IV69221: WINDOWS OS AGENT UNRESPONSIVE AND CONSUMING HIGH CPU IN CASE OF WMI ERRORS

IBM Tivoli Monitoring 6.3.0 Fix Pack 5 (6.3.0-TIV-ITM-FP0005)
IBM Tivoli Monitoring: Windows(R) OS Agent 6.3.0.4-TIV-ITM_WIN-IF0002
IBM Tivoli Monitoring: Windows(R) OS Agent 6.3.0.2-TIV-ITM_WIN-IF0004
IBM Tivoli Monitoring: Windows(R) OS Agent 6.3.0.4-TIV-ITM_WIN-IF0003

APAR status

• Closed as program error.

Error description

• APAR SEV: 2
  COMPID: 5724C040W
  REL: 6.30.03.00
  Symptom Code: LP
  Symptom Key Word: LOOP
  
  
  Abstract:
   Windows OS agent unresponsive and consuming high CPU in case of
   WMI errors
  
  Problem:
   While monitoring processes' CPU consumption through WMI
   performance classes and their related interfaces an error can
   be returned to the agent leading to a hard loop in one of its
   threads, and to a general deadlock of the data collection.
  
  Affected Platforms / Versions:
   This issue affects the Windows OS agent 6.30 and following
   fixpacks.
  
  Diagnostics:
   set (UNIT:proc ALL) (UNIT:cpu ALL) (UNIT:knt ALL) in KBB_RAS1
   and look for messages  like:
  
   procusage.cpp,100,"sample") Cannot get Process performance data
   from WMI (Error: 0x80070102)
   cpustats.cpp,378,"getWMIrefreshed_data") GetObjects failed
  
   or for an Entry in the "update" method never followed by an
   Exit:
     processstatisticstemplate.cpp,54,"update") Entry
  
  Initial Impact:
   High, monitoring no longer occurring and other applications
   impacted
  
  Additional Keywords:
   NTPROCESS
   NTPROCSSR
   PCTPRCSTME
  

Local fix

• Possible workaround is to stop any situations and historical
  data collections on the NTPROCESS, NTPROCRSUM, WTPROCESS and
  NTPROCSSR attribute groups (and avoid to query them e.g.
  refreshing TEP workspaces under Process and Processors).
  
  Also, the two background threads interfacing with WMI need to be
  disabled by setting in KNTENV:
    KNT_PROCESS_SAMPLE_SECS=0
    KNT_CPUSTAT_SAMPLE_SECS=0
  

Problem summary

• The Monitoring Agent for Windows OS does not properly handle an
  error condition that may be returned from WMI calls used for
  monitoring processes' CPU consumption.  As a result of this
  error the agent can start consuming high CPU and no longer
  respond to incoming queries.
  

Problem conclusion

• Code fixed to properly detect and manage such WMI errors.
  
  The fix for this APAR will be contained in the following
  maintenance packages:
  
  | FixPack    | 6.3.0-TIV-ITM-FP0005
  

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

  ITM AGENT WINDO

• Fixed component ID

  5724C040W

Applicable component levels

• R630 PSY

     UP

• R610 PSN

     UP

• R620 PSN

     UP

• R621 PSN

     UP

• R622 PSN

     UP

• R623 PSN

     UP