Skip to Content
avatar image
Former Member

Expanding MSCS to include Java stack (ABAP system becomes ABAP+JAVA)

We had a BW 3.x system (ABAP only) clustered and working fine. We have upgraded that to NW 70 BI and added the Java to the stack. So there is still one SID/system. The addition is the java processes.

I'm a little confused in the docs about configuring the cluster for this type of scenario - or if any additional cluster work must be done if the instance itself was already clustered.

Right now, I have no problems starting up the instance on node A and getting all the services running fine. I'm a little unsure how to proceed testing a failover to node B and what, if any, resources I should add to include the new Java stack in the pre-existing instance.

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

1 Answer

  • avatar image
    Former Member
    Jul 10, 2008 at 04:24 PM

    Was this a standard SAP install via SAPINST? Or did you manually create these instances? If SAPINST, then I think that everything should be taken care of. If manually, you may want to verify:

    1. JAVA_HOME variable on both nodes. Make sure node B is in sync with node A (active).

    2. The JRE/JDK version should be the same.

    3. In the MSCS resource properties, there's a timeout or wait value, by default I think is like 30 or 300 seconds (it's 3-something). You may want to increase this to around 10 minutes, since AS-JAVA takes a bit more time to start up.

    4. Verify all the SAP* NT services are in sync between the two nodes.

    Other than that, I would just take a downtime and test the failover to see if it works.

    Add comment
    10|10000 characters needed characters exceeded

    • Former Member Former Member

      I mean if the variables and items mentioned in my last post was set identical between the two nodes. Meaning, the JAVA version should be the same, installed to the same place, and the JAVA_HOME system variable should be set to the same value.

      Also, the SAP<SID>_## entries under the NT Services on both nodes should be the same.