Something that is bothering us is - during the Load testing when we increase Load on the application, JBoss increases the number of threads. This done based on the Jboss HTTP ThreadPool Configuration (Max upto 200 threads for HTTP requests). But when the application load comes back to a normal state we notice that all the threads that where created by JBoss still remain in 'Waiting State' ever after. This happens in both the environments (Windows as well as CentOS).
So My Question is :
Is it wrong to expect that the number of JVM Threads should come down ? (Ideally in my opinion JBoss should kill some threads to free up the memory.)
If No then how do we configure the same.
Reasons Why we are bothered about this issue is that we are using ThreadLocal for passing some objects around different containers.
After some searching the only relevant post that I could fond is following open issue in Jboss Bug Tracking System: https://issues.jboss.org/browse/JBAS-8595?page=com.atlassian.streams.streams-jira-plugin%3Aactivity-stream-issue-tab
Based on this, should I assume that This issue is the known bug in JBoss and we have to live with this util we get a fix for this bug.
Thanks and Regards