IV73936: CPU PERCENTAGES SUDDENLY WRONGLY REPORTED AS 0% OR 100%

IBM Tivoli Monitoring: Linux(R) OS Agent 6.2.3.3-TIV-ITM_LINUX-IF0005
IBM Tivoli Monitoring 6.3.0 Fix Pack 6 (6.3.0-TIV-ITM-FP0006)
IBM Tivoli Monitoring: Linux(R) OS Agent 6.2.3.5-TIV-ITM_LINUX-IF0003
IBM Tivoli Monitoring: Linux(R) OS Agent 6.2.3.5-TIV-ITM_LINUX-IF0004

APAR status

• Closed as program error.

Error description

• Problem:
  Attributes in the Linux CPU group may sometimes be reported
  incorrectly
   due to an unexpected modification of the sampling period. Since
   percentages are computed based on the increase in processors'
  ticks
   over the last two samples, if the last sample is too close in
  time to
   the previous one it might occur that no increase, or very
  small, is
   detected, thus leading to incorrect values. Even if the time
  to elapse
   is controlled by the Linux operating system, the agent does not
   implement any checks that it has actually elapsed.
  
  Affected Platforms / Versions:
   This issue can affect all the Linux OS agent versions, but
  it's been
   reported only on zLinux so far.
  
  Diagnostics:
   Have (UNIT:cpustats ALL) defined in KBB_RAS1 settings. Then
  look for
   the sequence of "Entry" into the "updateCpuStatistics" method.
  Under
   normal conditions they must be printed every 30 seconds, like:
  
  (Tue May 26 03:13:26
  2015.0078-2:cpustats.cpp,74,"updateCpuStatistics")
  Entry
  ...
  (Tue May 26 03:13:56
  2015.0000-2:cpustats.cpp,74,"updateCpuStatistics")
  Entry
  ...
  (Tue May 26 03:14:26
  2015.0000-2:cpustats.cpp,74,"updateCpuStatistics")
  Entry
  
   When this problem raises you can find multiple trace points on
  the same
   second or very close, like
  
  (Tue Jun  2 12:03:28
  2015.0000-2:cpustats.cpp,74,"updateCpuStatistics")
  Entry
  ...
  (Tue Jun  2 12:03:28
  2015.0006-2:cpustats.cpp,74,"updateCpuStatistics")
  Entry
  ...
  (Tue Jun  2 12:03:28
  2015.000C-2:cpustats.cpp,74,"updateCpuStatistics")
  Entry
  
  Initial Impact:
   Medium, as the issue seems to be temporary and abnormal CPU
  metrics are
   reported only on a very short timeframe.
  
  Additional Keywords:
   KLZCPU, USRCPU, SYSCPU, BUSYCPU
  
  Local Fix:
   None
  ----------------------------------------------------------------
  --------
  

Local fix

Problem summary

• Incorrect processors' CPU percentages in the Linux CPU attribute
   group.
  
  
  The Monitoring Agent for Linux OS may temporarily incorrectly
  report 0% or 100% for all of the CPU percent attributes in the
  Linux CPU attribute group, like User CPU (Percent), System CPU
  (Percent) or Busy CPU (Percent).  This is the effect of a
  failure in waiting for the expected time between two consecutive
  samples.  This behavior is not permanent and normal values are
  restored within few minutes.
  

Problem conclusion

• Code fixed to check the actual elapsed time between samples and
  retry waiting if needed.
  
  The fix for this APAR is contained in the following maintenance
  
  
  | FixPack    | 6.3.0-TIV-ITM-FP0006
  | InterimFix | 6.2.3.3-TIV-ITM_LINUX-IF0005
  

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 LINUX

• Fixed component ID

  5724C04LN

Applicable component levels

• R623 PSY

     UP

• R630 PSY

     UP

• R610 PSN

     UP

• R620 PSN

     UP

• R621 PSN

     UP

• R622 PSN

     UP