DB2 startup problem during HACMP takeover

[ggitlin]

Hello all,
I am running HACMP 5.1 cluster with two nodes in cascading failover, under AIX 5.1 ML3.
The HACMP failover without database (no application startup scripts defined in the resource group) works fine. Once I add the application start up script with DB2 start, I get the following message in /tmp/hacmp.out:
"
SQL6048N A communication error occurred during START or STOP DATABASE MANAGER processing.
"
The DB2 fails to start. As a concequence, when I try to reintegrate the first node (which had HACMP stopped with "takeover" option), I get resource group aquasition failure.

Appreciate any thoughts on this.
Thanks. Greg. AIX Administrator.

 

[sbix]

Try to stop all DB2 applications first.
In the stop script put something like:
db2 force application all
db2 stop database manager
db2stop

 

[ggitlin]

Thanks, sbix.

That's what I have in the stop scripts, provided to me by the DBA. But the real question is what causes the initial error with DB manager?

Do i need two DB managers running if i have two separate DB instances on the node?

 

[ggitlin]

Also, does anyone out there know of any guides/redbooks about running DB2 under HACMP?

 

[sbix]

Looks like you switch address before to stop applications, which should be avoided by the stop application server scripts.
Perhaps they don't work very well ... try to put some logging in them, to catch the DB2 stopping

 

[AIX5L]

Hi

Can you send your Database Start-Script ?
What you do (step-by-step) on takeover with your Resources ?
Do you set the Hostname in the Startsrcipt ?

 

[ggitlin]

Thanks for your responces. We had a consultant come in and he resolve the problem.