Clovertech

Manually killing an orphaned Process

› Clovertech Forums › Read Only Archives › Cloverleaf › Cloverleaf › Manually killing an orphaned Process

  • Creator
    Topic
  • October 4, 2011 at 2:29 pm #52733
    Vince Angulo
    Participant

      We are CIS 5.7R1 on AIX 5.3.

      This morning we had an issue where a process was stopped from the Network Monitor GUI and could not be re-started.  One of our SysAdmins (who normally doesn’t work with Cloverleaf) identified a PID that was still running and killed it so the process could be restarted.  

      I’m trying to get a little smarter on how this works and get more self-sufficient…so pls forgive the wordiness!

      When I ps -ef | grep , I get two PIDs,

      — an ‘hciengine’ entry followed by the site, process and thread details, and

      — an ‘hcienginewatch’ perl script entry with the same details.

      Based on the entry names, I take it that the former is the actual interface and the latter controls the GUI display.

      When the process was stopped this morning, the GUI showed the process block red and the thread boxes as dead.  I take it to mean that the ‘hcienginewatch’ stopped successfully, but ‘hciengine’ was still running so it wouldn’t allow a restart.  There were no readily apparent errors we could identify.

      So to bring down the ‘hciengine’ PID, a kill PID was executed, then ps -ef again to make sure it’s gone and then the process started normally from the GUI.

      If all that sounds right, should there be any other ‘housekeeping’ steps to clean this up, or other things we can check that might lead us to a cause or prevent future problems? If I’m not on the right track, let me know that too.

      Thanks in advance!!

    • Creator
      Topic
    Viewing 2 reply threads
    • Author
      Replies
      • October 5, 2011 at 10:24 am #75317
        Robert Kersemakers
        Participant

          Hi Vince,

          First thing I would like to know is why the thread went dead. Check out the process log (.old for the previous log file) for more details on this. It should have some indication on what went wrong.

          You did the right thing to kill the processes; no further cleanup is necessary.

          Zuyderland Medisch Centrum; Heerlen/Sittard; The Netherlands

        • October 5, 2011 at 3:31 pm #75318
          Vince Angulo
          Participant

            Thanks Robert.

            The cause was in the log…a socket error to the target system — not surprising as we seem to have frequent issues with because their windows-based interface app doesn’t always shut down cleanly nor reliably release the socket.

            Apparently it hasn’t manifested itself at the same time as someone turning off the Cloverleaf process before — not sure why that would’ve left the engine process running, but hopefully we’ll figure that out in time.

          • October 8, 2011 at 2:28 pm #75319
            bill bearden
            Participant

              Vince,

              In the Windows world, there is a file called pid in the execprocesses directory. This file contains the process id. We always check these files when looking at the status of processes in task manager (equiv of ps -ef). We then know the CL process the hciengine pid represents.

              Good luck,

              Bill

          • Author
            Replies
          Viewing 2 reply threads
          • The forum ‘Cloverleaf’ is closed to new topics and replies.