ocp_logo Creating Oracle 9iR2 (9.2.0.4) Real Application Cluster (RAC) Database on Oracle Cluster File System (OCFS)
By Bhavin Hingu
<<PREV                      <<HOME>>                      NEXT>>

This document explains the step by step process of Creating Oracle 9iR2 Real Application Cluster (RAC) database on OCFS storage option.
In the Pre-Installation task, I have formated /dev/sda2 (mounted on /u02/oradata/ocfs) with OCFS. I am going to use the same mount point to
create the CLustered Database RacDB on OCFS. If you want to create the Clustered database on Raw Devices/partitions, Please goto Creating
Clustered Database using Raw Devices / LVM.

Creating Oracle 9iR2 (9.2.0.4) Real Application Cluster
(RAC) Database on OCFS:

Before you begin, Please make sure that ClusterManager (oracm), Global Service Daemon (gsd) is running. Also initialize the
Server Configuration repository using srvconfig -init  command.

Connect as root and  start the Cluster Manager.

[root@node1-pub root]# $ORACLE_HOME/oracm/bin/ocmstart.sh
oracm </dev/null 2>&1 >/u01/app/oracle/product/9.2.0/oracm/log/cm.out &
[root@node1-pub root]# ps -ef | grep oracm
root      2845     1  0 19:17 pts/0    00:00:00 oracm
root      2847  2845  0 19:17 pts/0    00:00:00 oracm
root      2848  2847  0 19:17 pts/0    00:00:00 oracm
root      2849  2847  0 19:17 pts/0    00:00:00 oracm
root      2850  2847  0 19:17 pts/0    00:00:00 oracm
root      2851  2847  0 19:17 pts/0    00:00:00 oracm
root      2852  2847  0 19:17 pts/0    00:00:00 oracm
root      2853  2847  0 19:17 pts/0    00:00:00 oracm
root      2854  2847  0 19:17 pts/0    00:00:00 oracm
root      2858  2847  0 19:17 pts/0    00:00:00 oracm
root      2860  2701  0 19:17 pts/0    00:00:00 grep oracm

Connect as oracle and start  the gsd.

[oracle@node1-pub oracle]$ gsdctl start
Successfully started GSD on local node

NOTE: Do not start the gsd as root. If you do this then you will get the below error message when you  use srvctl utility to start/stop
database/instances. If you get the below error then, stop the gsd and restart it again using oracle user. Even after doing that if you
continue getting that error, contact Oracle Support for further assistant.

[root@node2-pub root]# gsdctl start
Successfully started GSD on local node
[oracle@node2-pub oracle]# srvctl start database -d RacDB
PRKP-1007 : Failed to start all the listeners associated with all the instances of cluster database RacDB
PRKP-1005 : Failed to start up cluster database RacDB
ORA-01031: insufficient privileges
uction on Tue Jan 3 19:18:47 2006
 
Copyright (c) 1982, 2002, Oracle Corporation.  All rights reserved.
ORA-01005: null password given; logon denied
connect_string>] | /
06
 
Copyright (c) 1982, 2002, Oracle Corporation.  All rights reserved.
ORA-01031: insufficient privileges
uction on Tue Jan 3 19:18:55 2006
 
Copyright (c) 1982, 2002, Oracle Corporation.  All rights reserved.
ORA-01005: null password given; logon denied
connect_string>] | /
06


Now initialize the Server Configuratio Repository on all the nodes.

srvconfig -init

Follow These Steps to Create Clustered Database on OCFS using dbca.

[oracle@node1-pub oracle]# dbca -datafileDestination /u02/oradata/ocfs

If you do not see this screen, then you need to make sure that oracm service is running under all the nodes.

1


Select "Oracle cluster database" option and Click Next.
2

Select "Create a database" and Click Next.

3

Select All the nodes and then Click Next. Make sure that you see all the Nodes in this window  before
 you continue to the further steps. If you see any missing nodes, check that oracm is running on all those nodes.

4

Select whatever option you want that best fits your application. I choose general purpose for the experimental
database.

5

I Entered RacDB, RacDB as Global Database name and SID Prefix.

6

I selected dedicated server mode and clicked Next.

7

Adjust the memory parameter as my server has only 512 MB ram, so I had to adjusted these parameter to
set the SGA around 180MB.

8

9

Here you need to adjust the Server Parameter File path to one that you have used for ocfs mountpoint.
It is '/u02/oradata/ocfs' in my case.

10

Check whether the datafiles, controlfiles and redologs will be created under the right mountpoint and
so on the right file system (ocfs). In my case it should be /u02/oradata/ocfs.

11

12

13

Click Finish.

14


15

Enter the password for SYS, System oraclke database users and Click Exit.

Confirm that database files are created on the ocfs file system and has all the nodes registerd with all the files.

Connect as root and execute ocfstool command. You see that under the "File Listing" tab, a new folder RacDB is created.

18

Open the RacDB folder and check whether all the files are assigned to all the nodes in the cluster.

19

If any node is missing, then the database will not be opened by that node and you will get the below error message
wheh you try to open the database on that node. I got this error when dbca tried to startup the database on the another node
at the very last step and could not open it and failed with this error. I then looked at the database filed using ocfstool to
and found that the second node was not registered to these files. I cleanly shutdown database, all the Cluster Services
and finally rebooted all the nodes. Then I restarted all the Cluster Services (oracm, gsd) and then open the database manually using sqlplus on all the nodes one by one. At this time, the missing nodes were registered automatically to the database files and finally I was successfully able to open the database without any kind of errors.

PRKP-1003: Startup operation partially failed.

21

Check the status and configuration of the new database using srvctl utility.

srvctl start database -d RacDB
srvctl status database -d RacDB

20

Please go to next step of configure TAF of RAC.

<<PREV                      <<HOME>>                      NEXT>>