What happened on my test was a call to close taking longer than PING_TIME. The test was sending 100 messages, in 10 threads. All the sends were being performed asynchronously, and every PersistentMessage sent was syncing the page (it would be same effect on Journal if using NIO). When close was called, it had to wait the pending commands to finish, what took longer than the PING_TIME. But you shouldn't have a DeadLock on that case.. right? View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4194180#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...