Red Hat Bugzilla – Bug 1021750
Redeploying BRMS components causes thread leak
Last modified: 2013-11-04 09:37:26 EST
Description of problem:
When you redeploy BRMS components (jbpm-human-task, business-central, business-central-server, jboss-brms), some threads are not cleaned up and keep live.
Steps to Reproduce:
(This is an example of business-central-server. You may test other components with similar steps)
1. Start brms-standalone
2. Login to http://localhost:8080/business-central/
3. Capture a thread dump
4. Redeploy business-central-server.war (e.g. touch web.xml)
5. Login to http://localhost:8080/business-central/
6. Capture a thread dump
7. Compare 2 thread dumps taken in step 3 and 6
The number of threads increases. (You may ignore http threads)
The number of threads is not increased. (You may ignore http threads)
If you use BRMS components in Tomcat 7.0.42, Tomcat will log more clear messages for this issue so maybe useful to verify.
SEVERE: The web application [/business-central-server] appears to have started a thread named [Thread-0 (HornetQ-client-global-scheduled-threads-1981488825)] but has failed to stop it. This is very likely to create a memory leak.