Veritas Cluster Steps before upgrading Oracle (freeze oracle agent) in VCS environment

** Credit Web
If you have 2 nodes of Oracle database, one production server and one standby server.
Prerequisites :
- Make sure that the existing data on shared disk is compatible with old version and new version of Oracle
- Make sure that the VCS Oracle Agent installed on both node is compatible with new version of Oracle that we are going to upgrade to

1. Persistent Freeze to Oracle Service Group with VCS’ Cluster Management Console. This will prevent automatic failover capability while upgrading.

Figure 1. Right-click at Oracle Service Group, and select Freeze/Persistent

Figure 2. Red Lock Icon will show to indicate that the service group is now Frozen

2. Start upgrading Oracle’s binary resides on local volume on standby node (Node B), and make sure that upgrading Oracle’s binary doesn’t change existing settings, such as location of PFILE and SPFILE after finish.
   

3. After you finish the upgrade on Node B, Oracle is still running on Node A. Check the attributes of VCS’s resources related to Oracle that they are correct.

1. (While Oracle’s configuration hasn’t been changed, so the Oracle related resources’ attributes should be usable and no need to reconfigured)

4. After you finish checking the configuration and make sure that they are the same, Unfreeze the Oracle Service Group and then try to failover Oracle Service Group on Node A to Node B. Oracle will be running on Node B with new binary after failover.

1. If errors happen during failover, check VCS log file to see the resource that is problem, and correct the attribute for that resource, and then try to failover again.

5. At this time, Oracle is running on Node A. Before starting upgrade Oracle’s binary on Node A, freeze the Oracle Service Group.

6. Start upgrading Oracle’s binary on Node A. When the process finish, check the configuration of Oracle and Oracle related resources if they are matched.

7. At this point, Oracle should be all upgraded to newer version on both nodes in VCS. You need to try failover between two nodes to make sure that VCS can control all of resource in Oracle Service Group.