Uploaded image for project: 'OpenShift Etcd'
  1. OpenShift Etcd
  2. ETCD-201 Ensure a safety net for the 3.4 to 3.5 etcd upgrade
  3. ETCD-226

Ensure cluster restore after upgrade does not trigger new upgrade

XMLWordPrintable

    • Icon: Sub-task Sub-task
    • Resolution: Done
    • Icon: Blocker Blocker
    • None
    • None
    • None
    • False
    • False
    • Undefined

      On restore, the CVO and etcd-operator pods should be killed (possibly by scaling the CVO deployment to 0 and scaling the etcd operator deployment to zero) before restoring etcd state.  Once the Kube API-server is up on the restored etcd, folks should us 'oc adm upgrade --clear' to clear out desiredUpdate before they launch the CVO pod and etcd operator.  This will prevent the CVO from starting in on the same update again, and will also prevent the etcd operator from trying to take a new snapshot that dups the one that was just used for the restore.

              sbatsche@redhat.com Sam Batschelet (Inactive)
              mnewby@redhat.com Maru Newby (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: