cancel
Showing results for 
Search instead for 
Did you mean: 

SUM EHP7 upgrade with RAC HA stopped in MAIN_UPTRANS/STARTSAP_PUPG

former_member188433
Participant
0 Kudos

Hi - We are upgrading our ERP system from NW 7.02/EHP5 to NW 7.40/EHP7. The downtime of the upgrade is finished but the upgrade is stopped with an error in MAIN_UPTRANS/STARTSAP_PUPG:

The instance is running.  All processes/services are up and I can log in to the upgraded system with no issues.

The message server/enqueue processes (along with enqueue replication) runs on an Oracle RAC HA (Clusterware) cluster with three server nodes.

The upgrade is executing sapcontrol commands. The commands to the SAP app servers run fine.  But when the sapcontrol commands are executed on the RAC server nodes there are issues. Sometimes the sapcontrol command runs on a server node where no SAP processes are running.  Sometimes an instance number is specified that is not running on the server node.

See attached SAPup.ECO log for details.

Can anyone provide guidance or a possible workaround so that I can complete the upgrade?  All required services are running.
Best Regards - Jeff

Accepted Solutions (1)

Accepted Solutions (1)

former_member188433
Participant
0 Kudos

Hi All - My issue had to do with changes in the RAC HA services.  The upgrade process "remembers" what the instance processes looked like (and where they were running) before the upgrade started and tries to ensure that all instances in the list start up ok after the upgrade.  In my case there had been changes on the RAC so the upgrade got confused.

I will mark this question as answered.

Thanks - Jeff

michael_sharrar
Participant
0 Kudos

Jeff, I am having the same exact issue when applying EHP6 with a RAC environment.  We are running a 2 node RAC cluster and the sapcontrol GetProcessList is failing on the second Oracle node.  Our system is up and healthy (I'm able to login from the SAPgui).

When I run the command manually from the OS level and substitute the first Oracle node the command does not throw the error.

Can you please let me know what you did to "un-confuse" the SUM tool?  I think I will be able to proceed if I can get SUM to check the first Oracle node rather than the second.

Regards,

Mike

former_member188433
Participant
0 Kudos

Hi Mike -

At the beginning of the upgrade the SUM tool stores the state of all the instances in a file in the SUM directory:

../SUM/abap/mem/INSTANCELIST.DMP

After the upgrade SUM wants to see that everything listed in this file us back up and running exactly as it was before the upgrade started (using INSTANCELIST.DMP as its reference).


Edit the file (make a copy first)

  • Remove all entries except the ones having to do with instances on SAP app servers.
  • Stop the upgrade (to clear any possible cached values in memory)
  • Restart the upgrade and restart the step.

This should allow the upgrade to continue.

Hope this works for you - Jeff

michael_sharrar
Participant
0 Kudos

Thanks for the reply Jeff.  I realized that I could move our ASCS instance to the second Oracle node.  Once the ASCS was in the "proper" location I was able to complete the SUM process successfully.

However, this is very useful information to have so thanks again.

Best regards,

Mike

Answers (1)

Answers (1)

Sriram2009
Active Contributor
0 Kudos