Hi,
We have an application which, it would appear, is not managing JDBC connections properly in some circumstances. The connection count in the pool gradually increases over the day, until eventually it runs out of available connections.
Code analysis has identified one unhandled exception which could trigger an unclosed connection.
In a test environment, we forced the above exception repeatedly in order to create the unclosed connections. In this test environment, we then used the "Inactive Connection Timeout" setting to determine if it would treat the symptoms by returning the idle connections to the pool, buying us time to fix and test the solution. It worked as expected, clearing up the idle connections, keeping the connection count down.
We tried the same "inactive connection timeout" setting in the production environment, but it hasn't stopped the leak.
This leads us to conclude that either something is different in the production weblogic environment, or the event which is causing the connection leakage is not the one we have identified.
Log analysis in the production environment shows that the "timeoutInactiveRes" entires appear as expected, but the connections aren't being reclaimed. Is there any other WebLogic setting which might be preventing the "inactive connection timeout" setting from clearing up the idle connections?
Or could there be something other than unclosed connections causing the connection count to creep up?
Thanks.