Skip to Content
avatar image
Former Member

Retries When a Jobfails

In our Scenario we are sending the Sales Order to an HTTP address (Receiver), using XI. This is a Synchronous Call from XI to the URL.

Sometimes the Receiver URL is down and due to that our Job fails immediately. What we wanted was, during these situations, we want to Try or wait for some stipulated times or period.

we have modified the one of the parameter in nSXMB_ADM and set the IS_RETRY_LIMIT & Set the current value to 35

Also using the transaction /smicm

Increased the time Limit for icm/conn_timeout

BUT it doesn't help us. Whenever the Receiver URL is Down our job fails instantaneously.

Any help is really appreciated.

Thanks friends.

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

3 Answers

  • Best Answer
    avatar image
    Former Member
    Nov 08, 2005 at 07:58 PM

    Hi Karen,

    I'm not sure if this is still the case, but I believe that it used to be impossible to do what you are asking. I think the way they do things in XI is that messages that are synchronous are also considered to be 'best effort', that is, if they fail they should not be retried. If a message, needs to be retried, I believe it needs to be asynchronously, and only then will the retry parameters you've specified take effect. Perhaps you could try switching it to test and see if this is the case.

    Cheers,

    Steve

    -


    If you find a post useful, please help keep the community going by setting a good example and rewarding the poster with points.

    Add comment
    10|10000 characters needed characters exceeded

  • Nov 09, 2005 at 05:08 PM

    Hi Karen,

    if your message would be asynchronous, everything would be easy by making some changes in transaction SMQR.

    If you really want a synchronous call (bad idea, i dont recommend that) build up a business process:

    1.) receive message from R/3

    2.) send message to third party

    3.) in case of error jump to a time controlled loop

    4.) where you again send that message and so on

    5.) you need a counter to make that not endless

    6.) you will have a bad performance (too synchron)

    7.) increase your WAS timeout parameters because of synchronous first process

    8.) in case of success send response back and finish process

    Regards,

    Udo

    Add comment
    10|10000 characters needed characters exceeded

    • Hi Karen,

      Deregister all queues. Then send a message, look in SMQ2 for the queue name. Register again all queues. Put only 2 attempsts for your queue and a short pause.

      Copy program RSXMB_RESTART_MESSAGES to your namespace an modify it. It has fields for start date and time, but you need a intervall (do not start older than). Only a little work for a good programmer. Schedule that programm as job, which is restarting your error messages.

      Regards,

      Udo

  • avatar image
    Former Member
    Nov 09, 2005 at 05:42 PM

    Hi Karen

    Check this OSS Note 813029

    Sameer

    Add comment
    10|10000 characters needed characters exceeded

    • Former Member

      Sameer I will request my basis team to apply this OSS note (they were telling me that this NOTE has been applied already) and we are investigating that right now, I will get back to you and to the forum soon.

      Thanks a lot Sameer.