Uploaded image for project: 'Fast Datapath Product'
  1. Fast Datapath Product
  2. FDP-1452

"ovn-controller" is not detecting a created VM port during a OpenStack live migration

    • Icon: Bug Bug
    • Resolution: Duplicate
    • Icon: Critical Critical
    • None
    • None
    • ovn24.03
    • None
    • False
    • Hide

      None

      Show
      None
    • False
    • rhel-9
    • None
    • rhel-net-ovn
    • ssg_networking

      Problem Description: During an OpenStack live migration, the Nova compute agent creates the port in the destination host. This port is detected by openvswitch but not by ovn-controller. The customer has around 60 compute nodes. Each one has 2TB RAM. The destination host has around 200 ports/VMs (1 port per VM, more or less).

       

      Impact Assessment: this is happening very often. The customer needs to evacuate some compute nodes and most of the live migrations fail.

       

      Software Versions:

      • OSP 17.1
      • OVN 24.03.4-20.33.0-72.6
      • RHEL 9.2

       

      Issue Type: it doesn't look like a regression but a specific error in this customer. This is happening with new compute nodes added to this environment. The container versions used are the same as in the older nodes. The openvswitch RPM (baremetal installation) is the same too.

       

      Reproducibility: very often. There are plenty of logs in customer case 04156954. The best ones, so far, are from Jun 5, uploaded between 20:00 and 21:00. These ones have the OVN DBs, the compute node sos_reports and the Neutron API logs.

       

      Reproduction Steps: live migrate a VM with one single port.

       

      RHOSPPRIO ticket: https://issues.redhat.com/browse/RHOSPPRIO-622

      Neutron ticket: https://issues.redhat.com/browse/OSPRH-17144

              amusil@redhat.com Ales Musil
              rodolfo_alonso Rodolfo Alonso
              Votes:
              0 Vote for this issue
              Watchers:
              4 Start watching this issue

                Created:
                Updated:
                Resolved: