Uploaded image for project: 'OpenShift GitOps'
  1. OpenShift GitOps
  2. GITOPS-5975

Openshift GitOps Redis HA pods in CrashLoopBackOff after upgrade to 1.14.2

XMLWordPrintable

    • 5
    • False
    • Hide

      None

      Show
      None
    • False
    • Hide
      Release Notes: Bug Fix
      Before this update, an upgrade issue caused the argocd-redis-ha-configmap and argocd-redis-ha-health-configmap and the Redis HA statefulset to remain unchanged, leading to Redis HA pods encountering an AUTH error.

      This update fixes the issue by ensuring the OpenShift GitOps Operator correctly updates the Redis HA configmaps and the Redis HA statefulset during an upgrade.

      Now, the Redis HA configmaps and the Redis HA statefulset are properly updated after an upgrade, preventing Redis HA pods from entering an AUTH error state.
      Show
      Release Notes: Bug Fix Before this update, an upgrade issue caused the argocd-redis-ha-configmap and argocd-redis-ha-health-configmap and the Redis HA statefulset to remain unchanged, leading to Redis HA pods encountering an AUTH error. This update fixes the issue by ensuring the OpenShift GitOps Operator correctly updates the Redis HA configmaps and the Redis HA statefulset during an upgrade. Now, the Redis HA configmaps and the Redis HA statefulset are properly updated after an upgrade, preventing Redis HA pods from entering an AUTH error state.
    • Bug Fix
    • GitOps Crimson - Sprint 3267, GitOps Crimson - Sprint 3268

      Description of Problem

       Openshift GitOps Redis HA pods in CrashLoopBackOff after upgrade to 1.14.2

      Additional Info

      PFA the GitOps must-gather

      Problem Reproduction

      1. On an OCP 4.16.4 cluster, install v 1.12.0 of Openshift GitOps operator

      2. Create a new namespace and create an ArgoCD inside it, with ha enabled (.spec.ha.enabled = true) and wait for the ArgoCD to become available

      3. Capture the state of these ConfigMaps for later comparison: 

        - argocd-redis-ha-configmap

        - argocd-redis-ha-health-configmap

      4. Upgrade Openshift GitOps operator to 1.14.0

      5. Observe no differences to config maps:

        - argocd-redis-ha-configmap

        - argocd-redis-ha-health-configmap

      6. Upgrade Openshift GitOps operator to 1.14.1

      7. Observe no differences to config maps:

        - argocd-redis-ha-configmap

        - argocd-redis-ha-health-configmap

      8. Upgrade Openshift GitOps operator to 1.14.2

      9. Observe no differences to config maps:

        - argocd-redis-ha-configmap

        - argocd-redis-ha-health-configmap

      10. Delete the argocd-redis-ha-configmap and argocd-redis-ha-health-configmap config maps

      11. Config maps are recreated, with the changes from[ https://github.com/redhat-developer/gitops-operator/pull/719/files#diff-022af0073691c01f617a482fbdfec043af07f9a966099d2e93907c54a3503e68|https://github.com/redhat-developer/gitops-operator/pull/719/files#diff-022af0073691c01f617a482fbdfec043af07f9a966099d2e93907c54a3503e68]

      12. The argocd-redis-ha-server pods show the AUTH error

      13. In the ArgoCD CR the .spec.ha.enabled is set to false, and then set to true again

      14. The argocd-redis-ha-server pods are recreated and all is ok

      Reproducibility

      Always

      Prerequisites/Environment

      OCP 4.16.2

      Expected Results

      The redis-ha confimaps should be updated by the operator after upgrading to 1.14.2.   

      Actual Results

      The configmaps do not get updated with the changes introduced in GitOps 1.14.2 version and on manual changes, the redis-ha-server pods go into pending state. 

      Workaround (If Possible)

      Disable HA in ArgoCD CR and enable it again. 

       

              rh-ee-mmeetei Mangaal Meetei
              rhn-support-disharma Diksha Sharma
              Votes:
              1 Vote for this issue
              Watchers:
              5 Start watching this issue

                Created:
                Updated:
                Resolved: