JR46582: DEADLOCK ON PROCESSCONTEXT TABLE WHEN TWO ASYNCHRONOUS RESPONSE MESSAGES ARRIVE SIMULTANEOUSLY

Download Version 8.0.1 Fix Pack 2 for the IBM Business Process Manager products
Version 7.5.1 Fix Pack 2 for the IBM Business Process Manager products
Version 8.5 Refresh Pack 5 for the IBM Business Process Manager products

APAR status

• Closed as program error.

Error description

• DEADLOCK ON PROCESSCONTEXT TABLE WHEN TWO ASYNCHRONOUS RESPONSE
  MESSAGES ARRIVE SIMULTANEOUSLY
  .
  In the processing of an asynchronous response message the
  corresponding row in process context table is first read and
  after that updated. When two such message are received at the
  same time a deadlock can occur.
  

Local fix

Problem summary

• Problem Description:
  
  The For Each activity repeatedly runs the activities that it
  contains either sequentially or in parallel
  for a specified number of iterations. In case of parallel
  execution, the iterations may deadlock on the
  For Each instance when they determine their individual iteration
  counter value. In that case the following
  stack trace appears in the logs whereas the database JDBC driver
  classes may vary depending on the
  database system being used:
  
  [07/06/13 03:14:30:234 GMT] 00000113 TraceBPE      3
  com.ibm.bpe.engine.BpelEngine.onMessage(BpelEngine.java:1564)
  <Null Message>
  
  com.microsoft.sqlserver.jdbc.SQLServerException: Transaction
  (Process ID 165) was deadlocked on lock resources with another
  process and has been chosen as the deadlock victim. Rerun the
  transaction.
          at
  com.microsoft.sqlserver.jdbc.SQLServerException.makeFromDatabase
  Error(SQLServerException.java:197)
          at
  com.microsoft.sqlserver.jdbc.SQLServerResultSet$FetchBuffer.next
  Row(SQLServerResultSet.java:4762)
          at
  com.microsoft.sqlserver.jdbc.SQLServerResultSet.fetchBufferNext(
  SQLServerResultSet.java:1682)
          at
  com.microsoft.sqlserver.jdbc.SQLServerResultSet.next(SQLServerRe
  sultSet.java:955)
          at
  com.ibm.ws.rsadapter.jdbc.WSJdbcResultSet.next(WSJdbcResultSet.j
  ava:3120)
          at
  com.ibm.bpe.database.DbAccActivityInstanceB.verifyVersionAndLock
  (DbAccActivityInstanceB.java:1005)
          at
  com.ibm.bpe.database.ActivityInstanceB.verifyVersionAndLock(Acti
  vityInstanceB.java:2433)
          at
  com.ibm.bpe.database.TomInstanceCache.verifyObjectInDb(TomInstan
  ceCache.java:385)
          at
  com.ibm.bpe.database.TomInstanceCache.get(TomInstanceCache.java:
  364)
          at
  com.ibm.bpe.database.ActivityInstanceB.selectCacheByPIIDATIDEHII
  DnullEnclosingFEIIDnull(ActivityInstanceB.java:2133)
          at
  com.ibm.bpe.database.Tom.getActivityInstanceB(Tom.java:12109)
          at
  com.ibm.bpe.engine.BpelVariableHandler.initializeVariable(BpelVa
  riableHandler.java:3213)
          at
  com.ibm.bpe.engine.BpelVariableHandler.createVariableInstancesFo
  rScope(BpelVariableHandler.java:1734)
          at
  com.ibm.bpe.engine.BpelScopeStateReady.startScopeInstance(BpelSc
  opeStateReady.java:65)
          at
  com.ibm.bpe.engine.BpelActivityKindComplexBegin.doActivate(BpelA
  ctivityKindComplexBegin.java:233)
          at
  com.ibm.bpe.engine.BpelActivityStateInactive.activateOrSkip(Bpel
  ActivityStateInactive.java:280)
          at
  com.ibm.bpe.engine.BpelEngineCore.continueControlLink(BpelEngine
  Core.java:291)
          at
  com.ibm.bpe.engine.BpelContinueLinkMessage3.execute(BpelContinue
  LinkMessage3.java:88)
          at
  com.ibm.bpe.engine.BpelEngine.onMessage(BpelEngine.java:1536)
          ...
  
  
  Problem Summary:
  
  The problem did happen because each individual iteration
  accesses the For Each activity to determine
  its individual iteration counter value.
  

Problem conclusion

• The problem has been fixed by passing the iteration counter
  value directly to each individual
  iteration and thus avoiding the need to access the For Each
  activity to determine the individual
  iteration counter value. Once installed, it is not advisable to
  remove the fix as long as there are
  messages in the BPEIntQueue. This is because the fix contains a
  new process navigation message that
  might stay on the BPEIntQueue and that would not be readable any
  more if the fix would be removed
  
  Problem will be fixed with V8012
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

  IV41165

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  BPM ADVANCED

• Fixed component ID

  5725C9400

Applicable component levels

• R801 PSY

     UP