Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-16889

CEO needs to handle optional MachineAPI

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Done-Errata
    • Icon: Critical Critical
    • 4.14.0
    • 4.14
    • Etcd
    • None
    • Important
    • No
    • 3
    • ETCD Sprint 240
    • 1
    • Approved
    • False
    • Hide

      machine API capability won't be able to release without this

      Show
      machine API capability won't be able to release without this

      Description of problem:

      Quoting Joel: In 4.14 there's been an effort to make Machine API optional, anything that that relies on the CRD needs to be able to detect that the CRD is not installed and then not error should that be the case. You should be able to use a discovery client to determine if the API group is installed or not

We have several controllers and informers that are depending on the machine API to be at least available to list and sync caches with. When the API is not installed at all the depending controllers are blocked forever and eventually get killed by the aliveness probe. That causes hot restart loops that cause installations to fail.

      https://redhat-internal.slack.com/archives/C027U68LP/p1690436286860899

       

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

      4.14

      How reproducible:

      always

      Steps to Reproduce:

      1. install a machineAPI=false cluster
2. ??? 
3. watch it fail

              tjungblu@redhat.com Thomas Jungblut
              tjungblu@redhat.com Thomas Jungblut
              Ge Liu Ge Liu
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

                Created:
                Updated:
                Resolved: