Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-55638

Avoid contention over bootimages between GitOps, etc. and the MCO

XMLWordPrintable

    • Quality / Stability / Reliability
    • False
    • Hide

      None

      Show
      None
    • 3
    • None
    • None
    • None
    • None
    • MCO Sprint 271
    • 1
    • Done
    • Release Note Not Required
    • N/A
    • None
    • None
    • None
    • None

      Description of problem:

      Release-repo MachineSet manifests have boot-image opinions, but starting in 4.19 (MCO-1361), the MCO will automatically manage those boot images on AWS and GCP (4.18 docs), and the MCO and GitOps like these release-repo opinions can wrestle over divergent opinions on what the boot-image should be.

      Version-Release number of selected component

      4.19

      How reproducible:

      100%, but unclear how many clusters will be impacted.

      Steps to Reproduce

      1. Install a 4.19 AWS or GCP cluster
      2. Patch the boot image for a MachineSet.
      3. See the MCO stomp the boot image back to its preference.
      4. Return to step 2.

      Actual results

      MCO keeps fighting, so GitOps or other actors may not realize that they keep getting stomped, leading to a warm loop, depending on how often they write.

      Expected results

      MCO can:

      • Warn folks on AWS/GCP about the change while they're on 4.18, via an admin-ack or other message.
      • Report surprising contention on 4.19, e.g. via alerting if some number of writes to the same MachineSet keep arguing over the same A/B boot-image options.
      • Probably additional options too...

              djoshy David Joshy
              trking W. Trevor King
              None
              None
              Sergio Regidor de la Rosa Sergio Regidor de la Rosa
              None
              Votes:
              0 Vote for this issue
              Watchers:
              8 Start watching this issue

                Created:
                Updated:
                Resolved: