AS Java instance complications on Oracle Dataguard

[Former Member]

Having read many threads on Oracle DR for SAP, still some points remain unclear. Would appreciate if some expert can clear the points below for Java specific Instances:

DR is setup & switchover / failover also takes place fine but Java instance startup fails.

1. Some forums suggested to also copy the usr/sap/SID directory to the standby.

But will it not raise more errors coz the primary server hostname will conflict with the hostname of standby server ?

Directory backup, I suppose works if it is restored on the same host. But what about the different host with different hostname ?

2. I also found that bootstrap gives error "Cluster id [IDXXXXX]" not found in database. (Its the same if we restore the Java back up on a different host with different Instance ID)

This error is also logical & I found that replacing the ID of standby instance by the ID of primary server Java instance in the instance profile parameter "j2ee/instance_id" helped me bring the sytem up & it worked fine. But is it the correct method ? Coz with replaced Instance ID, the system may give errors in future upgrades or deployments, if performed on standby servers due to elimination of primary site.

Instance ID depends on hostname during installation (I read so in some S-Notes; Box number of Java instance). If we replace the ID manually then later sapinst or upgrade operations will detect a conflict in existing & expected Java ID.

So again, whats the proper procedure for Java instances ? Any reference to S-Notes or Guides ? I havn't found yet.

3. Once the DR gets running, & some JSPM / SDM deployment is performed on Primary node, how will the DR system be synch-ed up with primary. DB changes would be updated but I think SDM puts some files on file level also. If we perform a switchover after SDM deployment, will the server come up properly with all deployments, coz some files would be missing on file level.

Whats the technique to resolve this difference ?

Best regards / Thanks for time.

Edited by: Muhammad Ali Ahmed Siddiqui on Jan 29, 2012 10:55 AM

[Former Member]

Hi Muhammad,

You should copy "/usr/sap/<SID>/J<Instance#>/j2ee" excluding the folders, below;

"/usr/sap/<SID>/J<Instance#>/j2ee/cluster/server<#>/temp"

"/usr/sap/<SID>/J<Instance#>/j2ee/cluster/log"

Then, start the config tool, search for old hostname the replace with the new one.

>> Instance ID depends on hostname during installation (I read so in some S-Notes; Box number of Java instance). If we replace the ID manually then later sapinst or upgrade operations will detect a conflict in existing & expected Java ID.

It is correct... You will not face a trouble during the java archive package deployment neither on SDM nor JSPM. This number is checked by the system during the SAP upgrade. In short, if you are planning to upgrade on the new system after the role transition, you will face with the problem, because of the box number. This is because this number is calculated by hostname, instance# and SID

As a last suggestion is; you can replicate the files between hosts, noted on my first paragraph, by using robocopy (Windows) or rsync (UX)

Best regards,

Orkun Gedik