JR34388: RECORD LOCK TABLE NOT REMOVING LOCKS, CAUSING PROJECT LOCKED ANDJOB RUN FAILURES

APAR status

• Closed as program error.

Error description

• Locks in the Universe record lock table are being removed
  correctly. The net result is that jobs can fail to start, and
  D/S CLients fail to connect.
  

Local fix

• A deadlock daemon was provided to clear up the record lock table
  but this has to be executed manually
  

Problem summary

• ****************************************************************
  USERS AFFECTED:
  Found on Solaris DataStage 7.5.2 but problem is most likely not
  limited to that particular platform or DataStage version
  ****************************************************************
  PROBLEM DESCRIPTION:
  Under certain conditions, shared record locks are not removed
  from their table in disk shared memory.  This can lead to a
  situation where new DataStage clients and jobs cannot be started
  because a row in the lock table is full.
  ****************************************************************
  RECOMMENDATION:
  A patch was built for DataStage 7.5.2 on Solaris.
  ****************************************************************
  

Problem conclusion

• New functionality was added to the deadlock daemon (dsdlockd) to
  check for and remove shared record locks which have been
  abandoned.  The following configuration options in
  $DSHOME/dsdlockd.config must present to enable this
  functionality:
  
  start=1
  recordlocks=1
  

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:

Modules/Macros

• ENGINE
  

Fix information

• Fixed component name

  WIS DATASTAGE

• Fixed component ID

  5724Q36DS

Applicable component levels

• R752 PSY

     UP

• R753 PSN

     UP

• R801 PSN

     UP

• R810 PSN

     UP