Hello, We have connection pool set at 50 currently. The app team wants to increase it to 100 as they exhausted last week. I want to know what needs to be looked at on the DB server end to make sure that the new change can be supported. Is looking at javaheap and maxconnections good enough? We are at 10.1 db2 on aix 7.1. Thanks
Sometimes the app-server connection-pool gets exhausted because db-tasks are either not completing, or are completing too slowly. If that is happening then doubling the pool size will not cure the problem but may delay the symptoms or expose different symptoms. It is wiser to get to the root cause before making changes. If root cause for pool exhaustion is already known and proved to be NOT a db2 issue then you are OK.
Apart from that, without knowing all your settings for the DB-CFG and DBM-CFG and db2-registry settings it is difficult to give specific advice. Also, would need to know if your application has deployed java-stored-procedures or java-functions/routines-inside-db2. Also, if you app-server is WAS then what is the db2-client version+fixpack deployed on the application-server(s), and what kind of regime for statistics and reorgs are regularly fully completed on your databases.
Thanks for the response. There are no stored procedures. The application was not slow and the app team came back and said that the connection pool is getting exhausted as they started noticing connection time out errors. Regarding the registry variable, it's the usual ones like DB2COMM ...When you say db, dbm cfg are you asking me for any specific ones? STMM is enabled and max connection setting is automatic. This is WAS 6.1. Reorgs/runstats are done on a regular basis.
How will doubling the size of the connection-pool cure timeout errors?
Sounds like root cause for these symptoms is not yet proved (i.e. why are the connections timing out, what exactly is happening in terms of specific error codes, what methods tend to be running on those connections when the symptom happens etc etc ).
The app team says they didn't see slowness in the app or any queries slowing down. From the TCP timeout, they are assuming its to do with the pool. One the DB server, the resource usage is normal. Do you think the app team's explanation is not convincing? They say there is no explicit message about the pool exhaustion.