Hi there Nope, you were looking at the wrong check-in. That doesn't contain the fix. Have a look at this commit log. Basically, I have delegated the restart process to a separate class, which ensures that concurrent restarts do not happen, nor do "back-to-back" restarts. It involves using a CAS to detect concurrent restarts, and then queueing waiters so that they wait for the concurrent restart before returning and being able to use a successfully restarted resource. Re: your changes to start(), I have not included these yet. These do make sense though, so I will make those changes accordingly. Cheers Manik View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4201821#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...