IC99032: MULTIPLE TOKEN NODES WHEN NODE1 GOES DOWN

5.2.6.0

APAR status

• Closed as program error.

Error description

• Multiple Token nodes when Node1 goes down
  The cluster needs to be recycled causing a full outage for a
  while.   there is a problem in checking if there is any active
  node already has token, the query should include constrain that
  TOKEN=1
  

Local fix

• cluster recycle
  

Problem summary

• USERS AFFECTED:
  ALL
  
  PROBLEM DESCRIPTION:
  MULTIPLE TOKEN NODES WHEN NODE1 GETS OOM AND IS NOT RESPONDING
  
  PLATFORMS AFFECTED :
  ALL
  

Problem conclusion

• RESOLUTION SUMMARY:
  THIS MULTIPLE TOKEN NODES ISSUE IS A TIMING ISSUE. THE FIX
  INCLUDES:
  1. HAVE A LOCK WHEN HEART BEAT THREAD UPDATES DATABASE TO CLEAN
  TOKEN NODE AND TAKES TOKEN FOR CURRENT NODE, SO WE DON'T HAVE
  TWO NODES TAKING THE TOKEN AT THE SAME TIME.
  2. UPDATE TO MAKE SURE NO ACTIVE NODE HAS THE TOKEN BEFORE
  TAKING THE TOKEN FOR CURRENT NODE.
  THESE CHANGE ENSURES ONLY ONE NODE TAKES TOKEN WHEN A NODE GOES
  DOWN.
  
  DELIVERED IN:
  5020500_1
  

Temporary fix

• Fix the multiple token issue in ops server
  

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

  STR B2B INTEGRA

• Fixed component ID

  5725D0600

Applicable component levels

• R525 PSY

     UP