IV70630: JMS HANG WHEN MESSAGECONSUMER AND MESSAGEPRODUCER ARE CREATED FROM SAME SESSION AND CLOSED SIMULTANEOUSLY ON TWO THREADS.

APAR status

• Closed as program error.

Error description

• A JMS consumer application hangs and stops processing messages
  from the queue
  causing messages to accumulate on the queue. There are no
  errors in the logs. Stopping queue processing has no effect,
  and restarting does not work because the thread is still
  attached to the queue. The only resolution is to stop and
  restart the JMS application.
  

Local fix

• Restart the JMS application.
  

Problem summary

• ****************************************************************
  USERS AFFECTED:
  This issue affects users of the WebSphere MQ classes for JMS
  v7.0.1.12.
  
  This includes users of WebSphere Application Server which bundle
  the WebSphere MQ Resource Adapter v7.0.1.12.
  
  
  Platforms affected:
  MultiPlatform, AIX, HP-UX Itanium, HP-UX PA-RISC, IBM iSeries,
  Linux on Power, Linux on S390, Linux on x86, Linux on x86-64,
  Linux on zSeries, Solaris SPARC, Solaris x86-64, Windows, z/OS
  
  ****************************************************************
  PROBLEM DESCRIPTION:
  APAR IV46878:
  
    http://www.ibm.com/support/docview.wss?uid=swg1IV46878
  
  introduced a code change to address a problem where a JMS
  application would hang in the case where:
  
  - an attempt was made to close a MessageConsumer but the
  connection to the queue manager had been disconnected, such that
  an MQRC 2009 'MQRC_CONNECTION_BROKEN' exception was thrown
  during the close request,
  
  and then subsequently
  
  - the application created a new MessageConsumer object (from the
  same JMS Session) and attempted to receive a message using it.
  
  The code change associated with that APAR introduced a new
  deadlock, which occurred when a request was made to close both a
  MessageProducer (or QueueSender, etc.) and a MessageConsumer (or
  QueueReceiver, etc.) simultaneously on two different threads.
  
  When this issue occurred, a Javacore would show two threads in
  the following Java stacktrace state:
  
  
  Thread 1: MessageProducer.close()
  ----------------------------------------------------------------
  ------
  com/ibm/msg/client/wmq/internal/WMQMessageProducer$SpiIdentified
  ProducerShadow.close(WMQMessageProducer.java:687(Compiled Code))
  com/ibm/msg/client/wmq/internal/WMQMessageProducer.close(WMQMess
  ageProducer.java:1147(Compiled Code))
  com/ibm/msg/client/jms/internal/JmsMessageProducerImpl.close(Jms
  MessageProducerImpl.java:205(Compiled Code))
  com/ibm/msg/client/jms/internal/JmsMessageProducerImpl.close(Jms
  MessageProducerImpl.java:165)
  com/ibm/mq/jms/MQQueueSender.close(MQQueueSender.java:87)
  ----------------------------------------------------------------
  ------
  
  
  Thread 2: MessageConsumer.close()
  ----------------------------------------------------------------
  ------
  java/lang/Object.wait(Native Method)
  java/lang/Object.wait(Object.java:167(Compiled Code))
  com/ibm/msg/client/commonservices/ReentrantMutex.acquire(Reentra
  ntMutex.java:136(Compiled Code))
     (entered lock:
  com/ibm/msg/client/commonservices/ReentrantMutex@0x0000000012018
  700, entry count: 2)
  com/ibm/msg/client/commonservices/ReentrantMutex.acquire(Reentra
  ntMutex.java:70(Compiled Code))
     (entered lock:
  com/ibm/msg/client/commonservices/ReentrantMutex@0x0000000012018
  700, entry count: 1)
  com/ibm/msg/client/wmq/internal/WMQConsumerShadow.close(WMQConsu
  merShadow.java:404(Compiled Code))
     (entered lock: java/lang/Object@0x0000000011DED090, entry
  count: 1)
  com/ibm/msg/client/wmq/internal/WMQMessageConsumer.close(WMQMess
  ageConsumer.java:389(Compiled Code))
  com/ibm/msg/client/jms/internal/JmsMessageConsumerImpl.close(Jms
  MessageConsumerImpl.java:326(Compiled Code))
  com/ibm/msg/client/jms/internal/JmsMessageConsumerImpl.close(Jms
  MessageConsumerImpl.java:190(Compiled Code))
  com/ibm/mq/jms/MQQueueReceiver.close(MQQueueReceiver.java:80(Com
  piled Code))
  ----------------------------------------------------------------
  ------
  
  
  
  This deadlock lasted indefinitely, meaning that the Java process
  must be killed in order to restart it and resume application
  operation.
  
  While APAR IV46878 was also included in WebSphere MQ 7.5.0.3 and
  7.1.0.5, the issue does not affect those releases. Only
  WebSphere MQ 7.0.1.12 is affected by by the issue.
  

Problem conclusion

• The WebSphere MQ classes for JMS locking logic has been updated
  to revert to the code behaviour prior to WebSphere MQ 7.0.1.12,
  while maintaining the fix which APAR IV46878 was intended
  resolve.
  
  ---------------------------------------------------------------
  The fix is targeted for delivery in the following PTFs:
  
  Version    Maintenance Level
  v7.0       7.0.1.13
  
  The latest available maintenance can be obtained from
  'WebSphere MQ Recommended Fixes'
  http://www-1.ibm.com/support/docview.wss?rs=171&uid=swg27006037
  
  If the maintenance level is not yet available information on
  its planned availability can be found in 'WebSphere MQ
  Planned Maintenance Release Dates'
  http://www-1.ibm.com/support/docview.wss?rs=171&uid=swg27006309
  ---------------------------------------------------------------
  

Temporary fix

Comments

APAR Information

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

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

Fix information

• Fixed component name

  WMQ LIN X86 V7

• Fixed component ID

  5724H7224

Applicable component levels