IT13098: CLIENT CRASH DURING SNAPDIFF INCREMENTAL BACKUPS

IBM Spectrum Protectâ„¢ Client 7.1.6 Downloads and READMEs

APAR status

• Closed as program error.

Error description

• A Tivoli Storage Manager client can crash with a segmentation
  fault during SnapDiff incremental backup operations under
  specific timing conditions.
  
  Customer/L2 Diagnostics:
  A callstack similar to the following (Linux) example can
  be extracted from the dump file:
  
     #0  raise ()
     #1  abort ()
     #2  psTrapHandler(int) ()
     #3  <signal handler called>
     #4  InitializeTxnItem ()
     #5  tlBackObj ()
     #7  baSnapDiffCallBack ()
     #8  cacheObject::walkCallback ()
     #9  bTree::WalkTree ()
    #10  bTree::WalkTree ()
    #11  bTree::WalkTree ()
    #12  bTree::WalkTree ()
    #13  bTree::WalkTree ()
    #14  bTree::WalkTree ()
    #15  bTree::WalkTree ()
    #16  cacheObject::dbQuery ()
    #17  SnapdiffDB::SnapdiffDBQuery ()
    #18  baProcessRequest ()
    #19  baRequestHandler ()
    #20  DccTxnProducer::HandleQueue ()
    #21  DccTxnProducer::Run ()
    #22  DccTxnProducer::DoThread ()
    #23  callDoPThread2 ()
    #24  startThread2 ()
    #25  startThread ()
    #26  start_thread ()
    #27  clone ()
  
  Errors similar to the following may be reported in the
  dsmerror.log just prior to the crash:
  
    11/19/2015 01:34:53 ANS3402W psGetHostName(): A system call
               Failed to determine fully qualified hostname
               failed with code 6201.
    11/19/2015 01:34:53 ANS5250E An unexpected error was
               encountered.
               TSM function name : nsGetNasVolumeInfo():
               TSM function      : dmMapNfsVolume() failed to map
               remote volume for path: </nas-backup/vol/data>.
               TSM return code   : 6201
               TSM file          : snapcommon.cpp (437)
  
  Initial Impact:
  Medium
  
  Tivoli Storage Manager Versions Affected:
  All supported Tivoli Storage Manager clients.
  
  Additional Keywords:
  TSM SPECTRUM PROTECT SNAPDIFF SNAP DIFF CRASH ABORT ABEND
  MSGANS3402W MSGANS5250E PSGETHOSTNAME GETHOSTNAME NETAPP
  

Local fix

• Reducing the number of producer threads may change the timing
  enough (in some environments) to avoid the conditions required
  to experience this crash.  Add the "TESTFLAG MAXPRODUCERS:1"
  option to your dsm.opt and initiate a new incremental SnapDiff
  backup to determine if this avoids the crash.  Note that this
  option may induce increased backup times for this client.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * Tivoli Storage Manager for Client version 6.4 and 7.1        *
  * running on AIX and Linux x86_64                              *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * See ERROR DESCRIPTION                                        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * Apply fixing level when available. This problem is currently *
  * projected to be fixed in level 7.1.6.                        *
  * Note that until these levels are available, this information *
  * is subject to change at the discretion of IBM.               *
  ****************************************************************
  *
  

Problem conclusion

• The fix replaced non-reentrant system function calls with
  appropriate reentrant ones. Also, added the checking for NULL
  pointers in snapdiff related routines. As a result, the snapdiff
  operation crash has been fixed.
  

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

• dsmc
  

Fix information

• Fixed component name

  TSM CLIENT

• Fixed component ID

  5698ISMCL

Applicable component levels

• R71L PSY

     UP

• R71A PSY

     UP