Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-12773

Include statefulsets description and behavior during a node NotReady state

XMLWordPrintable

    • Moderate
    • No
    • 3
    • OSDOCS Sprint 237, OSDOCS Sprint 238
    • 2
    • False
    • Hide

      None

      Show
      None
    • N/A
    • Release Note Not Required

      Description of problem:

      Include statefulsets description and its behavior during a node NotReady state.
      As of now there are no much data available about the Statefulsets and its default behavior in the official OCP documentation.
      
      Business Impact:
      - CU recently had an issue with the nodes where the nodes went to NotReady state and the statefulsets didn't actually rescheduled. 
      - According to the Official Openshift documentation the pods should reschedule to another working node once the tolerationtime in seconds are finished according to the documentation here[0]
      - But there is some manual intervention required to get the pods evicted and to get rescheduled into other nodes in case of Statefulsets and itsan expected behavior.
      - We have already this[1] upstream doc and a KCS[2] which actually describes the same and are the fundamental behavior of the same.
      - But it would be great if we could add the same as a note in the OCP documentation[0] as it already includes the information for the daemon-sets pods behavior
      
      
      [0] https://docs.openshift.com/container-platform/4.12/nodes/scheduling/nodes-scheduler-taints-tolerations.html#nodes-scheduler-taints-tolerations-about-taintBasedEvictions_nodes-scheduler-taints-tolerations
      [1] https://kubernetes.io/docs/tasks/run-application/force-delete-stateful-set-pod/#delete-pods
      [2] https://access.redhat.com/solutions/4254721

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

      4.12

      How reproducible:

      - Run a statefulset deployment
      - Make a node notReady where one of the statefulset pod runs
      - The statefulset pod will not automatically reschedule to another node
      - We may need to manually remove the pod which is marked for deletion and then reschedule it

      Steps to Reproduce:

      1.
      2.
      3.
      

      Actual results:

      The STS pods are not rescheduled to a working node without manual interventions.

      Expected results:

      Need to add the same into the documentation that this is the expected behavior in case of STS pods! 

      Additional info:

      Reference Case: 03460040

       

            rhn-support-ahoffer Andrea Hoffer
            rhn-support-vpavithr Vishnudutt Pavithran (Inactive)
            Sunil Choudhary Sunil Choudhary
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Created:
              Updated: