Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-4754

Race condition between PTP events and AMQ router startup

XMLWordPrintable

    • CNF RAN Sprint 231
    • 1
    • False
    • Hide

      None

      Show
      None
    • Rel Note for Telco: Not Required (4.12) - this is the 4.11.z fix

      Description of problem:

      When PTP events are configured and pods are restarted (for example, during an SNO reboot), there is a race condition between the cloud-events-proxy container in the linuxptp-daemon pod and the AMQ router pod.
      
      If the cloud-events-proxy container is started first, it will fail to connect to the AMQ router and avoid sending any events at all. The only workaround would be to restart the container or linuxptp-daemon pod.
      
      We see the following in the logs:
      
      time="2022-11-02T11:49:44Z" level=info msg="Starting AMQP server"
      time="2022-11-02T11:49:59Z" level=error msg="error starting amqp at amqp://router-mesh.amq-router.svc.cluster.local error: amqp connection error amqp connection error"
      time="2022-11-02T11:49:59Z" level=warning msg="requires QPID router installed to function fully amqp connection error amqp connection error"
      
      And then, anytime an event should be sent:
      
      time="2022-11-02T11:49:59Z" level=warning msg="amqp disabled,no action taken: ...

      Version-Release number of selected component (if applicable):

      Seen at least on OCP 4.9.37 and 4.11.5

      How reproducible:

      Very often 

      Steps to Reproduce:

      1. Set up an SNO with PTP and AMQ Interconnect operators
      2. Configure PTP to enable the event publisher, using an AMQP router in the same node
      3. Reboot
      

      Actual results:

      PTP Events fails to connect to the AMQ router and never retries

      Expected results:

      It should retry the connection and connect to the AMQ router once it is available

      Additional info:

       

            jacding@redhat.com Jack Ding
            jpena@redhat.com Javier Pena
            Ofer Bochan Ofer Bochan
            Votes:
            1 Vote for this issue
            Watchers:
            7 Start watching this issue

              Created:
              Updated:
              Resolved: