IT02187: ASSERT FAILURE AND CRASH ON PROMOTION OF HDR SECONDARY TO PRIMARY WITH HEURISTICALLY COMPLETED XA TRANSACTION.

APAR status

• Closed as program error.

Error description

• An orphaned XA transaction is heuristically completed, for
  example when it becomes a long tx, on the primary server of an
  HDR pair. If a failover then occurs the newly promoted HDR
  secondary will complete logical recovery with an open
  transaction and then attempt to abort that transaction. Two
  different types of Assert Failure may occur depending on whether
  the logical log containing the start of the transaction is still
  available on the server.
  
  For the case where the log is still available a rollback error
  126 will appear in the online log file with messages in the
  Assert Failure file similar to:
  
  doitem failed(iserrno=111): could not delete item
  Node 0x1 fragid 0x1002a1 rowid 0x202, key:
  ...
  Assert Warning: Error encountered during rollback.
  Who: Session(23, informix@gundabad, 0, 7000000203a7a30)
           Thread(50, clean_ltx, 70000002036cd30, 1)
           File: rskey.c Line: 1647
  Results: Index 'stores_demo:"informix".customer#zip_ix' is now
  unusable
  Action: Run 'oncheck -cI stores_demo:"informix".customer#zip_ix'
  
  The stack for the thread will be similar to:
  
  afstack
  afhandler
  afwarn_interface
  kybad
  doitem
  logundo
  rlogm_undo
  rollback
  rsrollback
  cleantx
  clean_ltx
  th_init_initgls
  startup
  
  This will result in multiple rollback errors accompanied by an
  Assert Failure and the instance shuts down. An attempt to
  restart the newly promoted instance will result in further
  rollback errors and the instance will be shut down.
  
  For the second case where the logical log is no longer available
  the Assert Failure will be similar to:
  
  log_get: get_logfile_by_id() failed
  log_get: cannot read loguniq 273 logpos 0x1fe1cc
  logm_read: cannot read loguniq 273 logpos 0x1fe1cc
  logerr('relock() - logread failed')
  tx 0x203a7d68, tx_flags 0x2008062b
  tx_loguniq 282, tx_logpos 0x80cc
  
  With the stack for the thread similar to:
  
  afstack
  afhandler
  afcrash_interface
  logerr
  relock
  txcleanup
  rsclose_lgr
  dr_lgr_end
  dr_finish_recovery
  dr_mode
  onmode_monitor
  th_init_initgls
  startup
  
  In this scenario restarting the server instance will result in
  it coming up as the HDR secondary once again.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * Users of secondary servers and XA transactions               *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * See Error Description                                        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * Update to IDS-11.70.xC9                                      *
  ****************************************************************
  

Problem conclusion

• Problem Fixed In IDS-11.70.xC9
  

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

  INFORMIX SERVER

• Fixed component ID

  5725A3900

Applicable component levels

• RB70 PSY

     UP

• RC10 PSY

     UP