To limit the scope of the discussion: The problem we're talking about has nothing to do with durable subscriptions. DurableCrashTest1 and DurableCrashTest2, while semantically correct, were designed to investigate a completely different bug, and Clebert stumbled over the current one (http://jira.jboss.org/jira/browse/JBREM-596), just by chance. Let's create a Messaging test that exhibits the problem using the minimal possible number of moving parts. Clebert is working on that. I would suggest "TwoConnectionsCrashTest" as name of the test. Let's post the conclusions on this thread, and not in private e-mails. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3970843#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...