PH29018: VSAM LOG READER CAUSES CAC00112E AND ABENDS DURING SHUTDOWN DUE TO INCORRECT THREAD COORDINATION

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• VSAM log reader causes CAC00112E and abends during shutdown due
  to incorrect thread coordination
  

Local fix

• NA
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: Classic CDC for VSAM, Version 11.3           *
  *                 Data Replication for VSAM, Version 11.3      *
  *                 VSAM Remote Source, Version 11.4             *
  ****************************************************************
  * PROBLEM DESCRIPTION: CAC00112E errors during shutdown of     *
  *                      source server with VSAM log reader and  *
  *                      subsequent 0C4 abend.                   *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  Timing issues during shutdown led to a second free of the
  log reader event and caused CAC00112E errors.  There are
  multiple threads involved and coordinating during shutdown:
  LRS (log reading service), LRI (log reading interface), SRBs
  .
  1.  LRS receives the request to disconnect the subscription
       change stream and sets a state of disconnect_waiting for
       the SRB to end its processing of the log stream.
  2.  The SRB will end its processing and remove the log stream
       read event.  The SRB sets a state of disconnecting
       indicating it's done and posts the LRI TCB to complete
       cleanup.
  3.  The LRI TCB does various cleanup activities, including
       removing the stream from the LRI list of streams and sets
       a state indicating disconnect was completed and posts LRS
       to complete the notification.
  .
  This normally works without problem.  However, if between
  numbers 2&3 above the LRS thread also initiates server shutdown
  then additional processing to clean up stream resources occurs.
  At this point, the stream is still on the list, but the reading
  event was removed.  Once the stream read event has been removed
  it is no longer safe to use the event storage.
  

Problem conclusion

• To avoid these timing issues the pointer to the event control
  block must be set to NULL when the event is initially removed
  so that a second attempt to delete the event is not done even
  if the change stream is still on the list.
  

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:

  UI71554

Modules/Macros

•    CECCLRVR
  

Fix information

• Fixed component name

  CLASSIC FED & E

• Fixed component ID

  5697I8200

Applicable component levels

• RB30 PSY UI71554

     UP20/09/16 P F009

Fix is available

• Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.