Oh, I forgot: was there any indication of Producer Flow Control kicking in on the broker prior to when the consumer stopped processing messages?
On Sun, Apr 26, 2015 at 9:06 AM, Tim Bain <tb...@alumni.duke.edu> wrote: > Did you ever solve this issue? > > If not, can you please clarify a couple things in your description? > > 1. Did you confirm (via a thread dump) that the MDB wasn't actually > hung somewhere in your consumer code? Clearly the broker believed it had > dispatched a prefetch buffer worth of messages to the consumer, so the > question is whether the consumer's ActiveMQ client/RAR code had attempted > to dispatch the next message to your MDB or not. And if the thread wasn't > stuck in your MDB code, where was it stuck instead? > 2. When you shut down Glassfish, was your MDB consumer connected or > disconnected? The first half of your description sounded like it was > connected, while the last paragraph sounded like it was disconnected. > 3. When you connected the additional consumer, did it process messages > as expected? You never said what happened when those additional consumers > connected. > 4. Did that ID match one of the consumers (or one of their > connections; I can never tell consumer IDs and connection IDs apart)? If > so, which one? > 5. If you can reproduce this reliably, does the exception happen if > you shut down Glassfish without attempting to connect any additional > consumers? > > Tim > > On Tue, Mar 24, 2015 at 3:24 PM, sbarlabanov <cont...@sergiy-barlabanov.de > > wrote: > >> Hi, >> >> this is the continuation of my forum entry >> >> http://activemq.2283324.n4.nabble.com/ActiveMQ-5-10-0-queue-slowed-down-restart-helped-td4690706.html#a4690723 >> . >> But since the discussion there went to another direction I start a new >> topic. >> We got the issue again several times in the last two month. >> More details: we have a standalone running ActiveMQ 5.10.0, Glassfish >> 3.1.2.2 with ActiveMQ RAR and an MDB consuming messages from an ActiveMQ >> queue. >> Suddenly after weeks successful processing the MDB stops processing any >> message (I mentioned that already). In jconsole we see the corresponding >> JMS >> consumer on the queue. MessageCountAwaitingAcknowledge of the consumer is >> equal the prefetch buffer (10 in our case). From time to time a new >> consumer >> occurs on the queue (this is seen in jconsole), but the first one stays >> with >> MessageCountAwaitingAcknowledge=prefetch buffer. >> When we stop Glassfish, the following interesting exception appears in >> activemq.log: >> >> java.lang.IllegalStateException: Cannot remove session from connection >> that >> had not been registered: ID:NOEUD-SI2-55792-1427222776632-47:1 >> at >> >> org.apache.activemq.broker.TransportConnection.processRemoveSession(TransportConnection.java:677)[activemq-broker-5.10.0-patch4.jar:5.10.0-patch4] >> at >> >> org.apache.activemq.command.RemoveInfo.visit(RemoveInfo.java:74)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.broker.TransportConnection.service(TransportConnection.java:294)[activemq-broker-5.10.0-patch4.jar:5.10.0-patch4] >> at >> >> org.apache.activemq.broker.TransportConnection$1.onCommand(TransportConnection.java:148)[activemq-broker-5.10.0-patch4.jar:5.10.0-patch4] >> at >> >> org.apache.activemq.transport.MutexTransport.onCommand(MutexTransport.java:50)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.transport.WireFormatNegotiator.onCommand(WireFormatNegotiator.java:113)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.transport.AbstractInactivityMonitor.onCommand(AbstractInactivityMonitor.java:270)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.transport.TransportSupport.doConsume(TransportSupport.java:83)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.transport.nio.NIOTransport.serviceRead(NIOTransport.java:138)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.transport.nio.NIOTransport$1.onSelect(NIOTransport.java:69)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.transport.nio.SelectorSelection.onSelect(SelectorSelection.java:94)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> org.apache.activemq.transport.nio.SelectorWorker$1.run(SelectorWorker.java:119)[activemq-client-5.10.0-patch2.jar:5.10.0-patch2] >> at >> >> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)[:1.7.0_45] >> at >> >> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)[:1.7.0_45] >> at java.lang.Thread.run(Thread.java:744)[:1.7.0_45] >> >> It seems that something goes wrong in ActiveMQ RAR. Seems that something >> hangs around in the RAR locking the messages. Because otherwise ActiveMQ >> will not complain about the session which is going to be removed from an >> unregistered session. So something went wrong before. And the broker did >> not >> get it and still thinks that the consumer/session/connection is open. >> >> >> >> >> -- >> View this message in context: >> http://activemq.2283324.n4.nabble.com/MDB-stops-consuming-messages-tp4693769.html >> Sent from the ActiveMQ - User mailing list archive at Nabble.com. >> > >
