Uploaded image for project: 'Infinispan'
  1. Infinispan
  2. ISPN-819

Deadlock detection seem to not detect deadlock case between two nodes on the same cache.

    XMLWordPrintable

Details

    • Hide

      1/ Launch two times the program TwoNodesDeadlocks.java.
      2/ The program will wait that two nodes to be launched.
      3/ Each node will print that deadlock detection is activated.
      4/ Each node will try to perform hundred times get and put operations on the same key in a transaction.
      5/ deadlock occurs.

      Show
      1/ Launch two times the program TwoNodesDeadlocks.java. 2/ The program will wait that two nodes to be launched. 3/ Each node will print that deadlock detection is activated. 4/ Each node will try to perform hundred times get and put operations on the same key in a transaction. 5/ deadlock occurs.

    Description

      Hello,
      after activating deadlock detection with 500 ms of spin duration, i launch a classic deadlock scenario.
      Node 1 acquire local lock.
      Node 1 try to acquire cluster wide lock.
      Node 2 acquire local lock.
      Node 2 try to acquire cluster wide lock.
      Then deadlock is not detected and the TimeoutException occurs:
      org.infinispan.util.concurrent.TimeoutException: Unable to acquire lock after [20 seconds] on key [counter_key] for requestor [DldGlobalTrans
      action

      {coinToss=3136454835479173852, isMarkedForRollback=false, lockIntention=counter_key, affectedKeys=[counter_key], locksAtOrigin=[]}

      Glob
      alTransaction:<nodeDeadLock-11465>:5:local]! Lock held by [DldGlobalTransaction

      {coinToss=3164596253027655679, isMarkedForRollback=false, lock Intention=null, affectedKeys=[], locksAtOrigin=[]}

      GlobalTransaction:<nodeDeadLock-16603>:1:remote].

      I give my configuration file for ISPN and a Class TwoNodesDeadlocks.java to reproduce the problem.

      Attachments

        1. AgentManager_InfiniSpan_Config_Generic.xml
          3 kB
          Thomas Jean
        2. log4j_Generic.properties
          0.3 kB
          Thomas Jean
        3. nodeDeadLock-43429.log
          11 kB
          Thomas Jean
        4. nodeDeadLock-45547.log
          7 kB
          Thomas Jean
        6. ThreeNodesDeadlocks.java
          3 kB
          Thomas Jean
        7. TwoNodesDeadlocks.java
          3 kB
          Thomas Jean

        Issue Links

          relates to

          Bug - A problem that impairs or prevents the functions of the product. ISPN-836 Three nodes leading in replication timeout exception.

          • Major - To be worked after higher priority work items are addressed. When the severity is high and the effort to change it is low to moderate, this term is chosen. Issues in this category likely have an existing workaround but implementation or execution may be non-trivial.
          • Resolved

          Activity

            People

              mircea.markus Mircea Markus (Inactive)
              thom28 Thomas Jean (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: