Clovertech

Cloverleaf 6.1.1.0 threads showing as dead when connected

› Clovertech Forums › Read Only Archives › Cloverleaf › Cloverleaf › Cloverleaf 6.1.1.0 threads showing as dead when connected

  • Creator
    Topic
  • October 26, 2015 at 7:52 am #54859
    Elisha Gould
    Participant

      I’ve just started migrating our test sites over to 6.1 from 5.8 and am seeing some odd things with the connection status.

      The site starts ok and everything shows as up and connected. When a message is sent through, some of the threads start showing as “dead” in network monitor, however the thread is processing as expected.

      I’ve done a quick check with msiAttach and msiGetStatSample, with the threads that show as dead having an empty list, and the others populated.

      hcimsiutil -dd shows everything ok.

      Has anyone else seen this or know what to do to fix it?

    • Creator
      Topic
    Viewing 4 reply threads
    • Author
      Replies
      • October 26, 2015 at 10:28 am #83226
        James Cobane
        Participant

          Hi Elisha,

          We’ve been running 6.1.1 in production since August (on AIX) and have not seen this issue.  It sounds like this is something you should probably open an incident with Infor on.

          Jim Cobane

          Henry Ford Health

        • October 26, 2015 at 4:05 pm #83227
          Russ Ross
          Participant

            Just curious is the problem across the board or just for potentially problematic named threads like ones that start with a number like 123_thread or threads that have capital letters in them?

            Russ Ross
            RussRoss318@gmail.com

          • October 27, 2015 at 12:02 am #83228
            Elisha Gould
            Participant

              We are running on RHEL 5.10. All the procs/threads only contain a-z and ‘_’.

              There are presently 5.8 and 6.1 cloverleaf sites running (the migrated sites have been stopped in 5.8 and started in 6.1)

              It seems to be the same threads that have the issue when it occurs, but there is almost no difference between the ones that show ok and the ones that stop showing status.

              I have several sites that are almost identical, with only the thread names/ports differing, and different threads show up as dead.

              It appears to be half shows up and half show dead.

              I’ll try turning on debug to see if it shows anything.

            • October 27, 2015 at 12:54 am #83229
              Elisha Gould
              Participant

                I think I may have figured it out.

                There is a script that runs as a cron job that executes a tcl script which can set the environment using hcisetenv and calls msiAttach/msiGetStatSample.

                This was using the 5.8 version of hcitcl, which apparently is incompatible with 6.1 for these calls.

                The result must have updated the shared memory, causing the issue.

                I fixed the script and it appears ok now.

              • October 27, 2015 at 1:27 pm #83230
                Russ Ross
                Participant

                  The way we get around using the older version of hcitl when doing a cloverleaf upgrade is to have all our TCL stand alone scripts start with this

                  as the first line:

                  Code:

                  #!/usr/bin/env hcitcl

                  some people and some of our scripts also use this

                  Code:

                  #!/usr/bin/env tcl

                  the outcome seems to be successful either way.

                  Russ Ross
                  RussRoss318@gmail.com

              • Author
                Replies
              Viewing 4 reply threads
              • The forum ‘Cloverleaf’ is closed to new topics and replies.