PH41075: ADD OPTION TO TERMINATE ALL THE OTHER PROCESSES IF THE PARENT PROCESS CRASHES

APAR status

• Closed as program error.

Error description

• If the parent process abnormally terminates for some reason, the
  parent process stops and started task associated with it will
  end.   But the child processes don't also terminate.  A new
  directive will be added to IHS to terminate the child processes
  if the parent process ends abnormally.
  
  An example of the parent process abnormally ending is seen in
  the joblog with:
  
  14.23.21 STC12345  IEF404I WEBSRVR1 - ENDED - TIME=14.23.21
  14.23.21 STC12345  $HASP395 WEBSRVR1 ENDED - RC=0131
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:  All users of IBM HTTP Server on z/OS        *
  ****************************************************************
  * PROBLEM DESCRIPTION: When the IHS parent process crashes,    *
  *                      the                                     *
  *                      started task ends but other child       *
  *                      processes are not automatically         *
  *                      terminated                              *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  While processing a fatal signal (SIGABRT, SIGSEGV) the parent
  process leaves other processes running. Normally these processes
  will continue to work.
  If rotatelogs is in use and subsequently exits or crashes, the
  remaining processes are likely to eventually hang.
  Both conditions are rare.
  

Problem conclusion

• 1. The directive "IHSTerminateChildrenOnParentCrash" was added
  with a default value of OFF. When set to "ON", a crash in the
  parent process will send SIGTERM to all children asking that
  they gracefully exit.
  
  2. When an assertion is triggered in the server, the assertion
  text will be sent to the operator console with prefix IHS00023.
  Assertions are one source of SIGABRT signals.
  
  3. 9.0 only: Some errors, such as maintaining the symlink with
  the -L argument, no longer cause rotatelogs process to exit. The
  error paths changed are ones where the replacement rotatelogs
  process is not expected to help the situation.  In the event the
  parent process has crashed, rotatelogs processes cannot be
  replaced and will lead to a hang.
  
  The fix for this APAR is targeted for inclusion in IBM HTTP
  Server fix packs 8.5.5.21 and 9.0.5.11. For more information,
  see 'Recommended Updates for WebSphere Application Server':
  https://www.ibm.com/support/pages/node/715553
  

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

  WAS IHS ZOS

• Fixed component ID

  5655I3510

Applicable component levels