Timeout handling – avoid duplication

Clovertech Forums Read Only Archives Cloverleaf Cloverleaf Timeout handling – avoid duplication

  • Creator
    Topic
  • #52652
    sylaja suseela
    Participant

      Hi,

      We have a thread configured as below. Can someone kindly answer my following query.

      OutBound Data –

      Retries = -1

      Interval = 10

      Inbound Replies –

      Await Replies = checked

      TimeOut=30

      TimeOut Handling = Resnd OB message

      TPS Inbound Reply = cl_check_ack

      This is resulting in duplicate messages at receiving end when time out occurs.

      How I can change the above configuration to stop re-sending (when end system times out) and receive ACK’s when all okay?

      What is the effect if I change Retries = 0 and leave all other values as it is?

      Many thanks,

      Sylaja

    Viewing 5 reply threads
    • Author
      Replies
      • #74998
        Levy Lazarre
        Participant

          Hi Sylaja,

          I had a similar issue with a charge interface. The first thing you should try is increase your TimeOut setting to 45 or even 60. This will not slow down your interface; it only makes you wait longer before timing out and gives you more flexibility in case an occasional ACK is delayed, before Cloverleaf attempts to resend.

          In my opinion, if you don’t get an ACK back after a TimeOut of 45-60, there must a problem on the receiving end, and they should investigate (assuming you have no network issue).

          I hope this helps.

        • #74999
          Jim Kosloskey
          Participant

            Another option if the receiving system cannot tolerate duplicate messages is to set the Timeout to -1 (the default) to wait forever.

            Then you will need to adjust your alerts to let you know when NO mesages are flowing (essentially a deep queue depth or time since last sent or a combination).

            Yet another option is using Tcl keep track of how many times the same message is resent and set the threshold to one – although you must decide what to do if the threshold is reached.

            email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

          • #75000
            sylaja suseela
            Participant

              Thanks Jim,Levy,

              I think I would go with Jim’s last option – tcl to track resent count and put the message into error db.

              Regards,

              Sylaja

            • #75001
              Jim Kosloskey
              Participant

                OK but remember to put something in to control a runaway situation (like when the system does not respond at all) – I don’t think you would wnat a lot of messages erroring out because the system is just not reponding.

                email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

              • #75002
                sylaja suseela
                Participant

                  Good point Jim. Thanks.

                  What is the effect/harm of setting OutBount Data -> Retries=0 in this situation?

                  Regards,

                  Sylaja

                • #75003
                  Jim Kosloskey
                  Participant

                    As far as I know those settings are for the protocol level handling not the application replies and so would not come into play for what you are experiencing. I think leaving those as defaulted is a good strategy.

                    email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

                Viewing 5 reply threads
                • The forum ‘Cloverleaf’ is closed to new topics and replies.