IJ29555: NSD RETRY LOOP CONTINUOUS RESIGN ISSUE

APAR status

• Closed as program error.

Error description

• current design of GNR, RG will resign due to vdisk with
  8+2p failing fault tolerance whenever RG is recovered,
  and after enough threshold of this pattern reached,
  I/O will be returned as error for any disks retrying the
  I/O.
  

Local fix

Problem summary

• In mirrored disk environment, I/O is expected to
  continue with the surviving disk if a disk experiences a
  problem. In the case of a disk being created with a
  recovery group vdisk, and the recovery group is in a state
  that it continues to resign due to some vdisk's fault
  tolerance exceeded after a successful recovery,
  a race condition exists which could cause the logic of
  checking this state to be skipped. As a result
  of this, I/O will  continue to be  retried to the problem
  disk instead of moving on to the surviving
  disks.
  

Problem conclusion

• Benefits of the solution:
  With the fix, I/O will continue with the surviving
  disks instead of hanging with the problem disk.
  
  Work around:
  None
  
  Problem trigger:
  A race condition in the logic that determines if a
  recovery group is repetitively experiencing
  a resign after a successful recovery due to
  a fault tolerance being reached.
  
  Symptom:
  I/O hang due to long waiters  'waiting for stateful
  NSD server error takeover (2)'
  
  Platforms affected:
  N/A
  
  Functional Area affected:
  GNR
  
  Customer Impact:
  Suggested
  Changed Externals:
  None
  

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

  SPEC SCALE STD

• Fixed component ID

  5737F33AP

Applicable component levels