Multiple ORACLE_HOMEs for 9iRAC on Linux

From: Hemant K Chitale <hkchital@xxxxxxxxxxxxxx>
To: oracle-l@xxxxxxxxxxxxx
Date: Sun, 19 Dec 2004 15:15:36 +0800

I have been trying to install additional ORACLE_HOMEs to run seperate 
Database environments
on an existing 9iRAC Linux Cluster.
The Oracle Cluster Manager ["CM"] can be running only from the first 
installed ORACLE_HOME.
However, I want to use additional ORACLE_HOMEs to run seperate databases.

10g, I believe, does require the CRS to be a seperate ORACLE_HOME.

On Tru64, I had no problems installing a seperate ORACLE_HOME owned by a 
seperate Unix account.
The second installation could detect the cluster.

However, on Linux, as the ClusterManager actually runs from the first 
ORACLE_HOME, when installing
a new ORACLE_HOME the Installer cannot detect that it is a 
cluster.  Apparently, it does not use
the "ps -ef" commands to search for a running CM.  It just runs the 
"lsnodes" command.  As there is
no "lsnodes" in the new ORACLE_HOME/bin, it thinks that I am not running a 
cluster.
I got around that by copying "lsnodes"  {and a corresponding library} from 
the CM ORACLE_HOME to
the new ORACLE_HOME.

After having detected a cluster and installed the RAC option, and then 
applied the 9.2.0.5 patchset
correctly,
I then manually created the database
[dbca kept failing on identifying the directory for the control files as a 
raw device inspite of passing
the -datafilesDirectory option].

Then, on startup of the second database instance, I get errors like
"SSKGXP_IP: Primary interface down
operation cannot open cmcfg.ora
location clusterip1
error No such file or directory
Wed Dec 15 16:21:58 2004
cluster interconnect IPC version:Oracle UDP/IP
IPC Vendor 1 proto 2 Version 1.0
SSKGXP_IP: Primary interface down
operation
location
error Block device required
Wed Dec 15 16:21:58 2004
SSKGXP_IP: Primary interface down
operation cannot open cmcfg.ora
location clusterip1
error No such file or directory"

Obviously, a "cmfcfg.ora" files does not exist the second ORACLE_HOME.  A 
soft-link to the "oracm"
directory in the first ORACLE_HOME allows this database instance to detect 
and start using the Cluster Manager.

A TAR with Oracle Support hasn't helped much.  Support first asked me to 
run "lsnodes" when it didn't
exist in the second ORACLE_HOME.  The next analyst pointed me to Note 
235932.1  which seems
to be about using the right installer to install a patchset [9.2.0.3] into 
the RAC ORACLE_HOME.



Hemant K Chitale
http://web.singnet.com.sg/~hkchital



               
--
//www.freelists.org/webpage/oracle-l
Multiple ORACLE_HOMEs for 9iRAC on Linux

Other related posts:
