Uploaded image for project: 'Infinispan'
  1. Infinispan
  2. ISPN-8646

PessimisticTxPartitionAndMergeDuringRollbackTest.testDegradedPartitionWithDiscard failures

    XMLWordPrintable

Details

    Description

      The discard filter isn't working properly. the discard is only expects (and discard) a single command but if the retries happens to fast, the discard let it go through.

      this may be the cause of the other discard related failures in this package

      log relevant parts

      #Node C discards the rollback (as expected)
      15:34:04,611 TRACE (jgroups-11,PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeC-27886) [org.infinispan.partitionhandling.BaseTxPartitionAndMergeTest] [Discard] Checking command RollbackCommand {gtx=GlobalTx:PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeA-2726:6, cacheName='pes-cache', topologyId=13}.
      15:34:04,611 TRACE (jgroups-11,PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeC-27886) [org.infinispan.partitionhandling.BaseTxPartitionAndMergeTest] [Discard] Discarding command RollbackCommand {gtx=GlobalTx:PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeA-2726:6, cacheName='pes-cache', topologyId=13}.
      
      #we split the cluster now
      15:34:04,612 DEBUG (testng-PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]) [org.infinispan.partitionhandling.PessimisticTxPartitionAndMergeDuringRollbackTest] Splitting cluster in equal partition
      
      #topology changed on originator and it retries the rollback
      15:34:04,623 TRACE (remote-thread-PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeA-p34-t6) [org.infinispan.remoting.transport.impl.MultiTargetRequest] Target PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeC-27886 of request 25 left the cluster view
      15:34:04,623 TRACE (remote-thread-PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeA-p34-t6) [org.infinispan.statetransfer.StateTransferInterceptor] Retrying command RollbackCommand {gtx=GlobalTx:PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeA-2726:6, cacheName='pes-cache', topologyId=14} for topology 14
      
      #command isn't discarded and it is enqueued. NodeC is still in topology 13.
      #this is wrong. the test expects the rollback to be lost.
      15:34:04,629 TRACE (jgroups-11,PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeC-27886) [org.infinispan.partitionhandling.BaseTxPartitionAndMergeTest] [Discard] Checking command RollbackCommand {gtx=GlobalTx:PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeA-2726:6, cacheName='pes-cache', topologyId=14}.
      15:34:04,629 TRACE (jgroups-11,PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeC-27886) [org.infinispan.statetransfer.StateTransferLockImpl] Checking if transaction data was received for topology 14, current topology is 13
      15:34:04,629 TRACE (jgroups-11,PessimisticTxPartitionAndMergeDuringRollbackTest[DIST_SYNC]-NodeC-27886) [org.infinispan.util.concurrent.BlockingTaskAwareExecutorServiceImpl]
       Added a new task to the queue: 1 task(s) are waiting
      

      Attachments

        Issue Links

          Activity

            People

              pruivo@redhat.com Pedro Ruivo
              pruivo@redhat.com Pedro Ruivo
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: