IT36110: Many PCF status messages flow through the channels of a Uniform Cluster even though no reconnectable applications are present

APAR status

• Closed as program error.

Error description

• When using a uniform cluster, a  large number of status messages
  are being sent through cluster channels although no
  re-connectable clients are present.
  These are internal MQPCF status messages which bounce between
  queue managers as a result of them believing that their system
  topics, are no longer a local TOPICHOST.
  The problem can also impact the ability for applications to be
  re-balanced correctly
  
  Use runmqsc to display the topic status and check the CLROUTE
  and PUBCOUNT
   - DIS TPSTATUS ('$SYS/MQ/INTERQM/BROADCAST') TYPE(TOPIC)
  

Local fix

Problem summary

• ****************************************************************
  USERS AFFECTED:
  Users of Uniform Cluster configurations.
  
  
  Platforms affected:
  MultiPlatform
  
  ****************************************************************
  PROBLEM DESCRIPTION:
  When a queue manager starts, it will restore locally defined
  topic objects to the topic tree.
  After the topic tree has been initialized with defined topics,
  it can be updated with entries from the cluster cache. During
  cache restoration, updates to in the topic tree are made but
  this should not include topics which had previously been marked
  as 'DELETED' as this can modify the internal flags for topics
  which are defined locally.
  
  If the topic being updated is defined with CLROUTE(TOPICHOST)
  set, the internal flags were incorrectly updated, removing the
  local flag. Messages published to the TOPICSTR of the topic,
  would attempt to find an alternative TOPICHOST elsewhere in the
  cluster configuration, excluding the locally defined topic.
  
  If all TOPICHOST's (for a given topic string) were in this
  confused state, messages would continually flow from one queue
  manager to another attempting to find a suitable host which was
  considered to be locally defined. This appeared as though a vast
  quantity of messages are flowing through the cluster.
  
  With respect to Uniform Clusters, there are system topics which
  are defined with CLROUTE(TOPICHOST) that can cause the issue
  when they are deleted from the cluster.
  These topics can be marked as deleted from the cluster as a
  result of
   - issuing REFRESH CLUSTER(<uniform cluster name>)
   - removing the queue manager from the Uniform Cluster (and
  subsequently re-introducing back into the cluster)
  

Problem conclusion

• The problem is resolved by preventing deleted topics from
  updating the topic tree during cache restore operations.
  The restore operations can result from :
  	- After restarting the Queue Manager.
  	- After enabling the SYSTEM.CLUSTER.COMMAND.QUEUE for MQGET
  operations.
  
  ---------------------------------------------------------------
  The fix is targeted for delivery in the following PTFs:
  
  Version    Maintenance Level
  v9.1 LTS   9.1.0.9
  v9.2 LTS   9.2.0.4
  v9.x CD    9.2.4
  
  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

  MQ BASE V9.2

• Fixed component ID

  5724H7281

Applicable component levels