[ https://issues.jboss.org/browse/JBIDE-20426?page=com.atlassian.jira.plugi... ] Rob Stryker commented on JBIDE-20426: ------------------------------------- https://github.com/jbosstools/jbosstools-server/pull/466 We will still get two threads trying to stop the server, potentially at different times. We had a check present to verify the 'shared process' (ie the process for the currently launched server) was identical to the one we're responding to terminating. However, this check was too early. It was possible for our check on the 'shared process' to return the old process for the server that was just terminated. Meanwhile, in other threads, our poller has marked the server as stopped, and a new process has been launched. Then, our 'process terminated' listener tries to stop it again, without re-checking that the process is the same yet again. I've traced through this extensively and tried a bunch of various paths in selectively progressing individual threads to see if it's possible to break, and my tests indicate the PR fixes the issue. What's more, if the 'process terminated' listener gets execution before the 2nd start of the server, it will prevent the 2nd start of the server until it finishes its checks. So that's safe. If on the other hand the process terminated listener gets execution after the 2nd start of the server, the check that the processes are identical will fail, so no redundant (or broken) setServerStopped() call will be made. This one feels good to fix. -- This message was sent by Atlassian JIRA (v7.2.3#72005)