Thursday, November 13, 2014

Load balanced 2nd node for Hyperion BI+ doesn't work independently

While building 11.1.2.3 from the  ground up, I have run into a string of issues. I will continue to post on all the issues but these tips can help someone who may be in a similar situation and some things to think about while troubleshooting them.

I have every web application load balanced on windows 2008 r2 64 bit . I began testing as soon as I completed configuring everything and started to test indepently to see if node 2 continues to run when node1 is completely down because in an actual DR situation these things matter.Thats when I ran into this message for the first time.

"The Startup Document does not exist in the repository" followed by "GSM_unabletoacquire LSM"

On a network deployment there are several reasons that can cause this:

1. The service account for the HyS9RaFrameworkAgent_epmsystem is not the same as the installer but the "local system" account. This is an absolute must since RM1 is a shared folder across 2 servers.

2. The Service Broker service and the GSM on the second server is not enabled.
To Resolve this,
- Log on to EPM Workspace on Server1
- Select Navigate, then Reporting and Analysis, and then Services.
- Right-click Reporting and Analysis for server2 , and then select Properties.
- On the Services tab, set GSM to Enabled (Yes) and Service Broker to Enable (Yes).

3. Verify that HAVA ID in EPM registry matched in css_provisioning_info table
a: Execute the following query against the Shared Services registry database:
b: select application_id from css_provisioning_info where lower(application_id) like 'hava%';
 note this application id, e.g. HAVA:0000000b0000bf00-0000-0000-0acc1bac
c: Run epmsys_registry.bat from \Oracle\Middleware\user_projects\epmsystem1\bin
Open EPM registry.xml and search for HAVA and compare the HAVA ID using the ID above

d: If there is a mismatch, edit the Shared Services Registry:
Go to EPM_ORACLE_INSTANCE/bin and run the following command:
epmsys_registry updateproperty RA_FRAMEWORK/@applicationId

Where applicationId is the id returned from the query above. For example, using the id above:

epmsys_registry updateproperty RA_FRAMEWORK/@applicationId HAVA: 0000000b0000bf00-0000-0000-0acc1bac

e. Restart all services on Server1 and ensure Shared Services and Workspace are accessible

However, in my case, all of these seemed perfect. I stumbled upon my resolution by chance . The Location for the RM1 shared folder (in the configurator, Eg: \\server1\c$\Oracle\Middleware\user_projects\epmsystem1\ReportingAnalysis\data\RM1 stopped being recognized from server 2 . 
The way in which server 1 is resolved from server 2 turned out to be the root cause of my issue.It used to be accessible by the short name earlier and suddenly started to accept only the FQDN owing to some DNS related changes.  In my opinion, We should go with the same servername that we issue in the host file as part of the preinstall. To override that, I added an entry on the server 2 host file and it started working.




No comments:

Post a Comment