FUSE Message Broker
  1. FUSE Message Broker
  2. MB-1039

temp queues looping between networked brokers

    Details

    • Type: Bug Bug
    • Status: Resolved
    • Priority: Major Major
    • Resolution: Done
    • Affects Version/s: 5.5.1-fuse-00-06, 5.5.1-fuse-01-11
    • Fix Version/s: 5.5.1-fuse-02-02
    • Component/s: broker
    • Labels:
      None
    • Similar Issues:
      Show 9 results 

      Description

      In a network of brokers information about temp destinations are broadcast inside the network.
      The attached JUnit testcase reveals a bug where the use of temp destinations may result in those temp destinations being constantly created/updated on all brokers rather than being destroyed. It causes the brokers to spin.

        Gliffy Diagrams

        1. testcase.tgz
          6 kB
          Torsten Mielke
        2. testcase-trimmed.tgz
          69 kB
          Torsten Mielke

          Activity

          Hide
          Torsten Mielke added a comment -

          Use mvn test to run the testcase. It may take a few test runs before reproducing the problem. In the two environments that I tested, almost every test run did reproduce it.

          When reproduced you will see loads of logging output of this kind:

          DEBUG AbstractRegion                 - Removing destination: temp-queue://ID:Mac.local-52310-1323267147531-11:161:1
          DEBUG Queue                          - ID:Mac.local-52310-1323267147531-11:183:1 toPageIn: 0, Inflight: 0, 
          pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0
          DEBUG Queue                          - ID:Mac.local-52310-1323267147531-11:171:1 toPageIn: 0, Inflight: 0, 
          pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0
          DEBUG AbstractRegion                 - BrokerA adding destination: temp-queue://ID:Mac.local-
          52310-1323267147531-11:183:1
          DEBUG AbstractRegion                 - BrokerB adding destination: temp-queue://ID:Mac.local-
          52310-1323267147531-11:171:1
          DEBUG AbstractRegion                 - assigning ownership of auto created temp : temp-queue://ID:Mac.local-
          52310-1323267147531-11:183:1 to connection:ID:Mac.local-52310-1323267147531-9:2
          DEBUG AbstractRegion                 - assigning ownership of auto created temp : temp-queue://ID:Mac.local-
          52310-1323267147531-11:171:1 to connection:ID:Mac.local-52310-1323267147531-8:2
          DEBUG AbstractRegion                 - Removing destination: temp-queue://ID:Mac.local-52310-1323267147531-11:182:1
          DEBUG Queue                          - ID:Mac.local-52310-1323267147531-11:183:1 toPageIn: 0, Inflight: 0, 
          pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0
          DEBUG AbstractRegion                 - BrokerC adding destination: temp-queue://ID:Mac.local-
          52310-1323267147531-11:183:1
          DEBUG AbstractRegion                 - BrokerB adding destination: temp-queue://ID:Mac.local-
          52310-1323267147531-11:193:1
          DEBUG AbstractRegion                 - assigning ownership of auto created temp : temp-queue://ID:Mac.local-
          52310-1323267147531-11:183:1 to connection:ID:Mac.local-52310-1323267147531-6:2
          DEBUG Queue                          - ID:Mac.local-52310-1323267147531-11:161:1 toPageIn: 0, Inflight: 0, 
          pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0

          Show
          Torsten Mielke added a comment - Use mvn test to run the testcase. It may take a few test runs before reproducing the problem. In the two environments that I tested, almost every test run did reproduce it. When reproduced you will see loads of logging output of this kind: DEBUG AbstractRegion - Removing destination: temp-queue://ID:Mac.local-52310-1323267147531-11:161:1 DEBUG Queue - ID:Mac.local-52310-1323267147531-11:183:1 toPageIn: 0, Inflight: 0, pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0 DEBUG Queue - ID:Mac.local-52310-1323267147531-11:171:1 toPageIn: 0, Inflight: 0, pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0 DEBUG AbstractRegion - BrokerA adding destination: temp-queue://ID:Mac.local- 52310-1323267147531-11:183:1 DEBUG AbstractRegion - BrokerB adding destination: temp-queue://ID:Mac.local- 52310-1323267147531-11:171:1 DEBUG AbstractRegion - assigning ownership of auto created temp : temp-queue://ID:Mac.local- 52310-1323267147531-11:183:1 to connection:ID:Mac.local-52310-1323267147531-9:2 DEBUG AbstractRegion - assigning ownership of auto created temp : temp-queue://ID:Mac.local- 52310-1323267147531-11:171:1 to connection:ID:Mac.local-52310-1323267147531-8:2 DEBUG AbstractRegion - Removing destination: temp-queue://ID:Mac.local-52310-1323267147531-11:182:1 DEBUG Queue - ID:Mac.local-52310-1323267147531-11:183:1 toPageIn: 0, Inflight: 0, pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0 DEBUG AbstractRegion - BrokerC adding destination: temp-queue://ID:Mac.local- 52310-1323267147531-11:183:1 DEBUG AbstractRegion - BrokerB adding destination: temp-queue://ID:Mac.local- 52310-1323267147531-11:193:1 DEBUG AbstractRegion - assigning ownership of auto created temp : temp-queue://ID:Mac.local- 52310-1323267147531-11:183:1 to connection:ID:Mac.local-52310-1323267147531-6:2 DEBUG Queue - ID:Mac.local-52310-1323267147531-11:161:1 toPageIn: 0, Inflight: 0, pagedInMessages.size 0, enqueueCount: 0, dequeueCount: 0
          Hide
          Torsten Mielke added a comment -

          Attaching a trimmed down testcase (testcase-trimmed.tgz).
          It only uses two brokers in a network and plain JMS APIs (no more stomp).
          Using this testcase the problem reproduces fairly every time now.

          Show
          Torsten Mielke added a comment - Attaching a trimmed down testcase (testcase-trimmed.tgz). It only uses two brokers in a network and plain JMS APIs (no more stomp). Using this testcase the problem reproduces fairly every time now.
          Hide
          Gary Tully added a comment -

          reopened https://issues.apache.org/jira/browse/AMQ-3253 to address this problem

          Show
          Gary Tully added a comment - reopened https://issues.apache.org/jira/browse/AMQ-3253 to address this problem
          Hide
          Gary Tully added a comment -

          fix on the 5.5.* fuse branches

          Show
          Gary Tully added a comment - fix on the 5.5.* fuse branches

            People

            • Assignee:
              Gary Tully
              Reporter:
              Torsten Mielke
            • Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: