Hi

I am having trouble installing 64-bit Oracle RDBMS version 9204 on NetApp
storage.



This is a 2 node configuration, sharing NetApp filer storage.

In each node SLES 9 SP2 and necessary packages have been installed. The
storage is NFS mounted and can be seen by each of the nodes.

The oracle software owner and group is also verified and validated.
Necessary environment variables were also set.

The quorum disk file is also created and all the required permissions were
granted.

Hangcheck-timer module is installed and verified.



When 9204 Cluster Manager is installed from node1, as part of the
installation necessary files were copied to the second node.

Cluster manager configuration file is also modified to include
KernelModuleName=hangcheck-timer.



After the above installation and configuration were completed, when
./ocmstart.sh (to invoke Oracle Cluster Manager) is executed, Oracle Cluster
Manager processes (oracm) get started in one node. The issue is, when I try
to start the Cluster Manager processes in the second node it would not,
coming out with errors. (The errors run into pages, for want of space, I did
not include them. If anyone wants, I would be more than happy to provide the
error logs.)



When Oracle is contacted, I was told to apply patch set 9206 Cluster Manager
to fix this issue. After applying 9206 patch set for Cluster Manager, the
Cluster Manager processes get started in each of the nodes successfully. As
a next step, when I try to install Oracle 9204 RDBMS in node1, the
installation fails with an error "There was an error loading some library
files". Oracle was contacted again and using their web conference tools,
the problems were demonstrated to their analysts.

It was later filed as bug to their internal development team.



I tried the workaround by trying to install 9204 RDBMS on 9204 Cluster
Manager with only one node running the Cluster Manager processes. (this is
done with the intention of upgrading to 9206 Cluster Manager and RDBMS at
the same time) However, the installation failed saying it requires "gsdctl"
processes be running. When tried to start the "gsdctl" process, it fails
since it requires both the nodes running the cluster manager processes.



While Oracle is working on this issue, I would appreciate any other work
around from anybody that came across similar situation.



Thanks

Shree.