This would suggest an uncaught exception could be killing the executor thread.
If you have a local build, Can you wrap the conn.fail method with a try catch (throwable t) See if that helps at all. Sent from my iPhone > On 28 Jun 2017, at 16:19, francesco81 <francesco.padov...@bticino.it> wrote: > > An update. > We have re-build "artemis-commons-2.1.0.jar", > "artemis-mqtt-protocol-2.1.0.jar" and "artemis-server-2.1.0.jar" in order to > improve the logs. It seems that there is a problem with the thread which has > in charge the activities to free the clientid, send lwt message, etc.. > For a few time all works good. Broken connection are correctly cleaned and I > see all the logs related to the clientid cleaning and lwt sending. > Then, at a certain point (we've not yet undestood why...) connections are no > more cleaned correctly. And there are no more logs about the "cleaner" > thread. And there's no error. It seems like the connection.fail method > simply is not called. > > logger.debug("connection id " + conn.getID() + " calling fail executor > thread "); flushExecutor.execute(new Runnable() { > > public void run() { logger.debug("starting thread > for removing connection "+ pair.getA()); > conn.fail(ActiveMQMessageBundle.BUNDLE.clientExited(conn.getRemoteAddress(), > pair.getB())); logger.debug("thread for connection > "+ pair.getA() + " ended correctly"); } > > ...I can't see the log above. Have you any idea about the cause? > > Thanks. > > Francesco > > > > -- > View this message in context: > http://activemq.2283324.n4.nabble.com/Artemis-rejects-the-clientid-after-client-disconnection-tp4727426p4728012.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com.