aug 2 10:05:00 2007
root@laika:~# at teatime tomorrow
at> tea
at> <EOT>
job 6 at Thu Aug 2 16:00:00 2007
root@laika:~# atq
6 Thu Aug 2 16:00:00 2007 a root
5 Thu Aug 2 10:05:00 2007 a root
root@laika:~#

9.2.3. atrm
Jobs in the at queue can be removed with atrm.

root@laika:~# atq
6 Thu Aug 2 16:00:00 2007 a root
5 Thu Aug 2 10:05:00 2007 a root
root@laika:~# atrm 5
root@laika:~# atq
6 Thu Aug 2 16:00:00 2007 a root
root@laika:~#

9.2.4. at.allow and at.deny

You can also use the /etc/at.allow and /etc/at.deny files to manage who can schedule
jobs with at.

The /etc/at.allow file can contain a list of users that are allowed to schedule at jobs.
When /etc/at.allow does not exist, then everyone can use at unless their username
is listed in /etc/at.deny.

If none of these files exist, then everyone can use at.

179
 Scheduling

9.3. cron

9.3.1. crontab file

The crontab(1) command can be used to maintain the crontab(5) file. Each user
can have their own crontab file to schedule jobs at a specific time. This time can be
specified with five fields in this order: minute, hour, day of the month, month and day
of the week. If a field contains an asterisk (*), then this means all values of that field.

The following example means : run script42 eight minutes after two, every day of the
month, every month and every day of the week.
8 14 * * * script42

Run script8472 every month on the first of the month at 25 past midnight.
25 0 1 * * script8472

Run this script33 every two minutes on Sunday (both 0 and 7 refer to Sunday).
*/2 * * * 0

Instead of these five fields, you can also type one of these: @reboot, @yearly or
@annually, @monthly, @weekly, @daily or @midnight, and @hourly.

9.3.2. crontab command

Users should not edit the crontab file directly, instead they should type crontab -e
which will use the editor defined in the EDITOR or VISUAL environment variable.
Users can display their cron table with crontab -l.

9.3.3. cron.allow and cron.deny

The cron daemon crond is reading the cron tables, taking into account the /etc/
cron.allow and /etc/cron.deny files.

These files work in the same way as at.allow and at.deny. When the cron.allow file
exists, then your username has to be in it, otherwise you cannot use cron. When the
cron.allow file does not exists, then your username cannot be in the cron.deny file
if you want to use cron.

9.3.4. /etc/crontab
The /etc/crontab file contains entries for when to run hourly/daily/weekly/monthly
tasks. It will look similar to this output.

180
 Scheduling

SHELL=/bin/sh
PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin

20 3 * * * root run-parts --report /etc/cron.daily

40 3 * * 7 root run-parts --report /etc/cron.weekly
55 3 1 * * root run-parts --report /etc/cron.monthly

9.3.5. /etc/cron.*
The directories shown in the next screenshot contain the tasks that are run at the times
scheduled in /etc/crontab. The /etc/cron.d directory is for special cases, to schedule
jobs that require finer control than hourly/daily/weekly/monthly.

paul@laika:~$ ls -ld /etc/cron.*

drwxr-xr-x 2 root root 4096 2008-04-11 09:14 /etc/cron.d
drwxr-xr-x 2 root root 4096 2008-04-19 15:04 /etc/cron.daily
drwxr-xr-x 2 root root 4096 2008-04-11 09:14 /etc/cron.hourly
drwxr-xr-x 2 root root 4096 2008-04-11 09:14 /etc/cron.monthly
drwxr-xr-x 2 root root 4096 2008-04-11 09:14 /etc/cron.weekly

181
 Scheduling

9.4. practice : scheduling

1. Schedule two jobs with at, display the at queue and remove a job.

2. As normal user, use crontab -e to schedule a script to run every four minutes.

3. As root, display the crontab file of your normal user.

4. As the normal user again, remove your crontab file.

5. Take a look at the cron files and directories in /etc and understand them. What is
the run-parts command doing ?

182
 Scheduling

9.5. solution : scheduling

1. Schedule two jobs with at, display the at queue and remove a job.
root@rhel55 ~# at 9pm today
at> echo go to bed >> /root/todo.txt
at> <EOT>
job 1 at 2010-11-14 21:00
root@rhel55 ~# at 17h31 today
at> echo go to lunch >> /root/todo.txt
at> <EOT>
job 2 at 2010-11-14 17:31
root@rhel55 ~# atq
2 2010-11-14 17:31 a root
1 2010-11-14 21:00 a root
root@rhel55 ~# atrm 1
root@rhel55 ~# atq
2 2010-11-14 17:31 a root
root@rhel55 ~# date
Sun Nov 14 17:31:01 CET 2010
root@rhel55 ~# cat /root/todo.txt
go to lunch

2. As normal user, use crontab -e to schedule a script to run every four minutes.
paul@rhel55 ~$ crontab -e
no crontab for paul - using an empty one
crontab: installing new crontab

3. As root, display the crontab file of your normal user.

root@rhel55 ~# crontab -l -u paul
*/4 * * * * echo `date` >> /home/paul/crontest.txt

4. As the normal user again, remove your crontab file.

paul@rhel55 ~$ crontab -r
paul@rhel55 ~$ crontab -l
no crontab for paul

5. Take a look at the cron files and directories in /etc and understand them. What is
the run-parts command doing ?
run-parts runs a script in a directory

183
Chapter 10. Logging
Table of Contents
10.1. about logging ............................................................................................. 184
10.2. login logging .............................................................................................. 185
10.3. syslogd daemon .......................................................................................... 187
10.4. logger ......................................................................................................... 189
10.5. watching logs ............................................................................................. 190
10.6. rotating logs ............................................................................................... 190
10.7. practice : logging ........................................................................................ 191
10.8. solution : logging ....................................................................................... 192

10.1. about logging

10.1.1. /var/log
The location for log files according to the FHS is /var/log. You will find a lot of log
files and directories for common applications in /var/log.

[paul@RHEL4b ~]$ ls /var/log

acpid cron.2 maillog.2 quagga secure.4
amanda cron.3 maillog.3 radius spooler
anaconda.log cron.4 maillog.4 rpmpkgs spooler.1
anaconda.syslog cups mailman rpmpkgs.1 spooler.2
anaconda.xlog dmesg messages rpmpkgs.2 spooler.3
audit exim messages.1 rpmpkgs.3 spooler.4
boot.log gdm messages.2 rpmpkgs.4 squid
boot.log.1 httpd messages.3 sa uucp
boot.log.2 iiim messages.4 samba vbox
boot.log.3 iptraf mysqld.log scrollkeeper.log vmware-tools-guestd
boot.log.4 lastlog news secure wtmp
canna mail pgsql secure.1 wtmp.1
cron maillog ppp secure.2 Xorg.0.log
cron.1 maillog.1 prelink.log secure.3 Xorg.0.log.old
[paul@RHEL4b ~]$

10.1.2. /var/log/messages
A typical first file to check when troubleshooting is the /var/log/messages file. By
default this file will contain information on what just happened to the system.

[root@RHEL4b ~]# tail /var/log/messages

Jul 30 05:13:56 anacron: anacron startup succeeded
Jul 30 05:13:56 atd: atd startup succeeded
Jul 30 05:13:57 messagebus: messagebus startup succeeded
Jul 30 05:13:57 cups-config-daemon: cups-config-daemon startup succeeded
Jul 30 05:13:58 haldaemon: haldaemon startup succeeded
Jul 30 05:14:00 fstab-sync[3560]: removed all generated mount points

184
 Logging

Jul 30 05:14:01 fstab-sync[3628]: added mount point /media/cdrom for...

Jul 30 05:14:01 fstab-sync[3646]: added mount point /media/floppy for...
Jul 30 05:16:46 sshd(pam_unix)[3662]: session opened for user paul by...
Jul 30 06:06:37 su(pam_unix)[3904]: session opened for user root by paul
[root@RHEL4b ~]#

10.2. login logging

To keep track of who is logging into the system, Linux can maintain the /var/log/
wtmp, /var/log/btmp, /var/run/utmp and /var/log/lastlog files.

10.2.1. /var/run/utmp (who)

Use the who command to see the /var/run/utmp file. This command is showing you
all the currently logged in users. Notice that the utmp file is in /var/run and not in /
var/log .

[root@rhel4 ~]# who

paul pts/1 Feb 14 18:21 (192.168.1.45)
sandra pts/2 Feb 14 18:11 (192.168.1.42)
inge pts/3 Feb 14 12:01 (192.168.1.33)
els pts/4 Feb 14 14:33 (192.168.1.19)

10.2.2. /var/log/wtmp (last)

The /var/log/wtmp file is updated by the login program. Use last to see the /var/
run/wtmp file.

[root@rhel4a ~]# last | head

paul pts/1 192.168.1.45 Wed Feb 14 18:39 still logged in
reboot system boot 2.6.9-42.0.8.ELs Wed Feb 14 18:21 (01:15)
nicolas pts/5 pc-dss.telematic Wed Feb 14 12:32 - 13:06 (00:33)
stefaan pts/3 pc-sde.telematic Wed Feb 14 12:28 - 12:40 (00:12)
nicolas pts/3 pc-nae.telematic Wed Feb 14 11:36 - 12:21 (00:45)
nicolas pts/3 pc-nae.telematic Wed Feb 14 11:34 - 11:36 (00:01)
dirk pts/5 pc-dss.telematic Wed Feb 14 10:03 - 12:31 (02:28)
nicolas pts/3 pc-nae.telematic Wed Feb 14 09:45 - 11:34 (01:48)
dimitri pts/5 rhel4 Wed Feb 14 07:57 - 08:38 (00:40)
stefaan pts/4 pc-sde.telematic Wed Feb 14 07:16 - down (05:50)
[root@rhel4a ~]#

The last command can also be used to get a list of last reboots.

[paul@rekkie ~]$ last reboot

reboot system boot 2.6.16-rekkie Mon Jul 30 05:13 (370+08:42)

wtmp begins Tue May 30 23:11:45 2006

[paul@rekkie ~]$

185
 Logging

10.2.3. /var/log/lastlog (lastlog)

Use lastlog to see the /var/log/lastlog file.

[root@rhel4a ~]# lastlog | tail

tim pts/5 10.170.1.122 Tue Feb 13 09:36:54 +0100 2007
rm pts/6 rhel4 Tue Feb 13 10:06:56 +0100 2007
henk **Never logged in**
stefaan pts/3 pc-sde.telematic Wed Feb 14 12:28:38 +0100 2007
dirk pts/5 pc-dss.telematic Wed Feb 14 10:03:11 +0100 2007
arsene **Never logged in**
nicolas pts/5 pc-dss.telematic Wed Feb 14 12:32:18 +0100 2007
dimitri pts/5 rhel4 Wed Feb 14 07:57:19 +0100 2007
bashuserrm pts/7 rhel4 Tue Feb 13 10:35:40 +0100 2007
kornuserrm pts/5 rhel4 Tue Feb 13 10:06:17 +0100 2007
[root@rhel4a ~]#

10.2.4. /var/log/btmp (lastb)

There is also the lastb command to display the /var/log/btmp file. This file is updated
by the login program when entering the wrong password, so it contains failed login
attempts. Many computers will not have this file, resulting in no logging of failed
login attempts.

[root@RHEL4b ~]# lastb

lastb: /var/log/btmp: No such file or directory
Perhaps this file was removed by the operator to prevent logging lastb\
info.
[root@RHEL4b ~]#

The reason given for this is that users sometimes type their password by mistake
instead of their login, so this world readable file poses a security risk. You can
enable bad login logging by simply creating the file. Doing a chmod o-r /var/log/
btmp improves security.

[root@RHEL4b ~]# touch /var/log/btmp

[root@RHEL4b ~]# ll /var/log/btmp
-rw-r--r-- 1 root root 0 Jul 30 06:12 /var/log/btmp
[root@RHEL4b ~]# chmod o-r /var/log/btmp
[root@RHEL4b ~]# lastb

btmp begins Mon Jul 30 06:12:19 2007

[root@RHEL4b ~]#

Failed logins via ssh, rlogin or su are not registered in /var/log/btmp. Failed logins
via tty are.

[root@RHEL4b ~]# lastb

HalvarFl tty3 Mon Jul 30 07:10 - 07:10 (00:00)
Maria tty1 Mon Jul 30 07:09 - 07:09 (00:00)
Roberto tty1 Mon Jul 30 07:09 - 07:09 (00:00)

186
 Logging

btmp begins Mon Jul 30 07:09:32 2007

[root@RHEL4b ~]#

10.2.5. su and ssh logins

Depending on the distribution, you may also have the /var/log/secure file being filled
with messages from the auth and/or authpriv syslog facilities. This log will include
su and/or ssh failed login attempts. Some distributions put this in /var/log/auth.log,
verify the syslog configuration.

[root@RHEL4b ~]# cat /var/log/secure

Jul 30 07:09:03 sshd[4387]: Accepted publickey for paul from ::ffff:19\
2.168.1.52 port 33188 ssh2
Jul 30 05:09:03 sshd[4388]: Accepted publickey for paul from ::ffff:19\
2.168.1.52 port 33188 ssh2
Jul 30 07:22:27 sshd[4655]: Failed password for Hermione from ::ffff:1\
92.168.1.52 port 38752 ssh2
Jul 30 05:22:27 sshd[4656]: Failed password for Hermione from ::ffff:1\
92.168.1.52 port 38752 ssh2
Jul 30 07:22:30 sshd[4655]: Failed password for Hermione from ::ffff:1\
92.168.1.52 port 38752 ssh2
Jul 30 05:22:30 sshd[4656]: Failed password for Hermione from ::ffff:1\
92.168.1.52 port 38752 ssh2
Jul 30 07:22:33 sshd[4655]: Failed password for Hermione from ::ffff:1\
92.168.1.52 port 38752 ssh2
Jul 30 05:22:33 sshd[4656]: Failed password for Hermione from ::ffff:1\
92.168.1.52 port 38752 ssh2
Jul 30 08:27:33 sshd[5018]: Invalid user roberto from ::ffff:192.168.1\
.52
Jul 30 06:27:33 sshd[5019]: input_userauth_request: invalid user rober\
to
Jul 30 06:27:33 sshd[5019]: Failed none for invalid user roberto from \
::ffff:192.168.1.52 port 41064 ssh2
Jul 30 06:27:33 sshd[5019]: Failed publickey for invalid user roberto \
from ::ffff:192.168.1.52 port 41064 ssh2
Jul 30 08:27:36 sshd[5018]: Failed password for invalid user roberto f\
rom ::ffff:192.168.1.52 port 41064 ssh2
Jul 30 06:27:36 sshd[5019]: Failed password for invalid user roberto f\
rom ::ffff:192.168.1.52 port 41064 ssh2
[root@RHEL4b ~]#

You can enable this yourself, with a custom log file by adding the following line tot
syslog.conf.
auth.*,authpriv.* /var/log/customsec.log

10.3. syslogd daemon

10.3.1. about syslog

The standard method of logging on Linux is through the syslogd daemon. Syslog
was developed by Eric Allman for sendmail, but quickly became a standard among

187
 Logging

many Unix applications and was much later written as rfc 3164. The syslog daemon
can receive messages on udp port 514 from many applications (and appliances), and
can append to log files, print, display messages on terminals and forward logs to
other syslogd daemons on other machines. The syslogd daemon is configured in /
etc/syslog.conf.

Each line in the configuration file uses a facility to determine where the message is
coming from. It also contains a level for the severity of the message, and an action
to decide on what to do with the message.

10.3.2. facilities
The man syslog.conf will explain the different default facilities for certain daemons,
such as mail, lpr, news and kern(el) messages. The local0 to local7 facility can be
used for appliances (or any networked device that supports syslog). Here is a list of
all facilities for syslog.conf version 1.3. The security keyword is deprecated.

auth (security)
authpriv
cron
daemon
ftp
kern
lpr mail
mark (internal use only)
news
syslog
user
uucp
local0-7

10.3.3. levels
The worst severity a message can have is emerg followed by alert and crit. Lowest
priority should go to info and debug messages. Specifying a severity will also log
all messages with a higher severity. You can prefix the severity with = to obtain only
messages that match that severity. You can also specify .none to prevent a specific
action from any message from a certain facility.

Here is a list of all levels, in ascending order. The keywords warn, error and panic
are deprecated.

debug
info
notice
warning (warn)
err (error)
crit
alert
emerg (panic)

188
 Logging

10.3.4. actions
The default action is to send a message to the username listed as action. When the
action is prefixed with a / then syslog will send the message to the file (which can be
a regular file, but also a printer or terminal). The @ sign prefix will send the message
on to another syslog server. Here is a list of all possible actions.

root,user1 list of users, separated by comma's

* message to all logged on users
/ file (can be a printer, a console, a tty, ...)
-/ file, but don't sync after every write
| named pipe
@ other syslog hostname

In addition, you can prefix actions with a - to omit syncing the file after every logging.

10.3.5. configuration
Below a sample configuration of custom local4 messages in /etc/syslog.conf.

local4.crit /var/log/critandabove
local4.=crit /var/log/onlycrit
local4.* /var/log/alllocal4

Don't forget to restart the server.

[root@rhel4a ~]# /etc/init.d/syslog restart

Shutting down kernel logger: [ OK ]
Shutting down system logger: [ OK ]
Starting system logger: [ OK ]
Starting kernel logger: [ OK ]
[root@rhel4a ~]#

10.4. logger
The logger command can be used to generate syslog test messages. You can aslo use
it in scripts. An example of testing syslogd with the logger tool.

[root@rhel4a ~]# logger -p local4.debug "l4 debug"

[root@rhel4a ~]# logger -p local4.crit "l4 crit"
[root@rhel4a ~]# logger -p local4.emerg "l4 emerg"
[root@rhel4a ~]#

The results of the tests with logger.

[root@rhel4a ~]# cat /var/log/critandabove

Feb 14 19:55:19 rhel4a paul: l4 crit

189
 Logging

Feb 14 19:55:28 rhel4a paul: l4 emerg

[root@rhel4a ~]# cat /var/log/onlycrit
Feb 14 19:55:19 rhel4a paul: l4 crit
[root@rhel4a ~]# cat /var/log/alllocal4
Feb 14 19:55:11 rhel4a paul: l4 debug
Feb 14 19:55:19 rhel4a paul: l4 crit
Feb 14 19:55:28 rhel4a paul: l4 emerg
[root@rhel4a ~]#

10.5. watching logs

You might want to use the tail -f command to look at the last lines of a log file. The
-f option will dynamically display lines that are appended to the log. You can do the
same with other commands by preceding them with the watch command.

10.6. rotating logs

A lot of log files are always growing in size. To keep this within bounds, you might
want to use logrotate to rotate, compress, remove and mail log files. More info on
the logrotate command in the scheduling chapter.

190
 Logging

10.7. practice : logging

1. Display the /var/run/utmp file with the proper command (not with cat or vi).

2. Display the /var/log/wtmp file.

3. Use the lastlog and lastb commands, understand the difference.

4. Examine syslog to find the location of the log file containing ssh failed logins.

5. Configure syslog to put local4.error and above messages in /var/log/l4e.log and

local4.info only .info in /var/log/l4i.log. Test that it works with the logger tool!

6. Configure /var/log/Mysu.log, all the su to root messages should go in that log. Test
that it works!

7. Send the local5 messages to the syslog server of your neighbour. Test that it works.

8. Write a script that executes logger to local4 every 15 seconds (different message).
Use tail -f and watch on your local4 log files.

191
 Logging

10.8. solution : logging

1. Display the /var/run/utmp file.
who

2. Display the /var/log/wtmp file.

last

3. Use the lastlog and lastb commands, understand the difference.

lastlog : when users last logged on

lastb: failed (bad) login attempts

4. Examine syslog to find the location of the log file containing ssh failed logins.

root@rhel53 ~# grep authpriv /etc/syslog.conf

authpriv.* /var/log/secure

Debian/Ubuntu: /var/log/auth.log

Ubuntu 9.10 and Debian Lenny have switched to using rsyslog.

root@ubuntu910:~# grep authpriv /etc/rsyslog.d/50-default.conf

auth,authpriv.* /var/log/auth.log

root@deb503:~# grep authpriv /etc/rsyslog.conf

auth,authpriv.* /var/log/auth.log

5. Configure syslog to put local4.error and above messages in /var/log/l4e.log and

local4.info only .info in /var/log/l4i.log. Test that it works with the logger tool!
echo local4.error /var/log/l4e.log >> /etc/syslog.conf

echo local4.=info /var/log/l4i.log >> /etc/syslog.conf

/etc/init.d/syslog restart

logger -p local4.error "l4 error test"

logger -p local4.alert "l4 alert test"

logger -p local4.info "l4 info test"

cat /var/log/l4e.log

cat /var/log/l4i.log

6. Configure /var/log/Mysu.log, all the su to root messages should go in that log. Test
that it works!
echo authpriv.* /var/log/Mysu.log >> /etc/syslog.conf

This will log more than just the su usage.

192
 Logging

7. Send the local5 messages to the syslog server of your neighbour. Test that it works.

On RHEL5, edit /etc/sysconfig/syslog to enable remote listening on the server.

On Debian/Ubuntu edit /etc/default/syslog or /etc/default/rsyslog.

on the client: logger -p local5.info "test local5 to neighbour"

8. Write a script that executes logger to local4 every 15 seconds (different message).
Use tail -f and watch on your local4 log files.

root@rhel53 scripts# cat logloop

#!/bin/bash

for i in `seq 1 10`

do
logger -p local4.info "local4.info test number $i"
sleep 15
done

root@rhel53 scripts# chmod +x logloop

root@rhel53 scripts# ./logloop &
[1] 8264
root@rhel53 scripts# tail -f /var/log/local4.all.log
Mar 28 13:13:36 rhel53 root: local4.info test number 1
Mar 28 13:13:51 rhel53 root: local4.info test number 2
...

193
Chapter 11. Library Management

Table of Contents
11.1. introduction ................................................................................................ 194
11.2. /lib and /usr/lib ........................................................................................... 194
11.3. ldd ............................................................................................................... 194
11.4. ltrace ........................................................................................................... 195
11.5. dpkg -S and debsums ................................................................................. 195
11.6. rpm -qf and rpm -V ................................................................................... 196

11.1. introduction
With libraries we are talking about dynamically linked libraries (aka shared objects).
These are binaries that contain functions and are not started themselves as programs,
but are called by other binaries.

Several programs can use the same library. The name of the library file usually starts
with lib, followed by the actual name of the library, then the chracters .so and finally
a version number.

11.2. /lib and /usr/lib

When you look at the /lib or the /usr/lib directory, you will see a lot of symbolic links.
Most libraries have a detailed version number in their name, but receive a symbolic
link from a filename which only contains the major version number.

root@rhel53 ~# ls -l /lib/libext*
lrwxrwxrwx 1 root root 16 Feb 18 16:36 /lib/libext2fs.so.2 -> libext2fs.so.2.4
-rwxr-xr-x 1 root root 113K Jun 30 2009 /lib/libext2fs.so.2.4

11.3. ldd
Many programs have dependencies on the installation of certain libraries. You can
display these dependencies with ldd.

This example shows the dependencies of the su command.

paul@RHEL5 ~$ ldd /bin/su

linux-gate.so.1 => (0x003f7000)
libpam.so.0 => /lib/libpam.so.0 (0x00d5c000)
libpam_misc.so.0 => /lib/libpam_misc.so.0 (0x0073c000)
libcrypt.so.1 => /lib/libcrypt.so.1 (0x00aa4000)
libdl.so.2 => /lib/libdl.so.2 (0x00800000)
libc.so.6 => /lib/libc.so.6 (0x00ec1000)

194
 Library Management

libaudit.so.0 => /lib/libaudit.so.0 (0x0049f000)

/lib/ld-linux.so.2 (0x4769c000)

11.4. ltrace
The ltrace program allows to see all the calls made to library functions by a program.
The example below uses the -c option to get only a summary count (there can be
many calls), and the -l option to only show calls in one library file. All this to see
what calls are made when executing su - serena as root.

root@deb503:~# ltrace -c -l /lib/libpam.so.0 su - serena

serena@deb503:~$ exit
logout
% time seconds usecs/call calls function
------ ----------- ----------- --------- --------------------
70.31 0.014117 14117 1 pam_start
12.36 0.002482 2482 1 pam_open_session
5.17 0.001039 1039 1 pam_acct_mgmt
4.36 0.000876 876 1 pam_end
3.36 0.000675 675 1 pam_close_session
3.22 0.000646 646 1 pam_authenticate
0.48 0.000096 48 2 pam_set_item
0.27 0.000054 54 1 pam_setcred
0.25 0.000050 50 1 pam_getenvlist
0.22 0.000044 44 1 pam_get_item
------ ----------- ----------- --------- --------------------
100.00 0.020079 11 total

11.5. dpkg -S and debsums

Find out on Debian/Ubuntu to which package a library belongs.

paul@deb503:/lib$ dpkg -S libext2fs.so.2.4

e2fslibs: /lib/libext2fs.so.2.4

You can then verify the integrity of all files in this package using debsums.

paul@deb503:~$ debsums e2fslibs

/usr/share/doc/e2fslibs/changelog.Debian.gz OK
/usr/share/doc/e2fslibs/copyright OK
/lib/libe2p.so.2.3 OK
/lib/libext2fs.so.2.4 OK

Should a library be broken, then reinstall it with aptitude reinstall $package.

root@deb503:~# aptitude reinstall e2fslibs

Reading package lists... Done
Building dependency tree
Reading state information... Done
Reading extended state information
Initializing package states... Done

195
 Library Management

Reading task descriptions... Done

The following packages will be REINSTALLED:
e2fslibs
...

11.6. rpm -qf and rpm -V

Find out on Red Hat/Fedora to which package a library belongs.

paul@RHEL5 ~$ rpm -qf /lib/libext2fs.so.2.4

e2fsprogs-libs-1.39-8.el5

You can then use rpm -V to verify all files in this package. In the example below
the output shows that the Size and the Time stamp of the file have changed since
installation.

root@rhel53 ~# rpm -V e2fsprogs-libs

prelink: /lib/libext2fs.so.2.4: prelinked file size differs
S.?....T /lib/libext2fs.so.2.4

You can then use yum reinstall $package to overwrite the existing library with an
original version.

root@rhel53 lib# yum reinstall e2fsprogs-libs

Loaded plugins: rhnplugin, security
Setting up Reinstall Process
Resolving Dependencies
--> Running transaction check
---> Package e2fsprogs-libs.i386 0:1.39-23.el5 set to be erased
---> Package e2fsprogs-libs.i386 0:1.39-23.el5 set to be updated
--> Finished Dependency Resolution
...

The package verification now reports no problems with the library.

root@rhel53 lib# rpm -V e2fsprogs-libs

root@rhel53 lib#

196
Chapter 12. Memory management

Table of Contents
12.1. about memory ............................................................................................ 197
12.2. /proc/meminfo ............................................................................................ 197
12.3. swap space ................................................................................................. 198
12.4. practice : memory ...................................................................................... 199

12.1. about memory

You can display information about ram memory with free -om, top and cat /proc/
meminfo. You should understand terms like swapping, paging and virtual memory.

12.2. /proc/meminfo
You will rarely want to look at /proc/meminfo...

paul@RHELv4u4:~$ cat /proc/meminfo

MemTotal: 255864 kB
MemFree: 5336 kB
Buffers: 42396 kB
Cached: 159912 kB
SwapCached: 0 kB
Active: 104184 kB
Inactive: 119724 kB
HighTotal: 0 kB
HighFree: 0 kB
LowTotal: 255864 kB
LowFree: 5336 kB
SwapTotal: 1048568 kB
SwapFree: 1048568 kB
Dirty: 40 kB
Writeback: 0 kB
Mapped: 33644 kB
Slab: 21956 kB
CommitLimit: 1176500 kB
Committed_AS: 82984 kB
PageTables: 960 kB
VmallocTotal: 761848 kB
VmallocUsed: 2588 kB
VmallocChunk: 759096 kB
HugePages_Total: 0
HugePages_Free: 0
Hugepagesize: 4096 kB

...since the free command displays the same information in a more user friendly
output.

paul@RHELv4u4:~$ free -om

197
 Memory management

total used free shared buffers cached

Mem: 249 244 5 0 41 156
Swap: 1023 0 1023
paul@RHELv4u4:~$

12.3. swap space

12.3.1. about swap space

When the operating system needs more memory than physically present in RAM, it
will use swap space. Swap space is located on slower but cheaper memory. Notice
that, although hard disks are commonly used for swap space, their access times are
one hundred thousand times slower.

The swap space can be a file, a partition, or a combination of files and partitions. You
can see the swap space with the free command, or with cat /proc/swaps.

paul@RHELv4u4:~$ free -om

total used free shared buffers cached
Mem: 249 245 4 0 125 55
Swap: 1023 0 1023
paul@RHELv4u4:~$ cat /proc/swaps
Filename Type Size Used Priority
/dev/mapper/VolGroup00-LogVol01 partition 1048568 0 -1
paul@RHELv4u4:~$

The amount of swap space that you need depends heavily on the services that the
computer provides.

12.3.2. creating a swap partition

You can activate or deactivate swap space with the swapon and swapoff commands.
New swap space can be created with the mkswap command. The screenshot below
shows the creation and activation of a swap partition.

root@RHELv4u4:~# fdisk -l 2> /dev/null | grep hda

Disk /dev/hda: 536 MB, 536870912 bytes
/dev/hda1 1 1040 524128+ 83 Linux
root@RHELv4u4:~# mkswap /dev/hda1
Setting up swapspace version 1, size = 536702 kB
root@RHELv4u4:~# swapon /dev/hda1

Now you can see that /proc/swaps displays all swap spaces separately, whereas the
free -om command only makes a human readable summary.

root@RHELv4u4:~# cat /proc/swaps

Filename Type Size Used Priority

198
 Memory management

/dev/mapper/VolGroup00-LogVol01 partition 1048568 0 -1

/dev/hda1 partition 524120 0 -2
root@RHELv4u4:~# free -om
total used free shared buffers cached
Mem: 249 245 4 0 125 54
Swap: 1535 0 1535
root@RHELv4u4:~#

12.3.3. creating a swap file

Here is one more example showing you how to create a swap file. On Solaris you
can use mkfile instead of dd.

root@RHELv4u4:~# dd if=/dev/zero of=/smallswapfile bs=1024 count=4096

4096+0 records in
4096+0 records out
root@RHELv4u4:~# mkswap /smallswapfile
Setting up swapspace version 1, size = 4190 kB
root@RHELv4u4:~# swapon /smallswapfile
root@RHELv4u4:~# cat /proc/swaps
Filename Type Size Used Priority
/dev/mapper/VolGroup00-LogVol01 partition 1048568 0 -1
/dev/hda1 partition 524120 0 -2
/smallswapfile file 4088 0 -3
root@RHELv4u4:~#

12.3.4. swap space in /etc/fstab

If you like these swaps to be permanent, then don't forget to add them to /etc/fstab.
The lines in /etc/fstab will be similar to the following.

/dev/hda1 swap swap defaults 0 0

/smallswapfile swap swap defaults 0 0

12.4. practice : memory

1. Use dmesg to find the total amount of memory in your computer.

2. Use free to display memory usage in kilobytes (then in megabytes).

3. On the Red Hat, create a swap partition on one of your new disks, and a swap file
on the other new disk.

4. Put all swap spaces in /etc/fstab and activate them. Use free again to verify that
it works.

199
Chapter 13. Installing Linux

Table of Contents
13.1. about ........................................................................................................... 200
13.2. installation by cdrom ................................................................................. 200
13.3. installation with rarp and tftp .................................................................... 200
13.4. about Red Hat kickstart ............................................................................. 201
13.5. using kickstart ............................................................................................ 202

13.1. about
The past couple of years the installation of linux has become a lot easier then before,
at least for end users installing a distro like Ubuntu, Fedora, Debian or Mandrake
on their home computer. Servers usually come pre-installed, and if not pre-installed,
then setup of a linux server today is very easy.

Linux can be installed in many different ways. End users most commonly use cdrom's
or dvd's for installation, most of the time with a working internet connection te receive
updates. Administrators might prefer network installations using protocols like tftp,
bootp, rarp and/or nfs or response file solutions like Red Hat Kickstart or Solaris
Jumpstart.

13.2. installation by cdrom

Installation of linux from cdrom is easy! Most distributions ask very few questions
during install (keyboard type, language, username) and detect all the hardware
themselves. There is usually no need to retrieve third-party drivers from the internet.
The GUI installation gives options like Desktop (for end users), Workstation (for
developers), Server or minimal (usually without graphical interface).

13.3. installation with rarp and tftp

Installing over the network involves powering on the machine, have it find a rarpd
server to get an ip-address, then let it find an tftps server to get an installation image
copied to the machine. This image can then boot. The procedure below demonstrates
how to setup three Sun SPARC servers with Ubuntu Linux, using a Debian Linux
machine to host the tftp, bootp and nfs daemons.

First we need to configure the mac to ip resolution in the /etc/ethers configuration

file. Each server will receive a unique ip-address during installation.

root@laika:~# cat /etc/ethers

200
 Installing Linux

00:03:ba:02:c3:82 192.168.1.71
00:03:ba:09:7c:f9 192.168.1.72
00:03:ba:09:7f:d2 192.168.1.73

We need to install the rarpd and tftpd daemons on the (Debian) machine that will be
hosting the install image.

root@laika:~# aptitude install rarpd

root@laika:~# aptitude install tftpd

The tftp services must be activated in inetd or xinetd.

root@laika:~# cat /etc/inetd.conf | tail -1

tftp dgram udp wait nobody /usr/sbin/tcpd /usr/sbin/in.tftpd /srv/tftp

And finally the linux install image must be present in the tftp served directory. The
filename of the image must be the hex ip-address, this is accomplished with symbolic
links.

root@laika:~# ll /srv/tftp/
total 7.5M
lrwxrwxrwx 1 root root 13 2007-03-02 21:49 C0A80147 -> ubuntu610.img
lrwxrwxrwx 1 root root 13 2007-03-03 14:13 C0A80148 -> ubuntu610.img
lrwxrwxrwx 1 root root 13 2007-03-02 21:49 C0A80149 -> ubuntu610.img
-rw-r--r-- 1 paul paul 7.5M 2007-03-02 21:42 ubuntu610.img

Time to enter boot net now in the openboot prompt. Twenty minutes later the three
servers where humming with linux.

13.4. about Red Hat kickstart

Automating Linux installations with response files can be done with Red Hat
kickstart. One way to set it up is by using the graphical tool /usr/sbin/system-config-
kickstart. If you prefer to set it up manually, read on.

You can modify the sample kickstart file RH-DOCS/sample.ks (can be found on the
documentation dvd). Put this file so anaconda can read it.

Anaconda is the Red Hat installer written in python. The name is chose because
anacondas are lizard-eating pythons. Lizard is the name of the Caldera Linux
installation program.

Another option is to start with the /root/anaconda-ks.cfg file. This is a sample

kickstart file that contains all the settings from your current installation.

Do not change the order of the sections inside your kickstart file! The Red Hat System
Administration Guide contains about 25 pages describing all the options, most of
them are easy ti understand if you already performed a couple of installations.

201
 Installing Linux

13.5. using kickstart

To use kickstart, name your kickstart file ks.cfg and put it in the root directory of your
installation cdrom (or on a usb stick or a floppy). For network based installations,
name the file $ip-address-kickstart and place the following in dhcpd.conf.

filename "/export/kickstart"
next-server remote.installation.server

Leaving out the next-server line will result in the client looking for the file on the
dhcp server itself.

Booting from cdrom with kickstart requires the following command at the boot:
prompt.
linux ks=cdrom:/ks.cfg

When the kickstart file is on the network, use nfs or http like in these examples.
linux ks=nfs:servername:/path/to/ks.cfg

linux ks=http://servername/path/to/ks.cfg

202
Chapter 14. Package management

Table of Contents
14.1. terminology ................................................................................................ 203
14.2. Red Hat package manager ......................................................................... 204
14.3. yum ............................................................................................................. 205
14.4. rpm2cpio .................................................................................................... 210
14.5. Debian package management ..................................................................... 210
14.6. alien ............................................................................................................ 212
14.7. Downloading software ............................................................................... 213
14.8. Compiling software .................................................................................... 213
14.9. Practice: Installing software ....................................................................... 213
14.10. Solution: Installing software .................................................................... 214

14.1. terminology

14.1.1. repositories
Most software for your Linux distribution is available in a central distributed
repository. This means that applications in the repository are tested for your
distribution and very easy to install with a GUI or command line installer.

The GUI is available via the standard menu (look for Add/Remove Software). The
command line is explained below in detail.

14.1.2. rpm based

Red Hat, Fedora, OpenSUSE, Mandriva, Red Flag and others use .rpm packages.
The tool to manage software here is called yum, which uses the rpm command in
background.

14.1.3. Debian based

Debian, Ubuntu, Linux Mint and all derivatives from Debian and Ubuntu use .deb
packages. To manage software on these systems, you can use aptitude. aptitude is
a front end for dpkg.

14.1.4. building from source

With open source software, there is always the option to compile software yourself.
This is discussed at the end of this chapter.

203
 Package management

14.1.5. dependency
Some packages need other packages to function. Tools like aptitude and yum will
install all dependencies you need. When using dpkg or the rpm command, or when
building from source, you will need to install dependencies yourself.

14.2. Red Hat package manager

14.2.1. about rpm

The Red Hat package manager can be used on the command line with rpm or in a
graphical way going to Applications--System Settings--Add/Remove Applications.
Type rpm --help to see some of the options.

Software distributed in the rpm format will be named foo-version.platform.rpm .

14.2.2. rpm -qa

To obtain a list of all installed software, use the rpm -qa command.

[root@RHEL52 ~]# rpm -qa | grep samba

system-config-samba-1.2.39-1.el5
samba-3.0.28-1.el5_2.1
samba-client-3.0.28-1.el5_2.1
samba-common-3.0.28-1.el5_2.1

14.2.3. rpm -q
To verify whether one package is installed, use rpm -q.

root@RHELv4u4:~# rpm -q gcc

gcc-3.4.6-3
root@RHELv4u4:~# rpm -q laika
package laika is not installed

14.2.4. rpm -q --redhatprovides

To check whether a package is provided by Red Hat, use the --redhatprovides option.

root@RHELv4u4:~# rpm -q --redhatprovides bash

bash-3.0-19.3
root@RHELv4u4:~# rpm -q --redhatprovides gcc
gcc-3.4.6-3
root@RHELv4u4:~# rpm -q --redhatprovides laika
no package provides laika

204
 Package management

14.2.5. rpm -Uvh

To install or upgrade a package, use the -Uvh switches. The -U switch is the same as
-i for install, except that older versions of the software are removed. The -vh switches
are for nicer output.

root@RHELv4u4:~# rpm -Uvh gcc-3.4.6-3

14.2.6. rpm -e
To remove a package, use the -e switch.

root@RHELv4u4:~# rpm -e gcc-3.4.6-3

rpm -e verifies dependencies, and thus will prevent you from accidentailly erasing
packages that are needed by other packages.

[root@RHEL52 ~]# rpm -e gcc-4.1.2-42.el5

error: Failed dependencies:
gcc = 4.1.2-42.el5 is needed by (installed) gcc-c++-4.1.2-42.el5.i386
gcc = 4.1.2-42.el5 is needed by (installed) gcc-gfortran-4.1.2-42.el5.i386
gcc is needed by (installed) systemtap-0.6.2-1.el5_2.2.i386

14.2.7. /var/lib/rpm
The rpm database is located at /var/lib/rpm. This database contains all meta
information about packages that are installed (via rpm). It keeps track of all files,
which enables complete removes of software.

14.3. yum

14.3.1. about yum

The Yellowdog Updater, Modified (yum) is an easier command to work with rpm
packages. It is installed by default on Fedora and Red Hat Enterprise Linux since
version 5.2.

14.3.2. yum list

Issue yum list available to see a list of available packages. The available parameter
is optional.

205
 Package management

[root@rhel55 ~]# yum list | wc -l

2471

Issue yum list $package to get all versions (in different repositories) of one package.

[root@rhel55 ~]# yum list samba

Loaded plugins: rhnplugin, security
Installed Packages
samba.i386 3.0.33-3.28.el5 installed
Available Packages
samba.i386 3.0.33-3.29.el5_5 rhel-i386-server-5

14.3.3. yum search

To search for a package containing a certain string in the description or name use
yum search $string.

[root@rhel55 ~]# yum search gcc44

Loaded plugins: rhnplugin, security
========================== Matched: gcc44 ===========================
gcc44.i386 : Preview of GCC version 4.4
gcc44-c++.i386 : C++ support for GCC version 4.4
gcc44-gfortran.i386 : Fortran support for GCC 4.4 previe

14.3.4. yum provides

To search for a package containing a certain file (you might need for compiling
things) use yum provides $filename.

[root@rhel55 ~]# yum provides /usr/share/man/man1/gzip.1.gz

Loaded plugins: rhnplugin, security
Importing additional filelist information
gzip-1.3.5-9.el5.i386 : The GNU data compression program.
Repo : rhel-i386-server-5
Matched from:
Filename : /usr/share/man/man1/gzip.1.gz
...

14.3.5. yum install

To install an application, use yum install $package. Naturally yum will install all
the necessary dependencies.

[root@rhel55 ~]# yum install sudo

Loaded plugins: rhnplugin, security
Setting up Install Process

206
 Package management

Resolving Dependencies
--> Running transaction check
---> Package sudo.i386 0:1.7.2p1-7.el5_5 set to be updated
--> Finished Dependency Resolution

Dependencies Resolved

=======================================================================
Package Arch Version Repository Size
=======================================================================
Installing:
sudo i386 1.7.2p1-7.el5_5 rhel-i386-server-5 230 k

Transaction Summary
=======================================================================
Install 1 Package(s)
Upgrade 0 Package(s)

Total download size: 230 k

Is this ok [y/N]: y
Downloading Packages:
sudo-1.7.2p1-7.el5_5.i386.rpm
Running rpm_check_debug
Running Transaction Test
Finished Transaction Test
Transaction Test Succeeded
Running Transaction
Installing : sudo 1/1

Installed:
sudo.i386 0:1.7.2p1-7.el5_5

Complete!

You can add more than one parameter here.

yum install $package1 $package2 $package3

14.3.6. yum update

To bring all applications up to date, by downloading and installing them, issue yum
update. All software that was installed via yum will be updated to the latest version
that is available in the repository.
yum update

If you only want to update one package, use yum update $package.

[root@rhel55 ~]# yum update sudo

Loaded plugins: rhnplugin, security
Skipping security plugin, no data
Setting up Update Process
Resolving Dependencies
Skipping security plugin, no data
--> Running transaction check
---> Package sudo.i386 0:1.7.2p1-7.el5_5 set to be updated
--> Finished Dependency Resolution

Dependencies Resolved

207
 Package management

=====================================================================
Package Arch Version Repository Size
=====================================================================
Updating:
sudo i386 1.7.2p1-7.el5_5 rhel-i386-server-5 230 k

Transaction Summary
=====================================================================
Install 0 Package(s)
Upgrade 1 Package(s)

Total download size: 230 k

Is this ok [y/N]: y
Downloading Packages:
sudo-1.7.2p1-7.el5_5.i386.rpm | 230 kB 00:00
Running rpm_check_debug
Running Transaction Test
Finished Transaction Test
Transaction Test Succeeded
Running Transaction
Updating : sudo 1/2
Cleanup : sudo 2/2

Updated:
sudo.i386 0:1.7.2p1-7.el5_5

Complete!

14.3.7. yum software groups

Issue yum grouplist to see a list of all available software groups.
[root@rhel55 ~]# yum grouplist
Loaded plugins: rhnplugin, security
Setting up Group Process
Installed Groups:
Administration Tools
Authoring and Publishing
DNS Name Server
Development Libraries
Development Tools
Editors
GNOME Desktop Environment
GNOME Software Development
Graphical Internet
Graphics
Legacy Network Server
Legacy Software Development
Legacy Software Support
Mail Server
Network Servers
Office/Productivity
Printing Support
Server Configuration Tools
System Tools
Text-based Internet
Web Server
Windows File Server
X Software Development
X Window System

208
 Package management

Available Groups:
Engineering and Scientific
FTP Server
Games and Entertainment
Java Development
KDE (K Desktop Environment)
KDE Software Development
MySQL Database
News Server
OpenFabrics Enterprise Distribution
PostgreSQL Database
Sound and Video
Done

To install a set of applications, brought together via a group, use yum groupinstall
$groupname.
[root@rhel55 ~]# yum groupinstall 'Sound and video'
Loaded plugins: rhnplugin, security
Setting up Group Process
Package alsa-utils-1.0.17-1.el5.i386 already installed and latest version
Package sox-12.18.1-1.i386 already installed and latest version
Package 9:mkisofs-2.01-10.7.el5.i386 already installed and latest version
Package 9:cdrecord-2.01-10.7.el5.i386 already installed and latest version
Package cdrdao-1.2.1-2.i386 already installed and latest version
Resolving Dependencies
--> Running transaction check
---> Package cdda2wav.i386 9:2.01-10.7.el5 set to be updated
---> Package cdparanoia.i386 0:alpha9.8-27.2 set to be updated
---> Package sound-juicer.i386 0:2.16.0-3.el5 set to be updated
--> Processing Dependency: libmusicbrainz >= 2.1.0 for package: sound-juicer
--> Processing Dependency: libmusicbrainz.so.4 for package: sound-juicer
---> Package vorbis-tools.i386 1:1.1.1-3.el5 set to be updated
--> Processing Dependency: libao >= 0.8.4 for package: vorbis-tools
--> Processing Dependency: libao.so.2 for package: vorbis-tools
--> Running transaction check
---> Package libao.i386 0:0.8.6-7 set to be updated
---> Package libmusicbrainz.i386 0:2.1.1-4.1 set to be updated
--> Finished Dependency Resolution
...

Read the manual page of yum for more information about managing groups in yum.

14.3.8. /etc/yum.conf and repositories

The configuration of yum repositories is done in /etc/yum/yum.conf and /etc/yum/
repos.d/.

Configurating yum itself is done in /etc/yum.conf. This file will contain the location
of a log file and a cache directory for yum and can also contain a list of repositories.

Recently yum started accepting several repo files with each file containing a list of
repositories. These repo files are located in the /etc/yum.repos.d/ directory.

One important flag for yum is enablerepo. Use this command if you want to use a
repository that is not enabled by default.

209
 Package management

yum $command $foo --enablerepo=$repo

An example of the contents of the repo file: MyRepo.repo

[$repo]
name=My Repository
baseurl=http://path/to/MyRepo
gpgcheck=1
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-MyRep

14.4. rpm2cpio
We can use rpm2cpio to convert an rpm to a cpio archive.

[root@RHEL53 ~]# file kernel.src.rpm

kernel.src.rpm: RPM v3 src PowerPC kernel-2.6.18-92.1.13.el5
[root@RHEL53 ~]# rpm2cpio kernel.src.rpm > kernel.cpio
[root@RHEL53 ~]# file kernel.cpio
kernel.cpio: ASCII cpio archive (SVR4 with no CRC)

But why would you want to do this ?

Perhaps just to see of list of files in the rpm file.

[root@RHEL53 ~]# rpm2cpio kernel.src.rpm | cpio -t | head -5

COPYING.modules
Config.mk
Module.kabi_i686
Module.kabi_i686PAE
Module.kabi_i686xen

Or to extract one file from an rpm package.

[root@RHEL53 ~]# rpm2cpio kernel.src.rpm | cpio -iv Config.mk

Config.mk
246098 blocks

14.5. Debian package management

14.5.1. about deb

Most people use aptitude or apt-get to manage their Debian/Ubuntu family of Linux
distributions. Both are a front end for dpkg and are themselves a back end for
synaptic or other graphical tools.

This is a screenshot of synaptic running on Debian 4.0.

210
 Package management

14.5.2. dpkg -l
The low level tool to work with .deb packages is dpkg. Here you see how to obtain
a list of all installed packages. The ii at the beginning means the package is installed.

root@laika:~# dpkg -l | grep gcc-4.2

ii gcc-4.2 4.2.4-1ubuntu3 The GNU C compiler
ii gcc-4.2-base 4.2.4-1ubuntu3 The GNU Compiler Collection (base package)

14.5.3. dpkg
You could use dpkg -i to install a package and dpkg -r to remove a package, but
you'd have to manually keep track of dependencies.

14.5.4. aptitude
Most people use aptitude for package management on Debian and Ubuntu Systems.

To synchronize with the repositories.

aptitude update

To patch and upgrade all software to the latest version on Debian.

aptitude upgrade

211
 Package management

To patch and upgrade all software to the latest version on Ubuntu and Mint.
aptitude safe-upgrade

To install an application with all dependencies.

aptitude install $package

To search the repositories for applications that contain a certain string in their name
or description.
aptitude search $string

To remove an application and all unused files.

aptitude remove $package

14.5.5. apt-get
We could also use apt-get, but aptitude is better at handling dependencies than apt-
get. Whenever you see apt-get in a howto, feel free to type aptitude.

14.5.6. /etc/apt/sources.list
The resource list for both apt-get and aptitude is located in /etc/apt/sources.list.
This file contains a list of http or ftp sources where packages for the distribution can
be downloaded.

root@barry:~# cat /etc/apt/sources.list

deb http://ftp.be.debian.org/debian/ etch main non-free contrib

# deb http://ftp.be.debian.org/debian/ sarge main non-free contrib
# deb http://ftp.be.debian.org/debian/ unstable main non-free contrib
# deb-src http://ftp.be.debian.org/debian/ etch main non-free contrib

deb http://security.debian.org/ etch/updates main

# deb-src http://security.debian.org/ etch/updates main

14.6. alien
alien is experimental software that converts between rpm and deb package formats
(and others).

Below an example of how to use alien to convert an rpm package to a deb package.

paul@barry:~$ ls -l netcat*
-rw-r--r-- 1 paul paul 123912 2009-06-04 14:58 netcat-0.7.1-1.i386.rpm
paul@barry:~$ alien --to-deb netcat-0.7.1-1.i386.rpm
netcat_0.7.1-2_i386.deb generated
paul@barry:~$ ls -l netcat*

212
 Package management

-rw-r--r-- 1 paul paul 123912 2009-06-04 14:58 netcat-0.7.1-1.i386.rpm

-rw-r--r-- 1 root root 125236 2009-06-04 14:59 netcat_0.7.1-2_i386.deb

In real life, use the netcat tool provided by your distribution, or use the .deb file from
their website.

14.7. Downloading software

First and most important, whenever you download software, start by reading the
README file!

Normally the readme will explain what to do after download. You will probably
receive a .tar.gz or a .tgz file. Read the documentation, then put the compressed file in
a directory. You can use the following to find out where the package wants to install.
tar tvzpf $downloadedFile.tgz

You unpack them like with tar xzf, it will create a directory called
applicationName-1.2.3
tar xzf $applicationName.tgz

Replace the z with a j when the file ends in .tar.bz2. The tar, gzip and bzip2
commands are explained in detail later.

If you download a .deb file, then you'll have to use dpkg to install it, .rpm's can be
installed with the rpm command.

14.8. Compiling software

First and most important, whenever you download source code for installation, start
by reading the README file!

Usually the steps are always the same three : running ./configure followed by make
(which is the actual compiling) and then by make install to copy the files to their
proper location.

./configure
make
make install

14.9. Practice: Installing software

1. Find the Graphical application on all computers to add and remove applications.

2. Verify on both systems whether gcc is installed.

213
 Package management

3. Use aptitude or yum to search for and install the 'dict', 'samba' and 'wesnoth'
applications. Did you find all them all ?

4. Search the internet for 'webmin' and install it.

5. If time permits, uninstall Samba from the ubuntu machine, download the latest
version from samba.org and install it.

14.10. Solution: Installing software

1. Find the Graphical application on all computers to add and remove applications.

2. Verify on both systems whether gcc is installed.

dpkg -l | grep gcc

rpm -qa | grep gcc

3. Use aptitude or yum to search for and install the 'dict', 'samba' and 'wesnoth'
applications. Did you find all them all ?
aptitude search wesnoth (Debian, Ubuntu and family)

yum search wesnoth (Red Hat and family)

4. Search the internet for 'webmin' and install it.

Google should point you to webmin.com.

There are several formats available there choose .rpm, .deb or .tgz .

5. If time permits, uninstall Samba from the ubuntu machine, download the latest
version from samba.org and install it.

214
Chapter 15. Backup

Table of Contents
15.1. About tape devices ..................................................................................... 215
15.2. Compression ............................................................................................... 216
15.3. tar ............................................................................................................... 217
15.4. Backup Types ............................................................................................ 219
15.5. dump and restore ....................................................................................... 219
15.6. cpio ............................................................................................................. 220
15.7. dd ................................................................................................................ 220
15.8. split ............................................................................................................. 222
15.9. Practice backup .......................................................................................... 222

15.1. About tape devices

Don't forget that the name of a device strictly speaking has no meaning since the
kernel will use the major and minor number to find the hardware! See the man page
of mknod and the devices.txt file in the linux kernel source for more info.

15.1.1. SCSI tapes

On the official Linux device list (http://www.lanana.org/docs/device-list/) we find
the names for SCSI tapes (major 9 char). SCSI tape devices are located underneath /
dev/st and are numbered starting with 0 for the first tape device.

/dev/st0 First tape device

/dev/st1 Second tape device
/dev/st2 Third tape device

To prevent automatic rewinding of tapes, prefix them with the letter n.

/dev/nst0 First no rewind tape device

/dev/nst1 Second no rewind tape device
/dev/nst2 Third no rewind tape device

By default, SCSI tapes on linux will use the highest hardware compression that is
supported by the tape device. To lower the compression level, append one of the
letters l (low), m (medium) or a (auto) to the tape name.

/dev/st0l First low compression tape device

/dev/st0m First medium compression tape device
/dev/nst2m Third no rewind medium compression tape device

215
 Backup

15.1.2. IDE tapes

On the official Linux device list (http://www.lanana.org/docs/device-list/) we find
the names for IDE tapes (major 37 char). IDE tape devices are located underneath
/dev/ht and are numbered starting with 0 for the first tape device. No rewind and
compression is similar to SCSI tapes.

/dev/ht0 First IDE tape device

/dev/nht0 Second no rewind IDE tape device
/dev/ht0m First medium compression IDE tape device

15.1.3. mt
To manage your tapes, use mt (Magnetic Tape). Some examples.

To receive information about the status of the tape.

mt -f /dev/st0 status

To rewind a tape...
mt -f /dev/st0 rewind

To rewind and eject a tape...

mt -f /dev/st0 eject

To erase a tape...
mt -f /dev/st0 erase

15.2. Compression
It can be beneficial to compress files before backup. The two most popular tools for
compression of regular files on linux are gzip/gunzip and bzip2/bunzip2. Below you
can see gzip in action, notice that it adds the .gz extension to the file.

paul@RHELv4u4:~/test$ ls -l allfiles.tx*
-rw-rw-r-- 1 paul paul 8813553 Feb 27 05:38 allfiles.txt
paul@RHELv4u4:~/test$ gzip allfiles.txt
paul@RHELv4u4:~/test$ ls -l allfiles.tx*
-rw-rw-r-- 1 paul paul 931863 Feb 27 05:38 allfiles.txt.gz
paul@RHELv4u4:~/test$ gunzip allfiles.txt.gz
paul@RHELv4u4:~/test$ ls -l allfiles.tx*
-rw-rw-r-- 1 paul paul 8813553 Feb 27 05:38 allfiles.txt
paul@RHELv4u4:~/test$

In general, gzip is much faster than bzip2, but the latter one compresses a lot better.
Let us compare the two.

paul@RHELv4u4:~/test$ cp allfiles.txt bllfiles.txt

216
 Backup

paul@RHELv4u4:~/test$ time gzip allfiles.txt

real 0m0.050s
user 0m0.041s
sys 0m0.009s
paul@RHELv4u4:~/test$ time bzip2 bllfiles.txt

real 0m5.968s
user 0m5.794s
sys 0m0.076s
paul@RHELv4u4:~/test$ ls -l ?llfiles.tx*
-rw-rw-r-- 1 paul paul 931863 Feb 27 05:38 allfiles.txt.gz
-rw-rw-r-- 1 paul paul 708871 May 12 10:52 bllfiles.txt.bz2
paul@RHELv4u4:~/test$

15.3. tar
The tar utility gets its name from Tape ARchive. This tool will receive and send
files to a destination (typically a tape or a regular file). The c option is used to create
a tar archive (or tarfile), the f option to name/create the tarfile. The example below
takes a backup of /etc into the file /backup/etc.tar .

root@RHELv4u4:~# tar cf /backup/etc.tar /etc

root@RHELv4u4:~# ls -l /backup/etc.tar
-rw-r--r-- 1 root root 47800320 May 12 11:47 /backup/etc.tar
root@RHELv4u4:~#

Compression can be achieved without pipes since tar uses the z flag to compress with
gzip, and the j flag to compress with bzip2.

root@RHELv4u4:~# tar czf /backup/etc.tar.gz /etc

root@RHELv4u4:~# tar cjf /backup/etc.tar.bz2 /etc
root@RHELv4u4:~# ls -l /backup/etc.ta*
-rw-r--r-- 1 root root 47800320 May 12 11:47 /backup/etc.tar
-rw-r--r-- 1 root root 6077340 May 12 11:48 /backup/etc.tar.bz2
-rw-r--r-- 1 root root 8496607 May 12 11:47 /backup/etc.tar.gz
root@RHELv4u4:~#

The t option is used to list the contents of a tar file. Verbose mode is enabled with
v (also useful when you want to see the files being archived during archiving).

root@RHELv4u4:~# tar tvf /backup/etc.tar

drwxr-xr-x root/root 0 2007-05-12 09:38:21 etc/
-rw-r--r-- root/root 2657 2004-09-27 10:15:03 etc/warnquota.conf
-rw-r--r-- root/root 13136 2006-11-03 17:34:50 etc/mime.types
drwxr-xr-x root/root 0 2004-11-03 13:35:50 etc/sound/
...

To list a specific file in a tar archive, use the t option, added with the filename
(without leading /).

217
 Backup

root@RHELv4u4:~# tar tvf /backup/etc.tar etc/resolv.conf

-rw-r--r-- root/root 77 2007-05-12 08:31:32 etc/resolv.conf
root@RHELv4u4:~#

Use the x flag to restore a tar archive, or a single file from the archive. Remember
that by default tar will restore the file in the current directory.

root@RHELv4u4:~# tar xvf /backup/etc.tar etc/resolv.conf

etc/resolv.conf
root@RHELv4u4:~# ls -l /etc/resolv.conf
-rw-r--r-- 2 root root 40 May 12 12:05 /etc/resolv.conf
root@RHELv4u4:~# ls -l etc/resolv.conf
-rw-r--r-- 1 root root 77 May 12 08:31 etc/resolv.conf
root@RHELv4u4:~#

You can preserve file permissions with the p flag. And you can exclude directories
or file with --exclude.

root ~# tar cpzf /backup/etc_with_perms.tgz /etc

root ~# tar cpzf /backup/etc_no_sysconf.tgz /etc --exclude /etc/sysconfig
root ~# ls -l /backup/etc_*
-rw-r--r-- 1 root root 8434293 May 12 12:48 /backup/etc_no_sysconf.tgz
-rw-r--r-- 1 root root 8496591 May 12 12:48 /backup/etc_with_perms.tgz
root ~#

You can also create a text file with names of files and directories to archive, and then
supply this file to tar with the -T flag.

root@RHELv4u4:~# find /etc -name *.conf > files_to_archive.txt

root@RHELv4u4:~# find /home -name *.pdf >> files_to_archive.txt
root@RHELv4u4:~# tar cpzf /backup/backup.tgz -T files_to_archive.txt

The tar utility can receive filenames from the find command, with the help of xargs.

find /etc -type f -name "*.conf" | xargs tar czf /backup/confs.tar.gz

You can also use tar to copy a directory, this is more efficient than using cp -r.

(cd /etc; tar -cf - . ) | (cd /backup/copy_of_etc/; tar -xpf - )

Another example of tar, this copies a directory securely over the network.

(cd /etc;tar -cf - . )|(ssh user@srv 'cd /backup/cp_of_etc/; tar -xf - ')

tar can be used together with gzip and copy a file to a remote server through ssh

cat backup.tar | gzip | ssh [email protected] "cat - > backup.tgz"

218
 Backup

Compress the tar backup when it is on the network, but leave it uncompressed at the
destination.

cat backup.tar | gzip | ssh [email protected] "gunzip|cat - > backup.tar"

Same as the previous, but let ssh handle the compression

cat backup.tar | ssh -C [email protected] "cat - > backup.tar"

15.4. Backup Types

Linux uses multilevel incremental backups using distinct levels. A full backup is
a backup at level 0. A higher level x backup will include all changes since the last
level x-1 backup.

Suppose you take a full backup on Monday (level 0) and a level 1 backup on Tuesday,
then the Tuesday backup will contain all changes since Monday. Taking a level 2
on Wednesday will contain all changes since Tuesday (the last level 2-1). A level 3
backup on Thursday will contain all changes since Wednesday (the last level 3-1).
Another level 3 on Friday will also contain all changes since Wednesday. A level 2
backup on Saturday would take all changes since the last level 1 from Tuesday.

15.5. dump and restore

While dump is similar to tar, it is also very different because it looks at the file
system. Where tar receives a lists of files to backup, dump will find files to backup
by itself by examining ext2. Files found by dump will be copied to a tape or regular
file. In case the target is not big enough to hold the dump (end-of-media), it is broken
into multiple volumes.

Restoring files that were backed up with dump is done with the restore command.
In the example below we take a full level 0 backup of two partitions to a SCSI tape.
The no rewind is mandatory to put the volumes behind each other on the tape.

dump 0f /dev/nst0 /boot

dump 0f /dev/nst0 /

Listing files in a dump archive is done with dump -t, and you can compare files with
dump -C.

You can omit files from a dump by changing the dump attribute with the chattr
command. The d attribute on ext will tell dump to skip the file, even during a full
backup. In the following example, /etc/hosts is excluded from dump archives.

chattr +d /etc/hosts

219
 Backup

To restore the complete file system with restore, use the -r option. This can be useful
to change the size or block size of a file system. You should have a clean file system
mounted and cd'd into it. Like this example shows.

mke2fs /dev/hda3
mount /dev/hda3 /mnt/data
cd /mnt/data
restore rf /dev/nst0

To extract only one file or directory from a dump, use the -x option.

restore -xf /dev/st0 /etc

15.6. cpio
Different from tar and dump is cpio (Copy Input and Output). It can be used to receive
filenames, but copies the actual files. This makes it an easy companion with find!
Some examples below.

find sends filenames to cpio, which puts the files in an archive.

find /etc -depth -print | cpio -oaV -O archive.cpio

The same, but compressed with gzip

find /etc -depth -print | cpio -oaV | gzip -c > archive.cpio.gz

Now pipe it through ssh (backup files to a compressed file on another machine)

find /etc -depth -print|cpio -oaV|gzip -c|ssh server "cat - > etc.cpio.gz"

find sends filenames to cpio | cpio sends files to ssh | ssh sends files to cpio 'cpio
extracts files'
find /etc -depth -print | cpio -oaV | ssh user@host 'cpio -imVd'

the same but reversed: copy a dir from the remote host to the local machine
ssh user@host "find path -depth -print | cpio -oaV" | cpio -imVd

15.7. dd

15.7.1. About dd
Some people use dd to create backups. This can be very powerful, but dd backups
can only be restored to very similar partitions or devices. There are however a lot of
useful things possible with dd. Some examples.

220
 Backup

15.7.2. Create a CDROM image

The easiest way to create a .ISO file from any CD. The if switch means Input File,
of is the Output File. Any good tool can burn a copy of the CD with this .ISO file.
dd if=/dev/cdrom of=/path/to/cdrom.ISO

15.7.3. Create a floppy image

A little outdated maybe, but just in case : make an image file from a 1.44MB floppy.
Blocksize is defined by bs, and count contains the number of blocks to copy.
dd if=/dev/floppy of=/path/to/floppy.img bs=1024 count=1440

15.7.4. Copy the master boot record

Use dd to copy the MBR (Master Boot Record) of hard disk /dev/hda to a file.
dd if=/dev/hda of=/MBR.img bs=512 count=1

15.7.5. Copy files

This example shows how dd can copy files. Copy the file summer.txt to
copy_of_summer.txt .
dd if=~/summer.txt of=~/copy_of_summer.txt

15.7.6. Image disks or partitions

And who needs ghost when dd can create a (compressed) image of a partition.

dd if=/dev/hdb2 of=/image_of_hdb2.IMG
dd if=/dev/hdb2 | gzip > /image_of_hdb2.IMG.gz

15.7.7. Create files of a certain size

dd can be used to create a file of any size. The first example creates a one MEBIbyte
file, the second a one MEGAbyte file.

dd if=/dev/zero of=file1MB count=1024 bs=1024

dd if=/dev/zero of=file1MB count=1000 bs=1024

15.7.8. CDROM server example

And there are of course endless combinations with ssh and bzip2. This example puts
a bzip2 backup of a cdrom on a remote server.

221
 Backup

dd if=/dev/cdrom |bzip2|ssh user@host "cat - > /backups/cd/cdrom.iso.bz2"

15.8. split
The split command is useful to split files into smaller files. This can be useful to fit
the file onto multiple instances of a medium too small to contain the complete file.
In the example below, a file of size 5000 bytes is split into three smaller files, with
maximum 2000 bytes each.

paul@laika:~/test$ ls -l
total 8
-rw-r--r-- 1 paul paul 5000 2007-09-09 20:46 bigfile1
paul@laika:~/test$ split -b 2000 bigfile1 splitfile.
paul@laika:~/test$ ls -l
total 20
-rw-r--r-- 1 paul paul 5000 2007-09-09 20:46 bigfile1
-rw-r--r-- 1 paul paul 2000 2007-09-09 20:47 splitfile.aa
-rw-r--r-- 1 paul paul 2000 2007-09-09 20:47 splitfile.ab
-rw-r--r-- 1 paul paul 1000 2007-09-09 20:47 splitfile.ac

15.9. Practice backup

!! Careful with tar options and the position of the backup file, mistakes can destroy
your system!!

1. Create a directory (or partition if you like) for backups. Link (or mount) it under /
mnt/backup.

2a. Use tar to backup /etc in /mnt/backup/etc_date.tgz, the backup must be gzipped.
(Replace date with the current date)

2b. Use tar to backup /bin to /mnt/backup/bin_date.tar.bz2, the backup must be

bzip2'd.

2c. Choose a file in /etc and /bin and verify with tar that the file is indeed backed up.

2d. Extract those two files to your home directory.

3a. Create a backup directory for your neighbour, make it accessible under /mnt/
neighbourName

3b. Combine ssh and tar to put a backup of your /boot on your neighbours computer
in /mnt/YourName

4a. Combine find and cpio to create a cpio archive of /etc.

4b. Choose a file in /etc and restore it from the cpio archive into your home directory.

222
 Backup

5. Use dd and ssh to put a backup of the master boot record on your neighbours
computer.

6. (On the real computer) Create and mount an ISO image of the ubuntu cdrom.

7. Combine dd and gzip to create a 'ghost' image of one of your partitions on another
partition.

8. Use dd to create a five megabyte file in ~/testsplit and name it biggest. Then split
this file in smaller two megabyte parts.
mkdir testsplit

dd if=/dev/zero of=~/testsplit/biggest count=5000 bs=1024

split -b 2000000 biggest parts

223
Chapter 16. Performance monitoring

Table of Contents
16.1. About Monitoring ...................................................................................... 224
16.2. top ............................................................................................................... 224
16.3. free ............................................................................................................. 225
16.4. watch .......................................................................................................... 225
16.5. vmstat ......................................................................................................... 225
16.6. iostat ........................................................................................................... 226
16.7. mpstat ......................................................................................................... 227
16.8. sadc and sar ............................................................................................... 227
16.9. ntop ............................................................................................................. 228
16.10. iftop .......................................................................................................... 228

16.1. About Monitoring

Monitoring means obtaining information about the utilization of memory, CPU
power, bandwidth and storage. You should start monitoring your system as soon as
possible, to be able to create a baseline. Make sure that you get to know your system.
Boys, just give your computer a girls name and get to know her. The baseline is
important, it allows you to see a steady growth in CPU utilization or a steady decline
in free disk space. It will allow you to plan for scaling up or scaling out.

Let us look at some tools that go beyond ps fax, df -h, lspci, fdisk -l and du -sh.

16.2. top
To start monitoring, you can use top. This tool will monitor Memory, CPU and
running processes. Top will automatically refresh. Inside top you can use many
commands, like k to kill processes, or t and m to toggle displaying task and memory
information, or the number 1 to have one line per cpu, or one summary line for all
cpu's.

top - 12:23:16 up 2 days, 4:01, 2 users, load average: 0.00, 0.00, 0.00
Tasks: 61 total, 1 running, 60 sleeping, 0 stopped, 0 zombie
Cpu(s): 0.3% us, 0.5% sy, 0.0% ni, 98.9% id, 0.2% wa, 0.0% hi, 0.0% si
Mem: 255972k total, 240952k used, 15020k free, 59024k buffers
Swap: 524280k total, 144k used, 524136k free, 112356k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND

1 root 16 0 2816 560 480 S 0.0 0.2 0:00.91 init
2 root 34 19 0 0 0 S 0.0 0.0 0:00.01 ksoftirqd/0
3 root 5 -10 0 0 0 S 0.0 0.0 0:00.57 events/0
4 root 5 -10 0 0 0 S 0.0 0.0 0:00.00 khelper
5 root 15 -10 0 0 0 S 0.0 0.0 0:00.00 kacpid
16 root 5 -10 0 0 0 S 0.0 0.0 0:00.08 kblockd/0

224
 Performance monitoring

26 root 15 0 0 0 0 S 0.0 0.0 0:02.86 pdflush

...

You can customize top to display the columns of your choice, or to display only the
processes that you find interesting.
[paul@RHELv4u3 ~]$ top p 3456 p 8732 p 9654

16.3. free
The free command is common on Linux to monitor free memory. You can use free
to display information every x seconds, but the output is not ideal.

[paul@RHELv4u3 gen]$ free -om -s 10

total used free shared buffers cached
Mem: 249 222 27 0 50 109
Swap: 511 0 511

total used free shared buffers cached

Mem: 249 222 27 0 50 109
Swap: 511 0 511

[paul@RHELv4u3 gen]$

16.4. watch
It might be more interesting to combine free with the watch program. This program
can also run commands with a delay, and can highlight changes (with the -d switch).

[paul@RHELv4u3 ~]$ watch -d -n 3 free -om

...
Every 3.0s: free -om Sat Jan 27 12:13:03 2007

total used free shared buffers cached

Mem: 249 230 19 0 56 109
Swap: 511 0 511

16.5. vmstat
To monitor CPU, disk and memory statistics in one line there is vmstat. The
screenshot below shows vmstat running every two seconds 100 times (or until the
Ctrl-C). Below the r, you see the number of processes waiting for the CPU, sleeping
processes go below b. Swap usage (swpd) stayed constant at 144 kilobytes, free
memory dropped from 16.7MB to 12.9MB. See man vmstat for the rest

[paul@RHELv4u3 ~]$ vmstat 2 100

procs ----------memory--------- --swap-- ---io--- --system-- ---cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa

225
 Performance monitoring

0 0 144 16708 58212 111612 0 0 3 4 75 62 0 1 99 0

0 0 144 16708 58212 111612 0 0 0 0 976 22 0 0 100 0
0 0 144 16708 58212 111612 0 0 0 0 958 14 0 1 99 0
1 0 144 16528 58212 111612 0 0 0 18 1432 7417 1 32 66 0
1 0 144 16468 58212 111612 0 0 0 0 2910 20048 4 95 1 0
1 0 144 16408 58212 111612 0 0 0 0 3210 19509 4 97 0 0
1 0 144 15568 58816 111612 0 0 300 1632 2423 10189 2 62 0 36
0 1 144 13648 60324 111612 0 0 754 0 1910 2843 1 27 0 72
0 0 144 12928 60948 111612 0 0 312 418 1346 1258 0 14 57 29
0 0 144 12928 60948 111612 0 0 0 0 977 19 0 0 100 0
0 0 144 12988 60948 111612 0 0 0 0 977 15 0 0 100 0
0 0 144 12988 60948 111612 0 0 0 0 978 18 0 0 100 0

[paul@RHELv4u3 ~]$

16.6. iostat
The iostat tool can display disk and cpu statistics. The -d switch below makes iostat
only display disk information (500 times every two seconds). The first block displays
statistics since the last reboot.

[paul@RHELv4u3 ~]$ iostat -d 2 500

Linux 2.6.9-34.EL (RHELv4u3.localdomain) 01/27/2007

Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn

hdc 0.00 0.01 0.00 1080 0
sda 0.52 5.07 7.78 941798 1445148
sda1 0.00 0.01 0.00 968 4
sda2 1.13 5.06 7.78 939862 1445144
dm-0 1.13 5.05 7.77 939034 1444856
dm-1 0.00 0.00 0.00 360 288

Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn

hdc 0.00 0.00 0.00 0 0
sda 0.00 0.00 0.00 0 0
sda1 0.00 0.00 0.00 0 0
sda2 0.00 0.00 0.00 0 0
dm-0 0.00 0.00 0.00 0 0
dm-1 0.00 0.00 0.00 0 0
...
[paul@RHELv4u3 ~]$

You can have more statistics using iostat -d -x, or display only cpu statistics with
iostat -c.

[paul@RHELv4u3 ~]$ iostat -c 5 500

Linux 2.6.9-34.EL (RHELv4u3.localdomain) 01/27/2007

avg-cpu: %user %nice %sys %iowait %idle

0.31 0.02 0.52 0.23 98.92

avg-cpu: %user %nice %sys %iowait %idle

0.62 0.00 52.16 47.23 0.00

avg-cpu: %user %nice %sys %iowait %idle

2.92 0.00 36.95 60.13 0.00

226
 Performance monitoring

avg-cpu: %user %nice %sys %iowait %idle

0.63 0.00 36.63 62.32 0.42

avg-cpu: %user %nice %sys %iowait %idle

0.00 0.00 0.20 0.20 99.59

[paul@RHELv4u3 ~]$

16.7. mpstat
On multi-processor machines, mpstat can display statistics for all, or for a selected
cpu.

paul@laika:~$ mpstat -P ALL

Linux 2.6.20-3-generic (laika) 02/09/2007

CPU %user %nice %sys %iowait %irq %soft %steal %idle intr/s
all 1.77 0.03 1.37 1.03 0.02 0.39 0.00 95.40 1304.91
0 1.73 0.02 1.47 1.93 0.04 0.77 0.00 94.04 1304.91
1 1.81 0.03 1.27 0.13 0.00 0.00 0.00 96.76 0.00
paul@laika:~$

16.8. sadc and sar

The sadc tool writes system utilization data to /var/log/sa/sa??, where ?? is replaced
with the current day of the month. By default, cron runs the sal script every 10
minutes, the sal script runs sadc for one second. Just before midnight every day, cron
runs the sa2 script, which in turn invokes sar. The sar tool will read the daily data
generated by sadc and put it in /var/log/sa/sar??. These sar reports contain a lot of
statistics.

You can also use sar to display a portion of the statistics that were gathered. Like this
example for cpu statistics.

[paul@RHELv4u3 sa]$ sar -u | head

Linux 2.6.9-34.EL (RHELv4u3.localdomain) 01/27/2007

12:00:01 AM CPU %user %nice %system %iowait %idle

12:10:01 AM all 0.48 0.01 0.60 0.04 98.87
12:20:01 AM all 0.49 0.01 0.60 0.06 98.84
12:30:01 AM all 0.49 0.01 0.64 0.25 98.62
12:40:02 AM all 0.44 0.01 0.62 0.07 98.86
12:50:01 AM all 0.42 0.01 0.60 0.10 98.87
01:00:01 AM all 0.47 0.01 0.65 0.08 98.80
01:10:01 AM all 0.45 0.01 0.68 0.08 98.78
[paul@RHELv4u3 sa]$

There are other useful sar options, like sar -I PROC to display interrupt activity per
interrupt and per CPU, or sar -r for memory related statistics. Check the manual page
of sar for more.

227
 Performance monitoring

16.9. ntop
The ntop tool is not present in default Red Hat installs. Once run, it will generate a
very extensive analysis of network traffic in html on http://localhost:3000 .

16.10. iftop
The iftop tool will display bandwidth by socket statistics for a specific network
device. Not available on default Red Hat servers.

1.91Mb 3.81Mb 5.72Mb 7.63Mb 9.54Mb

--------------|-------------|--------------|-------------|--------|----
laika.local => barry 4.94Kb 6.65Kb 69.9Kb
<= 7.41Kb 16.4Kb 766Kb
laika.local => ik-in-f19.google.com 0b 1.58Kb 14.4Kb
<= 0b 292b 41.0Kb
laika.local => ik-in-f99.google.com 0b 83b 4.01Kb
<= 0b 83b 39.8Kb
laika.local => ug-in-f189.google.com 0b 42b 664b
<= 0b 42b 406b
laika.local => 10.0.0.138 0b 0b 149b
<= 0b 0b 256b
laika.local => 224.0.0.251 0b 0b 86b
<= 0b 0b 0b
laika.local => ik-in-f83.google.com 0b 0b 39b
<= 0b 0b 21b

228