Hello, Yong Hao Gao.
I never said that JBM will forcibly close a active connection, how can you deduce this opinion from this post? On the contrary, the problem is either JBM or JBoss Remoting doesn't close (clean up) the failure connections that were caused by pulling out network cable.
Please ignore the timeout setting, the timeout setting was just our original suspicion. Previously we thought changing the timeout setting might solve this problem, but it didn't.
I think I had already described this problem clearly via JBOSS JIRA website and this forum. Please refer to https://jira.jboss.org/jira/browse/JBREM-1169 to know about the discusson between Ron Sigal and me. Also refer to https://jira.jboss.org/jira/browse/JBMESSAGING-1763 (you should already seen this JIRA issue). I had explained how to reproduce this problem and our environment configuration, I think it's sufficient for your guys to investigate this problem. What additional info are you required? Cannot you understand my description?
I thought this issue should be caused by either JBoss Remoting or JBoss Messaging, so I post the same question onto 2 forums and open 2 JIRA issues, I also spent some effort to reply and elaborate on this forum and JIRA website, but I regretted that nobody can help solve it for such a long time, eventually we have to use degrade JBM to bypass this problem as stated in this post. BTW, considering the instability of JBM, our customer has no confidence on JBoss, they said they might migrate onto other app server