IT23781: VM BACKUP CRASH

Download Information: Version 8.1.4.x Data Protection for VMware

APAR status

• Closed as program error.

Error description

• In rare cases, VM backup of multiple guests in parallel can
  crash when two different threads attempt to create a lock file
  at the same time.
  The dsmerror.log will display this error :
  
  ANS0361I DIAG: Overlapped IO Monitor thread, fatal error,
  signal 11
  
  IBM Spectrum Protect Versions Affected:
  IBM Spectrum Protect for Virtual Environments - Data protection
  for VMware 8.1.4 and higher on Linux and Windows
  
  Customer/L2 Diagnostics
  
  A service trace will display the following :
  
                  B/A Txn Consumer thread
  xx================>
  <timestamp> [PID] [TID_1] : psfilelock.cpp  (157): ENTER =====>
  psFileLock::createLockFile
  
                  B/A Txn Consumer thread
  yy================>
  <timestamp> [PID] [TID_2] : PasswordFile.cpp(491):
  PasswordFile::unlockPasswordFile(): Enter.
  
                  B/A Txn Consumer thread
  xx================>
  <timestamp> [PID] [TID_1] : psfilelock.cpp  (157): EXIT  <=====
  psFileLock::createLockFile
  
  We see here two threads (TID_1 and TID_2 in the above example)
  share pointers simultaneously when attempting to lock the
  password file and thus, one thread crashed after the pointer
  cleanup is done by the other thread. The stack trace cannot be
  used to identify this defect as the application might crash in
  different places due to timing and different memory access
  points and by the time the crash appears the memory can be
  corrupted and therefore not display useful information..
  
  Initial Impact: High
  
  Additional Keywords: tdp4ve tsm4ve crash lock password
  
  Products affected:
  IBM Spectrum Protect for Virtual Environments: Data Protection
  for VMware version 8.1.4 and higher on Microsoft Windows x64 and
  Linux x86_64 platforms.
  
  IBM Spectrum Protect for Virtual Environments: Data Protection
  for Microsoft Hyper-V version 8.1.4 and higher on Microsoft
  Windows x64 platforms.
  
  If you are using Data Protection for Microsoft Hyper-V version
  8.1.4 and higher refer to APAR IT24798.
  

Local fix

• Mitigate the problem by reducing or not using backup VM
  parallelism.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * Data Protection for VMware Data Mover version 8.1 running on *
  * all Microsoft Windows x64 and Linux x86 platforms.           *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * See ERROR DESCRIPTION                                        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * Apply fixing level when available. This problem is currently *
  * projected to be fixed in levels 8.1.4.1 and 8.1.6. Note that *
  * this is subject to change at the discretion of IBM.          *
  ****************************************************************
  

Problem conclusion

• User might experience program crash due to the fact that some
  functions for password file management are not thread-safe.
  Fix: make the involved functions thread-safe.
  

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:

  IT24798 IT25632

Modules/Macros

• dsmc
  

Fix information

• Fixed component name

  TSM FOR VE DP V

• Fixed component ID

  5725TVEVM

Applicable component levels

• R81L PSY

     UP

• R81W PSY

     UP