IZ61329: ITMCMD AGENT START PC FAILS TO START THE AGENT.

APAR status

• Closed as program error.

Error description

• PROBLEM DESCRIPTION:
  itmcmd agent start PC fails to start the agent.
  For example
  itmcmd agent start px
  and got the error:
  Starting Premium Monitoring Agent for AIX ...
  KCIIN0195E Agent already started...
  
  Then run cinfo -r, we got
  hostmachine  px     286980  someuser   Sep     None
  ...running
  
  But actually, the Premium Agent is not running at all.
  
  The reason the itmcmd failed is that in
  $CANDLEHOME/config/.ConfigData/RunInfo,
  there is an entry for 286980 for px, like the following
  hostmachine|px|286980|None|aix523|
  
  RECREATE INSTRUCTIONS:
  The issue is hard to recreate, we can use the following steps to
  simulate the issue.
  - Install the OS Agent in UNIX
  - Start the OS Agent
  - End the OS Agent
  - Run a process that sleeps for a while and get its PID
  - Change the ux entry in RunInfo and use the PID from your sleep
  process
  - Run: itcmd agent start ux
    You will get the agent is already running
  - Run: "cinfo -r" and it will show the ux agent is running
  - Run: itmcmd agent -f stop ux
    It kills your process and it should not.
  
  
  
  LOCAL FIX:
  (1)Make sure the agent is not running on the system by using ps
  -ef |
  grep xxx command.
  (2)edit $CANDLEHOME/config/.ConfigData/RunInfo and remove the
  entry for
  the agent.
  (3)start the agent by issuing itmcmd agent start command.
  

Local fix

Problem summary

• When an agent exits abnormally and if the pid of this agent is
  reused by the OS,i.e the pid is assigned to another process, the
  problem will happen when IBM Tivoli Monitoring starts/stops the
  agent. Start/stop server has the same issue.
  
  When starting an agent, a similar message like the following is
  displayed
  
  Starting Monitoring Agent for Linux OS ...
  KCIIN0195E Agent already started...
  
  When stopping the agent, a similar message like the following is
  displayed
  
  Stopping Monitoring Agent for Linux OS ...
  KCIIN2525W The saved PID description
  $CANDLEHOME/li6263/lz/bin/klzagent] does not match the current
  running process description [vi] for PID [19152] for application
  product code [lz]. This means that unless you use the [-n] flag
  (CLI mode only), it will not be stopped. This is to prevent
  re-used PIDs from being killed and unintentionally stopping
  non-IBM Tivoli Monitoring processes.
  
  KCIIN0205E Unable to stop agent or process...
  Exceptions.UnabletoKillProcess Running and saved PID
  descriptions do not match. Not continuing with the stop command.
  

Problem conclusion

• Codes are added to deal with the reused pid. If it is a reused
  pid, the start/stop will continue and the entry of the pid is
  deleted from RunInfo and RunInfo_Description.
  
  
  The fix for this APAR is included in the following maintenance
  vehicle:
      | fix pack | 6.2.1-TIV-ITM-FP0002
      | fix pack | 6.2.2-TIV-ITM-FP0002
  

Temporary fix

• (1)Make sure the agent is not running on the system by using
   ps -ef |grep xxx
  command.
  (2)edit $CANDLEHOME/config/.ConfigData/RunInfo and remove the
  entry for the agent.
   for example
   hostmachine|ux|286980|None|aix523|
  (3)start the agent by issuing itmcmd agent start command.
  

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

  OMEG DIST INSTA

• Fixed component ID

  5608A41CI

Applicable component levels

• R621 PSY

     UP