11g R1 Rac – Solaris – Installation Checklist

Important Documents
Clusterware installation guide for solaris operating system
Real Application Clusters Installation Guide for Unix and Linux
Release Notes for the solaris operating system
Storage Administration Guide
This checklist strives to document the minimum requirements to be met in order to
accomplish a successful Oracle Rac Installation.This checklist can be used as a companion
to the documents in the previous section. Please be aware that there could be requirements
documented in the Official Oracle documentation that is not present in this document.
Hardware requirements
Procure Certified Hardware Confirm with the hardware vendor (server
and disk) that the combination of hardware
you have procured, is certified to run with
Oracle RAC (ie the vendor has tested the
configuration and have successful production
implementations). Components include but
are not limited to
– Server make and model
– Type of HBA
– HBA driver to be used
– O/S kernel version to be used
– Multipathing driver to be used.
Make sure that you have redundant network
interface cards to be used for the private nic
and the public nic.

The private interconnect should be connected

using a switch (Crossover cable not
supported).
Technology compatibility matrix Review and familiarize yourself with the Rac
technology compatibility matrix.
Memory Requirements Atleast 1gb of physical ram
Solaris Containers Oracle Real Application clusters is only
certified to run in Global Containers (If you
are using solaris containers) (Cannot run in
local containers).
Host bus adapters Ensure that you have atleast 2 HBA's on
each node (So that you have redundant
access paths to the disk subsystem)
Network Requirements
Network interface requirements Each node must have atleast 2 network
 interfaces.

We recommend having atleast 4 network

interfaces. 2 for the private network and 2 for
the public network.
NIC Bonding If you have more than one network interface
card for each type of network interfae (private
and public), then bond a pair of interfaces
together using IPMP.
Refer to the following metalink notes for
details on how to setup ipmp
368464.1
283107.1

Interface Names The public interface names associated with

the network adapters for each network must
be the same on all nodes, and the private
interface names associated with the network
adaptors should be the same on all nodes.
For example: With a two-node cluster, you
cannot configure network adapters on node1
with eth0 as the public interface, but on
node2 have eth1 as the public interface.
Public interface names must be the same, so
you must configure eth0 as public on both
nodes. You should configure the private
interfaces on the same network adapters as
well. If eth1 is the private interface for node1,
then eth1 should be the private interface for
node2.

Network Protocol (Public Nic) For the public network, each network adapter
must support TCP/IP.

Network Protocol (Private Nic) For the private network, the interconnect
must support the user datagram protocol
(UDP) using high-speed network adapters
and switches that support TCP/IP (Gigabit
Ethernet or better recommended).

Public IP address Assign One IP address with an associated

host name (or network name) registered in
the DNS for the public interface. Record the
host name and IP address in the system
 hosts file also, /etc/hosts.

Each node needs a public IP address

This can be an interface bonded using IPMP.

Virtual IP address Assign One virtual IP (VIP) address with an

associated host name registered in a DNS.
Record the host name and VIP address in the
system hosts file, /etc/hosts.
Each node needs a virtual IP address.
Select an address for your VIP that meets the
following requirements:
– The IP address and host name are
currently unused (it can be registered in a
DNS, but should not be accessible by a
ping command)
– The VIP is on the same subnet as your
public interface

Private IP address Configure the private interconnect network

interface cards to have a private node name
and a private IP address.

This can be an interface bonded using IPMP.

Choose a private IP address that is in the

address range 10.*.*.* or 192.168.*.*

The private IP address should be on a

separate subnet than the public IP address.

Oracle strongly recommends using a

physically separate, private network (You can
use a vlan on a shared switch if you must).
You should ensure that the private IP
addresses are reachable only by the cluster
member nodes.

Operating System Requirements

Install Correct O/S version Install Solaris 10, with the latest patch cluster.
Swap space Equal to physical memory if physical memory
< 8gb.
.75 x physical memory if physical memory >
8gb
Temp space Atleast 400mb of space in /tmp, preferably
atleast 2gb
Network Time protocol Configure network time protocol (ntp), so that
the times on the rac hosts are synchronized
with the same clock. Ensure that each
member node of the cluster is set as closely
as possible to the same date and time.
Operating system packages required SUNWarc
SUNWbtool
SUNWhea
SUNWlibC
SUNWlibm
SUNWlibms
SUNWsprot
SUNWtoo
SUNWi1of
SUNWi1cs
SUNWi15cs
SUNWxwfnt

Make sure that the package SUNWxcu4 is

also installed on the machine.
If you are using sun cluster SUNWscucm
SUNWudlmr
SUNWudlm
If you are using Clustered solaris volume SUNWscmd
manager
If you are using sun cluster and hardware raid SUNWschwr
If you use any of the following oracle featuers Install Sun ONE Studio 11 (C and C++ 5.8)
Pro*C/C++, Oracle Call Interface, Oracle C++
Call Interface, Oracle XML Developer's Kit
(XDK)
If you use the oracle ODBC driver Install gcc 3.4.2
Operating system Patches Required 127111-02 SunOS 5.10: libc patch
Configure kernel parameters Configure the following kernel parameters
using both /etc/system and resource controls.

Use the instructions in the section

“Configuring kernel parameters on Solaris 10”
 from the link
http://download.oracle.com/docs/cd/B28359_
01/install.111/b28262/presolar.htm#BABCHA
ED

noexec_user_stack 1
semsys:seminfo_semmni 100
semsys:seminfo_semmns 1024
semsys:seminfo_semmsl 256
semsys:seminfo_semvmx 32767
shmsys:shminfo_shmmax 4294967295
shmsys:shminfo_shmmni 100
Set UDP parameters udp_xmit_hiwat=65536
udp_recv_hiwat=65536
Configure SSH on all cluster nodes Ssh should be configured on all nodes of the
rac cluster, for the oracle user, and you
should be able to ssh to every node without
being prompted for the password.

Make sure that there is no banner displayed

when you login (or ssh) as oracle.

Softlink ssh and scp to /usr/bin from

/usr/local/bin
Add all relavent node names and ip Check /etc/hosts to verify the following entries
addresses in the /etc/hosts exist

127.0.0.1 localhost localhost.localdomain

The /etc/hosts should have the following

entries for each database node
– Your public node name, public node
name.domainname
– Your private node name,private node
name.domainname
– Your vip node name, vip node
name.domainname.

Create the dba group Create the group named dba with the same
gid on each server
Create Oracle Software Owner Create the Unix user oracle, on each node
and assign it to the group dba. Setup bash or
ksh to be the default login shell
Set shell limits for the oracle user rlim_fd_max 65536
maxuprc 16384
Set umask Set the default umask for the oracle user to
be 022
Allocate a local disk mount This mount point should be different from
point for the CRS_HOME that allocated for ORACLE_HOME and
ASM_HOME (eg: /crs).

Allocate atleast 1gb of space for this mount

point.
Allocate a local disk mount This mount point should be different from
point for the ASM_HOME that allocated for ORACLE_HOME and
CRS_HOME
(eg: /u01/asm). Allocate atleast 3gb of
space for this mount point.
Allocate a local disk mount This mount point should be different from
point for the ORACLE_HOME that allocated for the CRS_HOME (eg:
/u01/app). Allocate atleast 3Gb of space for
this directory.
Allocate three sharable raw This device should preferably mirrored at
devices for the voting disk. the hardware level. Size 280Mb. Configure
this device by creating a partition on the
lun, that skips the first 1Mb on the LUN.
For best availability, keep these three luns
on physically separate disks.

The devices can be emc powerpath devices.

The device names that point to the same lun,

should be exactly the same on both nodes.

Eg: if /dev/rdsk/xxdisk1 points to lun1 on

node 1 then /dev/rdsk/xxdisk1 has to point to
lun1 on the second node.

After creating the devices run the following

command to
make absolutely sure that you have 280Mb of
free space.
dd if=/dev/zero of=/dev/raw/raw1 bs=+1M
count=280 (Substitute raw1 with your
rawdevice name)

Change the ownership and permissions of

the disks

chown root:dba /dev/rdsk/c3t4d12s6

chmod 660 /dev/rdsk/c3t4d12s6
Allocate two sharable raw This device should preferably mirrored at
devices for the cluster the hardware level.Size 280 Mb.Configure
registry. this device by creating a partition on the
lun, that skips the first 1Mb on the LUN.
Device should preferably mirrored at the
hardware level.

The devices can be emc powerpath devices.

Eg: if /dev/rdsk/xxdisk1 points to lun1 on

node 1 then /dev/rdsk/xxdisk1 has to point to
lun1 on the second node.

After creating the devices run the following

command to
make absolutely sure that you have 100Mb of
free space.
dd if=/dev/zero of=/dev/raw/raw1 bs=+1M
count=280 (Substitute raw1 with your
rawdevice name)

Change the ownership and permissions on

the devices

chown oracle:dba /dev/rdsk/c3t4d15s6

chmod 660 /dev/rdsk/c3t4d15s6

Allocate devices for the ASM Data disk group Take the total amount of space you require to
place your database files (data, index,
system, undo , temp, redo) and divide this up
to create as many luns as possible. Rule of
thumb is to present at least 4 luns and have
the number of luns as close to the number of
spindles you are allocating.

Present these luns to the operating system.

Eg: if you need to present 500gb of space

then create and present 10 luns of 50gb
each.

These luns can be emc powerpath devices.

Eg: if /dev/rdsk/xxdisk1 points to lun1 on

node 1 then /dev/rdsk/xxdisk1 has to point to
lun1 on the second node.
 After the powerpath devices are configured,
use fdisk to create a partition of the disk,
skipping the first 1mb of the disk. This new
partition is what ASM will use.

Change the ownership and permissions on

the devices

chown oracle:dba /dev/rdsk/c3t4d15s6

chmod 660 /dev/rdsk/c3t4d15s6
Create the devices for the flash recovery area Take the total amount of space you require to
disk group place your Flash recovery area and divide
this up to create as many luns as possible.

Rule of thumb is to present at least 4 luns

and have the number of luns as close to the
number of spindles you are allocating.

Present these luns to the operating system.

Eg: if you need to present 500gb of space

then create and present 10 luns of 50gb
each.

These luns can be emc powerpath devices.

Eg: if /dev/rdsk/xxdisk1 points to lun1 on

node 1 then /dev/rdsk/xxdisk1 has to point to
lun1 on the second node.

After the powerpath devices are configured,

use fdisk to create a partition of the disk,
skipping the first 1mb of the disk. This new
partition is what ASM will use.

Change the ownership and permissions on

the devices

chown oracle:dba /dev/rdsk/c3t4d15s6

chmod 660 /dev/rdsk/c3t4d15s6
Download latest cluster verification utility Cluster Verification Utility – Solaris
(cvu)
Download Oracle Software Download
solaris.sparc64_11gR1_database_1013.zip
solaris.sparc64_11gR1_clusterware.zip
Download Oracle Patchset Download 11.1.0.7, patch number 6890831
from metalink
Run Cluster Verification Utility and send me As the oracle user
the output. ./runcluvfy.sh stage -pre crsinst -n
node1,node2
./runcluvfy.sh comp nodecon -n node1,node2
./runcluvfy.sh stage –post hwos –n
node1,node2
./runcluvfy.sh comp sys -n node1,node2 -p
crs
-osdba dba -orainv dba

Replace the node names with your actual

public node names.

Xwindows client Make sure that, on the desktop from which

you are going to login to the servers and
perform the installation, that you have a good
xwindows client like exceed. (Open source
clients like xming and/or xmanager will not
work). If you are using reflections, then make
sure that you have the latest version of
reflections.