IT36500: Transfer item fails with a BFGIO0006E error if the source agent abends and restarts partway through a transfer

APAR status

• Closed as program error.

Error description

• An MQ Managed File Transfer agent is acting as the source agent
  for a message-to-file transfer, where:
  
  - The message-to-file transfer contains multiple items.
  - And each item transfers an individual message to a unique
  file.
  - And the -de (destination file behaviour) parameter for each
  item is set to "Error".
  - And the -sd (source disposition) parameter for each item is
  set to "Delete".
  
  Partway through the managed transfer, the source agent abends
  due to an OutOfMemoryError and shuts down. When the source agent
  is restarted, the message-to-file resumes and completes.
  However, even though all of the destination files have been
  written successfully, one of the transfer items is marked as
  "Failed" with the supplementary message:
  
  BFGIO0006E:  File "<destination filename>" already exists.
  
  As a result, the message for this item is not deleted from the
  source queue.
  

Local fix

• 1. use enableMemoryAllocationChecking=true to enabled memory
  allocation checking before new transfers are submitted
  2. increase the Java heap size available to the agent on startup
  

Problem summary

• ****************************************************************
  USERS AFFECTED:
  This issue affects users of:
  
  - MQ 9.1 Managed File Transfer
  - MQ 9.2 Managed File Transfer
  
  who have agents that:
  
  - Process managed transfer requests containing multiple transfer
  items.
  
  
  Platforms affected:
  MultiPlatform
  
  ****************************************************************
  PROBLEM DESCRIPTION:
  During a managed transfer, the source agent reads data for a
  transfer item and sends it to the destination agent. The
  destination agent receives the data, and writes it to the
  destination item. As part of this processing, the two agents
  store checkpoint records - these contain information about the
  amount of data that has been transferred. If the managed
  transfer goes into recovery before it has finished, the source
  and destination agents use their respective checkpoint records
  to determine where the managed transfer should be restarted
  from.
  
  Checkpoints are taken at regular intervals, based on the value
  of the following agent properties:
  
  - agentCheckpointInterval 	
  - agentChunkSize 	
  - agentFrameSize 	
  - agentWindowSize
  
  At the start of a managed transfer, the source and destination
  agents perform some negotiation to determine what values to use
  for these properties, and when to take checkpoints.
  
  In addition to this:
  
  - The source agent will take a checkpoint after it has sent the
  last piece of data for a transfer item to the destination agent.
  - And the destination agent will take a checkpoint after it has
  received the last piece of data for a transfer item, and written
  it out successfully.
  
  The checkpoint records are stored in messages on the
  SYSTEM.FTE.STATE.agent_name queue. Both the source and
  destination agents will have a single message for a managed
  transfer on their respective queues, and each message will
  contain a maximum of three checkpoint records.
  
  For more information on the agent properties mentioned above,
  see the "The MFT agent.properties file"  topic in the MQ
  sections of IBM Documentation. For reference, the URI of this
  topic in the MQ 9.2 section of IBM Documentation is
  https://www.ibm.com/docs/en/ibm-mq/9.2?topic=reference-mft-agent
  properties-file
  
  
  Now, when the issue reported in this APAR occurred, the source
  agent stopped unexpectedly due to an abend:
  
  - After it had sent the last piece of data for a transfer item
  to the destination agent.
  - And before it had written the checkpoint record indicating
  that the data had been sent.
  
  When the destination agent received the data, it wrote it to a
  temporary file, renamed that file to be the specified
  destination filename and stored a checkpoint record indicating
  that the data had been processed successfully. At this point,
  the destination agent had one more checkpoint than the source
  agent, as the source agent had stopped due to an abend before it
  could write its checkpoint.
  
  When the source agent restarted, it contacted the destination
  agent to determine where the managed transfer should be
  restarted from. Because the destination agent had one more
  checkpoint than the source agent, the source agent decided to
  restart the managed transfer from its checkpoint and resend the
  data for the transfer item.
  
  The destination agent received the data and wrote it to a
  temporary file.  However, when it tried to rename that file to
  be the specified destination filename, it was unable to do so as
  the destination file already existed. As the managed transfer
  request was submitted with the -de (destination file behaviour)
  parameter set to "Error" for this transfer item, , the
  destination agent marked the item as "Failed" with the
  supplementary information:
  
  BFGIO0006E:  File "<destination filename>" already exists.
  

Problem conclusion

• To resolve this issue, MQ Managed File Transfer has been updated
  so that:
  
  - If a managed transfer goes into, and then comes out of,
  recovery due to the source agent stopping unexpectedly.
  - And, during the recovery processing for a managed transfer,
  the source agent determines that it has written one less
  checkpoint record that the destination agent.
  
  then the source agent will create a dummy checkpoint record and
  restart from the position requested by the destination agent.
  This ensures that the source agent doesn't resend any data that
  has already been processed by the destination agent, which
  prevents the issue reported in this APAR from occurring.
  
  Because of the dummy checkpoint, the source agent is unable to
  confirm that the data that it sent before it abended is exactly
  the same as the data that was successfully written by the
  destination agent (for example, the source file might have
  changed while the source agent was down). Because of this, the
  source agent will mark any items that are affected in this way
  as "Failed", with the new supplementary message:
  
  BFGSS0090E: Failed to generate audit information for the source
  transfer item.
  
  If a managed transfer contains one or more transfer items that
  have been marked as "Failed" with this supplementary messages,
  systems administrators will have to perform manual checks to
  verify the destination item contains all of the expected data.
  
  ---------------------------------------------------------------
  The fix is targeted for delivery in the following PTFs:
  
  Version    Maintenance Level
  v9.1 LTS   9.1.0.10
  v9.2 LTS   9.2.0.4
  v9.x CD    9.2.4
  
  The latest available MQ 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

  IBM MQ MFT V9.1

• Fixed component ID

  5724H7272

Applicable component levels