IV66774: CPU UTILIZATION METRICS ARE INCORRECT DUE TO MISSING WMI IDLE TIME

IBM Tivoli Monitoring 6.3.0 Fix Pack 5 (6.3.0-TIV-ITM-FP0005)

APAR status

• Closed as program error.

Error description

• Problem:
  It has been reported that percentages of CPU time used by
  processors on Windows servers do not match the actual CPU
  utilization as reported by operating system tools. This depends
  on the possible lack of the processors' idle time information
  from WMI. With this APAR it will be possible to workaround this
  issue by restoring the same perfmon interface for these kind of
  metrics as used by pre 6.30 KNT agents.
  
  The perfmon style of data will be accessible only through an
  environment variable to be set. Current WMI interface remains
  the default.
  
  Affected Platforms / Versions:
  It looks like only 32 bit WMI DLL on Windows 2003 servers may
  be affected, so this issue is only with the 32 bit agent
  running on 32 or 64 bit servers.
  
  Diagnostics:
   KBB_RAS1: ERROR (UNIT:knt05agt ALL) (UNIT:cpu ALL). Look for:
  
   "getWMIrefreshed_data") Read Raw IdleTime: 0
  
  Example: TBD
  
  Initial Impact:
   High, the wrong CPU metrics may trigger false alerts on
   high cpu consumption.
  
  Additional Keywords:
   PRIVTME
   PROCTIME
   USERTME
   NTPROCSSR
  

Local fix

• Run the 64 bit version of the KNT agent on 64 bit servers.
  

Problem summary

• Overestimated percentages of CPU utilization from the Monitoring
   Agent for Windows OS.
  
  
  The Monitoring Agent for Windows OS can return incorrect values
  for attributes related to CPU percent utilization at system and
  processor level.  This is due to possible missing information
  about Idle Time of processors from the underlying WMI provider,
  when the agent runs in 32-bit mode on Windows 2003 servers.
  
  
  In order for this APAR to be properly implemented in your
  environment, a new environment variable has been added.  See the
  "Install Actions" section of the APAR conclusion for more
  details.
  

Problem conclusion

• Introduced new environment variable that allows to switch from
  the WMI interface to the perfmon interface when collecting CPU
  related metrics.  Default behavior is unchanged.
  
  Impacted attributes are NT_System\%_Total_Privileged_Time,
  NT_System\%_Total_Processor_Time,NT_System\%_Total_User_Time,
  NT_Processor\%_Privileged_Time, NT_Processor\%_Processor_Time,
  NT_Processor\%_User_Time,
  NT_Processor_Summary\High_%_Privileged_Time,
  NT_Processor_Summary\High_%_Processor_Time,
  NT_Processor_Summary\High_%_User_Time,
  NT_Processor_Summary\Low_%_Privileged_Time,
  NT_Processor_Summary\Low_%_Processor_Time,
  NT_Processor_Summary\Low_%_User_Time.
  
  Install Actions:
  Set environment variable: NT_SKIP_WMICPU=1
  
  allowed values:
  0=default, continue to collect the data using WMI
  1=collect the data using perfmon
  
  
  The fix for this APAR will be contained in the following
  maintenance packages:
  
  | FixPack    | 6.3.0-TIV-ITM-FP0005
  | InterimFix | 6.3.0.2-TIV-ITM_WIN-IF0003
  

Temporary fix

• Run 64-bit native Monitoring Agent for Windows OS on 64-bit
  servers.
  

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