FTP / UPOC / timing issues

Clovertech Forums Read Only Archives Cloverleaf Cloverleaf FTP / UPOC / timing issues

  • Creator
    Topic
  • #48228
    Dan Loch
    Participant

      Hello,

      We are having a problem with FTP pickup of a pharmacy orders file on our engine box.  We are in the process of moving all of our interfaces from 3.8.1 to 5.3.  Our new configuration is dual WIN 2003 boxes running clustered.

      I have an outbound thread that is setup with a UPOC protocol running a file build TCL procedure.  This file is picked up by our AS400 via a script that is kicked off every 45 seconds.  The script resides and runs on the AS400 box.  The script renames the file on the engine, then copies in the renamed file for processing.  The original file is then deleted.

      This process worked perfectly on our old single server WIN 2000 box, but when we moved the interface to the new hardware and version of Cloverleaf, strange things started to happen.

      It seems that the script would pick up the first file OK, then on the second pull it would grab a copy of the same one it just processed, ignoring the newly created file on the engine.  Maybe the file was in a buffer or something and reprocessed the same data again.

      When checking the process on the engine, we can see that the engine thread was working perfectly and no duplicate data was being sent to the file.  We could also see by watching and refreshing our view of the engine directory that the file would be created, we could see the rename and then the deletion of the original.

      We changed the script kickoff to 90 seconds and it seemed to correct the problem, but we still felt uncomfortable with the change and moved the interface back to the old engine.

      On the AS400, the only change made was that the address of the new machine was changed from hard IP to DNS name.

      Could there be some timing issue here that would cause these strange things to occur?  If anyone has seen any issue like this, I would really appreciate a response, any suggestion would help.

    Viewing 2 reply threads
    • Author
      Replies
      • #58078
        Bill Bertera
        Participant

          Not sure I completely understand what you’re doing, but are you using the Cloverleaf FTP protocol at all or just a upoc? I know there is a bug with the FTP-timing introduced somewhere in late 3.8 or early version 5.

        • #58079
          Dan Loch
          Participant

            William,

            I am using UPOC protocol to write the file to disk on the engine server.

            The FTP is running on our AS400.  This previously worked OK on our old server, now that we changed to 5.3 and WIN 2003, we see this issue.

            Dan

          • #58080
            Dirk Engels
            Participant

              Dan,

              we had a similar behaviour at one of our customers. There we have Cloverleaf running on Tru64 Unix and we are FTP’ing to Windows.

              We saw, that when Cloverleaf has processed a file and delete it on the Windows machine, Windows is caching the disk accesses. Therefore the Tru64 got the file again in the next read cycle and then got an error when trying to process it.

              I’ve written a tcl script as a dir-parse procedure which checks if we have processes a specific filename in the last 270 seconds and if so the filename will be sorted out.

              It sound that you have the same problem.

              Dirk

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