Uploaded image for project: 'JBoss Enterprise Application Platform'
  1. JBoss Enterprise Application Platform
  2. JBEAP-12566

Unresponsive broker caused by synchronization issue in QueueImpl

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done
    • Icon: Blocker Blocker
    • 7.1.0.CR1
    • 7.1.0.ER3
    • ActiveMQ
    • None

      The issue was revealed by SpecJMS 2007. There is synchronization issue in QueueImpl which causes that the system temporary hangs and then clients timeout.

      Customer impact: When the Artemis is under the load, it may get into the state when the system temporary hangs and clients timeout.

      We didn't see the issue with 7.1.0.ER2. It is regression against 7.0 and 7.1.0.ER2.

      In the SpecJMS 2007, it sometimes happens that all threads are waiting for directDeliveryGuard lock, whereas the thread holding this lock is waiting (10 sec) in flushExecutor until all threads are finished, but they cannot be finished because of the lock.

      In the log it can be seen a lot of following warnings.

      WARN  [org.apache.activemq.artemis.core.server] (Thread-24 (ActiveMQ-server-org.apache.activemq.artemis.core.server.impl.ActiveMQServerImpl$3@15159858)) AMQ222172: Queue 95f2b86f-2bfc-45fb-b1d6-fe2d3c66d6b7 was busy for more than 10,000 milliseconds. There are possibly consumers hanging on a network operation
      

      See thread-dump.txt for more details.

              csuconic@redhat.com Clebert Suconic
              eduda_jira Erich Duda (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

                Created:
                Updated:
                Resolved: