PH19731: Managed File Transfer source agent abends after a managed transfer resumes after coming out of recovery

APAR status

• Closed as duplicate of another APAR.

Error description

• An IBM MQ Managed File Transfer V9.1.0.0 agent is acting as the
  source agent for managed transfers. After a managed transfer
  comes out of recovery and restarts, the agent abends and
  generates an ABEND file containing information similar to the
  information shown below:
  
  Filename:
  /SYSTEM/var/mqmfte/mqft/logs/MFT/agents/AGENT1/logs/ffdc/ABEND.F
  TE.20190409111213969.7128783014550594708.log
  Level:      MQMFT 9.1.0.0 V910-GA910-L180705
  Time:       09/04/2019 11:12:13:969 GMT
  Thread:     37 (ReplyProcessor)
  Class:      com.ibm.wmqfte.thread.FTEThread
  Instance:   f75cd29e
  Method:     uncaughtException
  Probe:      ABEND_001
  Cause:      java.lang.NullPointerException
  java.lang.NullPointerException
  	at
  com.ibm.wmqfte.transfer.frame.impl.TransferFrameSenderImpl.updat
  e(TransferFrameSenderImpl.java:861)
  	at
  com.ibm.wmqfte.transfer.impl.TransferSenderRunnable.processAckno
  wledgement(TransferSenderRunnable.java:1406)
  	at
  com.ibm.wmqfte.transfer.impl.TransferSenderImpl.processAcknowled
  gement(TransferSenderImpl.java:172)
  	at
  com.ibm.wmqfte.statestore.impl.FTEStateStoreImpl.notifyAcknowled
  gement(FTEStateStoreImpl.java:803)
  	at
  com.ibm.wmqfte.transfer.impl.ReplyProcessorRunnable.run(ReplyPro
  cessorRunnable.java:258)
  	at java.lang.Thread.run(Thread.java:812)
  	at com.ibm.wmqfte.thread.FTEThread.run(FTEThread.java:70)
  

Local fix

• N/A
  

Problem summary

• This issue affects users of IBM MQ Managed File Transfer.
  
  When performing a managed transfer, a source agent maintains a
  list of:
  - The source items (files or messages) for the managed transfer.
  - And the file channels associated with each source item.
  
  The list is initialized when a managed transfer starts, and the
  source items added to it. However, the file channels associated
  with the source items are only added to the list when the source
  agent starts processing that particular item.
  
  The following sequence of events led up to the issue reported in
  this APAR:
  
  - The source agent started processing a managed transfer
  containing multiple transfer items.
  - The source agent opened a source file for the first transfer
  item (for example, "source-file-1"), read file data from it
  and sent it to the destination agent. At this point, the source
  agent's internal list for the managed transfer looked like this:
  
  Items List:
  Item 1:  source-file-1], Channel channel1]
  Item 2:  source-file-2], Channel null]
  Item 3:  source-file-3], Channel null]
  ...
  - After waiting 5 minutes, the source agent hadn't had an
  acknowledgement from the destination agent indicating that it
  had received the file data, and so it put the managed transfer
  into recovery. As part of this processing, the channel to
  "source-file-1" was closed.
  - Some time later, the source agent received a internal message
  from the destination agent indicating that it was OK to resume
  the managed transfer. This caused the source agent to bring
  the managed transfer out of recovery, and build a new internal
  list of transfer items and channels, which looked like this:
  
  Items List:
  Item 1:  source-file-1], Channel null]
  Item 2:  source-file-2], Channel null]
  Item 3:  source-file-3], Channel null]
  ...
  - Shortly afterwards, the source agent received an internal
  Acknowledgement message from the destination agent, indicating
  that it had received the file data that was sent before the
  managed transfer went into recovery. This message also
  indicated that the first transfer item failed due to an issue
  writing the destination file.  As a result, the source agent
  got the channel for this item from the its internal list,
  and then tried to use it to get the name of the source file to
  include in the supplementary message:
  
  "BFGTR0012E: The transfer of <source_file> has failed because of
  a problem at the destination."
  
  for the managed transfer. However, there wasn't a channel for
  this transfer item, as the source file had not been opened
  since the managed transfer restarted, and so the attempt to get
  the source file name from it failed with a NullPointerException
  - As the NullPointerException was unexpected, the agent
  generated an FDC containing the following information:
  Thread:     <number> (ReplyProcessor)
  Class:      com.ibm.wmqfte.thread.FTEThread
  Method:     uncaughtException
  Probe:      ABEND_001
  Cause:      java.lang.NullPointerException
  java.lang.NullPointerException
   at com.ibm.wmqfte.transfer.frame.impl.TransferFrameSenderImpl.
  update()
   at com.ibm.wmqfte.transfer.impl.TransferSenderRunnable.process
  Acknowledgement()
   at com.ibm.wmqfte.transfer.impl.TransferSenderImpl.processAckn
  owledgement()
   at com.ibm.wmqfte.statestore.impl.FTEStateStoreImpl.notifyAckn
  owledgement()
   at com.ibm.wmqfte.transfer.impl.ReplyProcessorRunnable.run()
   at java.lang.Thread.run(Thread.java:812)
   at com.ibm.wmqfte.thread.FTEThread.run(FTEThread.java:70)
  
  and shut itself down.
  

Problem conclusion

• To resolve this issue, the agent has been modified to check if
  there is a channel associated with a transfer item, before
  trying to use that channel to get the source file name to
  include in the supplementary message:
  
  "BFGTR0012E: The transfer of <source_file> has failed because
  of a problem at the destination."
  
  for a failed managed transfer. If the item does not have an
  associated channel, then the agent will use a different
  mechanism to get the source file name. In the event that the
  agent is still unable to determine the source file name,
  the following supplementary message will be logged for the
  transfer item:
  
  "BFGTR0083E: The transfer has failed because of a problem at
  the destination."
  
  The fix is included in fix pack 9.1.0.6, APAR PH26933,
  and fix pack 9.0.0.10, APAR PH25070.
  

Temporary fix

Comments

•  The fix is included in fix pack 9.1.0.6, APAR PH26933,
  and fix pack 9.0.0.10, APAR PH25070.
  

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

  IBM MQ MFT Z/OS

• Fixed component ID

  5655MF900

Applicable component levels