PM79860: Failover placement can occur prematurely if a container fails before the grid is initially placed.

APAR status

• Closed as program error.

Error description

• Even with placement suspended via suspendBalance, placement can
  occur if a container fails. This is by design to avoid
  scenarios in which failover does not occur.
  
  However, the failover placement can occur if a container fails
  before the data grid is initially placed.  This can cause more
  issues in larger environments if placement is triggered
  prematurely.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:  Any eXtreme Scale user, particularly those  *
  *                  with large numbers of partitions (1000 or   *
  *                  more).                                      *
  ****************************************************************
  * PROBLEM DESCRIPTION: When the catalog server submits         *
  *                      placement work that takes over 30       *
  *                      seconds to complete, some of that       *
  *                      work might time out.                    *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  Timed out placement that is sent to container servers might
  not complete.  Artifacts of timed out worked are cleaned
  up on the catalog server. However, the work is not resubmitted
  to the container servers.
  

Problem conclusion

• The catalog server resends the timed out work.
  

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

  WS EXTREME SCAL

• Fixed component ID

  5724X6702

Applicable component levels

• R850 PSY

     UP

• R860 PSY

     UP