Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-26071

connection to iSCSI boot volume lost when ovs configure script is triggered

    XMLWordPrintable

Details

    • Bug
    • Resolution: Not a Bug
    • Normal
    • None
    • 4.15.z
    • None
    • Important
    • No
    • SDN Sprint 248
    • 1
    • Rejected
    • Hide

      None

      Show
      None

    Description

      .SeDescription of problem:

      When installing OCP on a machine with an iSCSI boot volume, the ovs configure script does some network configuration that disconnect the boot volume. The machine becomes unusable after the volume is lost.

      Tested on Oracle Cloud as it's the main use case for this feature at the moment.
       

      Version-Release number of selected component (if applicable): iSCSI boot volume is available in RHCOS from OCP 4.15

       

      How reproducible: Always

       

      Steps to Reproduce:

      1. Install OCP on machines with an iSCSI boot volume

      2. Boot RHCOS with rd.iscsi.firmware=1 kernel argument

      I used the assisted installer to install OCP on Oracle Cloud (you have to ensure to use an iSCSI boot volume when importing the custom OS image): https://access.redhat.com/articles/7039183

      The iSCSI feature is not yet available in assisted-installer on console.rh.com, so if you decide to follow this path, deploy you own assisted-installer using podman: https://github.com/openshift/assisted-service/tree/master/deploy/podman

      Alternatively, you can also trigger a CI job from the assisted-service repo: https://github.com/openshift/assisted-service/pull/5804#issuecomment-1877163076

      Actual results:
      Connection to iSCSI boot volume is lost when configure_ovs.sh script is triggered. The connection to the boot volume is not recoverable.

      Expected results:

      Connection to iSCSI boot volume is not lost / is recoverable.

      Additional info:

      The machines used for my tests are Oracle Cloud VMs with iSCSI boot volumes, they have only one network interface.

      I tried to configure OVN with routingViaHost: true and routingViaHost: false (default), but it did not help.

      It's hard for me to get a sosreport as the machine gets un-usable after the boot volume is lost.

      Console output routingViaHost=false: https://drive.google.com/file/d/1-GjO0t3rjWj5LTVdxe7ykt1BkQOWbr0i/view?usp=drive_link
      Console output routingViaHost=true: https://drive.google.com/file/d/1VFVNOa2ievUxTZlY7hDvC65Q_KajF3aX/view?usp=drive_link
      journalctl: https://drive.google.com/file/d/1gwaVs5nz5-qtSWhDpjEHZNSkgZt7txCE/view?usp=drive_link

      Thread: https://redhat-internal.slack.com/archives/C05UQ2Y7HKR/p1700669629478229

      Attachments

        Issue Links

          Activity

            People

              jcaamano@redhat.com Jaime Caamaño Ruiz
              agentil@redhat.com Adrien Gentil
              Anurag Saxena Anurag Saxena
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: