PI27256: A queue manager's cluster operations such as REFRESH CLUSTER appear not to be working. Updates are not processed.

Obtain the fix for this APAR.

APAR status

• Closed as unreproducible in next release.

Error description

• The queue manager process amqrrmfa is not processing its work on
  SYSTEM.CLUSTER.COMMAND.QUEUE in
  a timely manner. Commands like RESET and REFRESH cluster
  commands
  seem not to work. No updates to cluster objects are being
  reflected in the local
  queue manager's cluster cache.
  ADDITIONAL SEARCH KEYWORD SCCQ
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All users of WebSphere MQ for z/OS Version 7 *
  *                 Release 1 Modification 0                     *
  ****************************************************************
  * PROBLEM DESCRIPTION: Incorrect chaining in the subscription  *
  *                      hash chains create subscriptions which  *
  *                      are never deleted. This may lead to     *
  *                      backlogs on the full repository's       *
  *                      SYSTEM.CLUSTER.COMMAND.QUEUE            *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  When rfxAddClSub adds a subscription it places it alphabetically
  on a hash chain. If there is already an existing subscription it
  will point to it with a pointer pPrevClSub, and will remember
  the previous entry to that one with another pointer, plClSub.
  plClSub therefore will point to the last entry in the chain that
  was not the same as the one that is being added.
  However, rfxAddClsub then goes on to set plClSub to pPrevClSub,
  and this is incorrect.
  Then the pPrevClSub entry is given a cache flag of CHANGED.
  Normally this works fine because we would not expect a new
  subscription to arrive for another month, by which time the
  CHANGED entry will be deleted by garbage collection and the
  chaining error will be resolved.
  However if a new subscription arrives in the meantime (because
  of some bug, or perhaps because a refresh is done on a partial
  repository), then the new subscription will be added to the
  chain between the CHANGED entry and the existing LIVE entry.
  This means that we have two live entries, the second of which
  will never be removed.
  If the hash chains contain a large number of these superfluous
  live subscriptions then the chains can take a long time to
  navigate, and hardening to the SYSTEM.CLUSTER.REPOSITORY.QUEUE
  may take a long time, leading to possible backlogs in the
  SYSTEM.CLUSTER.COMMAND.QUEUE.
  

Problem conclusion

Temporary fix

• *********
  * HIPER *
  *********
  

Comments

• rfxAddClSub has been changed so that new subscriptions are
  correctly chained in the hash chain.
  

APAR Information

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

  IC93358

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

  UI22469

Modules/Macros

• CMQXRFXC
  

Fix information

• Fixed component name

  WMQ Z/OS V7

• Fixed component ID

  5655R3600

Applicable component levels

• R100 PSY UI22469

     UP14/12/04 P F412 Ž

Fix is available

• Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.