IJ17503: AFTER TIEBREAKER DECISION, LOSING NODE(S) MIGHT NOT REBOOT

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• When a sundered domain occurs in a PowerHA v7 cluster
  with TieBreaker configured, there is a risk of the
  node(s) losing TieBreaker not rebooting when they
  should, if certain notifications to clients such as
  PowerHA's clstrmgr process are not handled as expected.
  
  The result will be both "sides" of the split will remain
  running, and a Domain Merge (GS_DOM_MERGE_ER) will occur
  later when connectivity between all nodes is restored,
  which will cause nodes to be rebooted at that time.
  

Local fix

Problem summary

• While processing the TieBreaker action, if a client of Group
  Services (HAGS) is delayed in responding or has other issues,
  HAGS may not continue processing the necessary TB action on
  its part, resulting in the node not being stopped as expected.
  

Problem conclusion

• The client logic in HAGS has been corrected so that necessary
  TB activity can be handled even if a client being warned of
  the action does not respond to the notification.
  

Temporary fix

Comments

APAR Information

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

  IJ16923

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

  U889479

Fix information

• Fixed component name

  RSCT FOR AIX

• Fixed component ID

  5765F07AP

Applicable component levels

• R324 PSY U889479

     UP21/04/30 I 1000

PTF to Fileset Mapping

• U889479 rsct.basic.rte 3.2.4.3

• U889479 rsct.basic.rte 3.2.4.3

• U889479 rsct.basic.rte 3.2.4.3