Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-46380

Static pod operator API accepts invalid node statuses and node status transitions

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done-Errata
    • Icon: Critical Critical
    • 4.19.0
    • 4.18.0, 4.19.0
    • kube-apiserver
    • Quality / Stability / Reliability
    • False
    • Hide

      None

      Show
      None
    • None
    • None
    • None
    • Proposed
    • None
    • Done
    • Bug Fix
    • Hide
      * Previously, the Static Pod Operator API allowed invalid node statuses with unset `currentRevision` and multiple nonzero `targetRevision` entries, which led to failures in node and installer controllers. With this release, new validation rules were added to enforce correct revision fields to ensure stable and consistent static pod status handling. (link:https://issues.redhat.com/browse/OCPBUGS-46380[OCPBUGS-46380])
      Show
      * Previously, the Static Pod Operator API allowed invalid node statuses with unset `currentRevision` and multiple nonzero `targetRevision` entries, which led to failures in node and installer controllers. With this release, new validation rules were added to enforce correct revision fields to ensure stable and consistent static pod status handling. (link: https://issues.redhat.com/browse/OCPBUGS-46380 [ OCPBUGS-46380 ])
    • None
    • None
    • None
    • None

      Description of problem:

      The StaticPodOperatorStatus API validations permit:
- nodeStatuses[].currentRevision can be cleared and can decrease
- more than one entry in nodeStatuses can have a targetRevision > 0
But both of these signal a bug in one or more of the static pod controllers that write to them.

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

      This has been the case ~forever but we are aware of bugs in 4.18+ that are resulting in controllers trying to make these invalid writes. We also have more expressive validation mechanisms today that make it possible to plug the holes.

      How reproducible:

      Steps to Reproduce:

          1.
    2.
    3.

      Actual results:

       The bug in 4.18+ that are resulting in some static pod node/installer controllers trying to make these invalid write requests.

      Expected results:

      To add some validation rules to help to see them

      Additional info:

              bluddy Ben Luddy
              bluddy Ben Luddy
              None
              None
              Ke Wang Ke Wang
              None
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

                Created:
                Updated:
                Resolved: