PK11200: RECORD COUNT DIFFERENCE BETWEEN FULL AND PITR

APAR status

• Closed as program error.

Error description

• If the recovery of 4 or more Fast Path areas are compared after
  a Full Recovery, to a Point in Time Recovery there could be a
  noticiable difference in the unload results. The cause is the
  handling of 5947x log records. Data Integrity is not affected
  but a visual comparison of the unloaded database appears
  different between a Full Recovery and a PITR.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: IMS Database Recovery Facility Version 2     *
  *                 Release 1 users are affected.                *
  ****************************************************************
  * PROBLEM DESCRIPTION: Data corruption was identified after a  *
  *                      DRF PITR of a group of Fast Path AREAs  *
  *                      with a large count of log updates.      *
  ****************************************************************
  * RECOMMENDATION: INSTALL CORRECTIVE SERVICE FOR APAR/PTF      *
  ****************************************************************
  At the end of DRF reading log data sets, multi threaded read
  drops log records under high stress conditions.  Each read
  instance consists of 5 read threads.  A secondary problem was
  detected under stress conditions in log read when a log read
  instance does not complete termination, recovery hangs and the
  MSGFRD4219I is not issued for all read instances.  A similar
  hang can occur while reading change accumulation data sets.
  

Problem conclusion

• AIDS: RIDS/UTIL RIDS/DBS DBS/UTIL
    DEP: NONE
    GEN:
  
  *** END IMS KEYWORDS ***
  The DRF log read process is simplified to be a single thread.
  Even though the reads are designed to be in parallel, they end
  up being dispatched single thread.  The condition of dropped
  records is eliminated without affecting performance.
  The secondary problem of the read instance hang is fixed by this
  change since the read process does not wait for multiple threads
  to complete termination processing.  A similar change is made
  for the change accumulation read instance for a hang reading
  change accumulation data sets and the MSGFRD4219I is not issued
  for all change accumulation read instances.  A change is made in
  FRXUORM0 to add diagnostics to count the number of log records
  detected during the recovery process.  Additionally, if a buffer
  is unable to be processed by FRXUORM0, DRF abnormally terminates
  with an ABENDU0385 RSN000A.
  

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:

  UK11983

Modules/Macros

• FRXCAMG0 FRXCARD0 FRXCON   FRXRBUF0 FRXRCTL0
  FRXRDTH0 FRXUORM0 FRXWSPM0 FRXYALL0
  

Fix information

• Fixed component name

  IMS DB RECOVERY

• Fixed component ID

  5655I4400

Applicable component levels

• R210 PSY UK11983

     UP06/02/23 P F602