MQ

MQ

Join this online group to communicate across IBM product users and experts by sharing advice and best practices with peers and staying up to date regarding product enhancements.

 View Only

  • 1.  MQ/MFT got a MQ reason code of 2102 saying it a connection could not be esablished to the queue manager

    Posted Mon May 10, 2021 12:05 PM
    We are running MQ 9.1 on z/OS 2.3 and MQ/MFT 9.1 also.  This past weekend, we had a job fail, when the MQ/MFT agent was trying to connect to the queue manager, on that system.  The MQ completion code was 2 and reason code wass 2102 with a description of "A connection could not be established to queue manager QP01" .

    I can't find any error messages that tell me what exactly the problem was.

    Any help would be appreciated.

    Thanks,
    Peter

    ------------------------------
    Peter Vander Woude
    ------------------------------


  • 2.  RE: MQ/MFT got a MQ reason code of 2102 saying it a connection could not be esablished to the queue manager

    Posted Mon May 10, 2021 12:22 PM
    Hi,

    Take a look at https://www.ibm.com/docs/en/ibm-mq/9.2?topic=codes-2102-0836-rc2102-mqrc-resource-problem
    Are shared queues being used here? Are there any messages in the queue manager job log that could help?

    It could also be a JVM thread scheduling delay. Try setting the com.ibm.mq.jmqi.threadWaitTimeout JVM system property to a value higher than 20000 (20 seconds in milliseconds) which is the default. This page describes how to set system properties for MFT: https://www.ibm.com/docs/en/ibm-mq/9.0?topic=reference-java-system-properties-mft

    So something like export BFG_JVM_PROPERTIES="-Dcom.ibm.mq.jmqi.threadWaitTimeout=60000" sets the scheduling timeout to 60 seconds. Assuming that's what the problem was of course.

    Regards, Matt.
    Matt Leming
    STSM, IBM MQ for z/OS

    Email: lemingma@uk.ibm.com
     
     





  • 3.  RE: MQ/MFT got a MQ reason code of 2102 saying it a connection could not be esablished to the queue manager

    Posted Mon May 10, 2021 12:40 PM
    Matthew,

    No, we are not using shared queues.  This is a standalone queue manager (i.e. not in a cluster).  There are no messages in the joblog of the queue manager master or channel initiator started tasks that give any indication of what was going on.

    I checked to see if that lpar was busy at the time, and it does not appear to have been anywhere near to being 100% busy.

    The only other thing was that the mq/mft transfer is using fteant, and the start of the process is to ping the remote agent, to make sure it's alive and responding, and the ping command does time out, but no where near the 180 seconds that is configured for the ping to wait... it actually fails a little over 1 minute after it started the ping request.

    Peter

    ------------------------------
    Peter Vander Woude
    ------------------------------

    Original Message


  • 4.  RE: MQ/MFT got a MQ reason code of 2102 saying it a connection could not be esablished to the queue manager

    Posted Wed May 12, 2021 04:07 AM
    Hi Peter,

    If this is a one off it's likely to be quite hard to resolve as that is a pretty generic error message. When I have seen these in the past, if the shared queue side of things is ruled out, it has been a thread scheduling delay in the JVM and the property I mentioned has been able to solve the problem.
    The fact that you said the command fails after a minute implies that thread scheduling, could potentially be an issue, maybe.

    Regards, Matt.
    Matt Leming
    STSM, IBM MQ for z/OS

    Email: lemingma@uk.ibm.com
     
     



    Original Message


  • 5.  RE: MQ/MFT got a MQ reason code of 2102 saying it a connection could not be esablished to the queue manager

    Posted Wed May 12, 2021 09:53 AM
    Edited by Peter Vander Woude Wed May 12, 2021 09:55 AM
    Matt,

    In the MQ/MFT agent, we have that setting set to 0.  I know that there was an apar for mq/fte 7.0 where even if the user had this set to 0, it was using the 20 second default, and saw in the apar that the issue did not apply to 8.5.  I am guessing that it does not apply to the 9.1 version as well.

    In the profile that get's used by the batch jobs, we do not have that setting at all.  Is it needed in the batch jobs that execute the fte commands?

    Peter

    ------------------------------
    Peter Vander Woude
    ------------------------------

    Original Message


Related Content