On Mon, Nov 8, 2010 at 10:42 PM, David Blevins <[email protected]> wrote: > On Thu, Nov 4, 2010 at 12:27 PM, Matthias Wessendorf <[email protected]> > wrote: >> I am seeing this, when shutting down "tomtom" >> >> SEVERE: The web application [/ejb-examples] appears to have started a >> thread named [poolevict...@17704464] but has failed to stop it. This >> is very likely to create a memory leak. > > Looks like we may need to hook in the undeploy logic on context > lifecycle stop even. Not sure if that would cause the rest of the > issues, but it would be the source of the pool eviction timer not > being cancelled. > > Will take a look tomorrow. Hopefully can stamp it out and reroll.
Ok, looks like the log message relates to some checks that Tomcat added just after the STOP event and just before the AFTER_STOP event. We were doing our undeploy in AFTER_STOP. So bottom line is the log messages were reporting things that weren't actually issues. I moved our undeploy a bit sooner to happen in STOP and that does take care of most the log messages. In the itests it does report some ActiveMQ related threads as still running, but those seem to be for the resource adapter and broker and should run as long as tomcat runs. Will have to investigate that further, but it shouldn't be an issue. -David
