PK77121: ABEND1C5 RC=804 IN WEBSPHERE CONTROLLER FOLLOWED BY ABENDEC3 RC=160D000D WHEN LOGSTREAM IS USED FOR XA PARTNER LOG

APAR status

• Closed as program error.

Error description

• When logstream is defined for XA partner log and logstream was
  created with retention period greater than 0, ie.
  RETPD(#ofDays), WebSphere controller may fail to start.  The
  likely scenario is that the server will start fine first couple
  of times, but fails on a subsequent restart that occurred within
  the retention period specified in the logstream.  Once the
  server does fail to start, it will not be able start again
  unless logstream is redefined.
  
  Symptoms include ABEND1C5 RC=804 in logger module IXGF2DEL
  issued in WebSphere application controller address space:
  
  10 04 59 49 BBOS001  SFTABN 00 PIDS/5752SCLOG RIDS/IXGINLPA#L
      RIDS/IXGF2DEL AB/S01C5 PRCS/00000804 REGS/0BE47 REGS/0A0C2
  10 04 59 49   N/A    SFTSYM 00 PIDS/5655I3500 RIDS/BBOGLOG
   RIDS/BBOGLOG  PRCS/00000008
  
  followed by ABENDEC3 RC=160D000D in WAS module BBOGLOG
  followed by ABEND U4087 RC=8
  
  The traceback for the TCB with EC3 completion code:
  
  Native Stack:
  Java_com_ibm_ws390_recoverylog_spi_RecoveryLogCompressionAla
  mmipInvoke_V_V
  mmipInvokeLazyJniMethod
  ...
  
  Java Stack:
  com/ibm/ws390/recoverylog/spi/RecoveryLogCompressionAlarm.
    compressMain
  com/ibm/ws390/recoverylog/spi/RecoveryLogCompressionAlarm.alarm
  com/ibm/ejs/util/am/_Alarm.run
  com/ibm/ws/util/ThreadPool$Worker.run
  
  The reason the server fails is because during startup, server
  obtains a copy of the logstream from the logger with all
  records, which includes records marked for deletion.  With the
  fix on, WebSphere will ask logger to pass active records only.
  
  Note: Similar symptoms can occur if you specify AUTODELETE(YES).
  XA Partner log logstreams need to be defined with AUTODELET(NO).
  The fix for this apar does not address this scenario.
  

Local fix

• Redefine the logstream with retention period of zero, ie.
  RETPD(0).  This has no side effect on WebSphere in any way, so
  there is no need to reset RETPD to a non-zero value even with
  this apar applied.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED:  All users of WebSphere Application Server   *
  *                  V6.0.1 for z/OS who defined and are using   *
  *                  logstreams for the XA partner log.          *
  ****************************************************************
  * PROBLEM DESCRIPTION: WebSphere application                   *
  *                      server controller region terminates     *
  *                      with abend code EC3 RC=160D000D.        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  WebSphere's Application Server controller region terminates
  with abend code EC3 RC=160D000D, when transaction recovery
  services attempts to delete records that have been logically
  deleted.
  For this abend to take place, the WebSphere Application
  Server must be configured to use logstreams to store XA
  resource information and the retention period must be set to a
  number greater than zero.
  If the above conditions are met and there are
  logically deleted records in the logs, the transaction recovery
  services incorrectly reads these records and attempts to
  process them.
  
  The following stack traces can be seen for the
  failing TCB:
  
  Native Stack:
  Java_com_ibm_ws390_recoverylog_spi_RecoveryLogCompressionAla
  mmipInvoke_V_V
  mmipInvokeLazyJniMethod
  ...
  
  Java Stack:
  com/ibm/ws390/recoverylog/spi/RecoveryLogCompressionAlarm.
  compressMain
  com/ibm/ws390/recoverylog/spi/RecoveryLogCompressionAlarm.alarm
  com/ibm/ejs/util/am/_Alarm.run
  com/ibm/ws/util/ThreadPool$Worker.run
  
  ================================================================
  Another symptom fixed by this apar is a NPE -
  NullPointerException:
  
  BBOO0220E: J2CA0009E: An exception occurred while trying to
  instantiate the ManagedConnectionFactory class com.ibm.connec
  tor2.ims.ico.IMSManagedConnectionFactory used by resource
  eis/IMSBA3 : java.lang.NullPointerException
  .at com.ibm.ws390.recoverylog.spi.IXGRecoveryLogImpl.filteredRec
    overableUnits(IXGRecoveryLogImpl.java:544)
  .at com.ibm.ws390.recoverylog.spi.IXGRecoveryLogImpl.recoverable
    Units(IXGRecoveryLogImpl.java:509)
  .at com.ibm.ws.Transaction.JTA.PartnerLogTable.findEntry(Partner
    LogTable.java:322)
  .at com.ibm.ws.Transaction.JTA.TranManagerSet.registerResourceIn
    fo(TranManagerSet.java:609)
  .at com.ibm.ws.Transaction.JTA.TranManagerSet.registerResourceIn
    fo(TranManagerSet.java:590)
  .at com.ibm.ejs.j2c.ConnectionManager.<init>(ConnectionManager.j
    ava:480)
  .at com.ibm.ejs.j2c.ConnectionFactoryBuilderServerImpl.createCFD
    _Common(ConnectionFactoryBuilderServerImpl.java:359)
  

Problem conclusion

• Code was updated to only read, during restart, active data and
  not logically deleted data from the logs.
  
  APAR PK77121 is currently targeted for inclusion in Service
  Level (Fix Pack) 6.0.2.35 of WebSphere Application Server V6.0.1
  for z/OS.
  
  Please refer to URL:
  //www.ibm.com/support/docview.wss?rs=404&uid=swg27006970
  for Fix Pack availability.
  

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:

  PK77123 PK77124

Fix information

• Fixed component name

  WEBSPHERE FOR Z

• Fixed component ID

  5655I3500

Applicable component levels

• R601 PSY UK46540

     UP09/05/30 P F905

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.