Uploaded image for project: 'OpenShift Pipelines'
  1. OpenShift Pipelines
  2. SRVKP-3731

tekton-pipelines-webhook HorizontalPodAutoscaler should be set to minReplicas:2 to prevent disruption when pods are being restarted

    XMLWordPrintable

Details

    • 3
    • False
    • None
    • False
    • Hide
      - Added support to update/create HorizontalPodAutoscaler via `options`.
      - Added support for adding labels and annotations on Deployments and StatefulSets podTemplate via `options`.
      Show
      - Added support to update/create HorizontalPodAutoscaler via `options`. - Added support for adding labels and annotations on Deployments and StatefulSets podTemplate via `options`.
    • Pipelines Sprint 255, Pipelines Sprint 256, Pipelines Sprint 257, Pipelines Sprint 258
    • Important

    Description

      Description of problem:

      According to Availability, webhook should be made highly available to improve reliability and overall performance. Currently HorizontalPodAutoscaler configures tekton-pipelines-webhook with minReplicas:1 which can cause disruption when the specific pod is being restarted and thus trigger undesired problems for PipelinesRuns.

      In order to prevent issues, it's requested to have HorizontalPodAutoscaler for tekton-pipelines-webhook configured with minReplicas:2 by default to achieve fault tolerance. If that is not possible, it should be possible to configure minReplicas in HorizontalPodAutoscaler for tekton-pipelines-webhook so customers can set it to desired value.

      Workaround

      N/A

      Prerequisites (if any, like setup, operators/versions):

      Install OpenShift Pipelines (any version) on OpenShift Container Platform 4 and check HorizontalPodAutoscaler for tekton-pipelines-webhook

      Steps to Reproduce

      Install OpenShift Pipelines (any version) on OpenShift Container Platform 4 and check HorizontalPodAutoscaler for tekton-pipelines-webhook
       

      Actual results:

      When HorizontalPodAutoscaler for tekton-pipelines-webhook does not trigger scaling of tekton-pipelines-webhook and the associated pod is restarted, errors may occur during PipelinesRuns and thus impact overall customer experience.

      Expected results:

      As per Availability, webhook should be made highly available by default to prevent disruption when maintenance on the respective pods or infrastructure is being done.

      Reproducibility (Always/Intermittent/Only Once):

      Always

      Acceptance criteria: 

      Either have HorizontalPodAutoscaler for tekton-pipelines-webhook configured with minReplicas:2 or allow to configure it, if the customer would like to achieve recommended practice.
       

      Definition of Done:

      Build Details:

      N/A

      Additional info (Such as Logs, Screenshots, etc):

      $ oc get csv openshift-pipelines-operator-rh.v1.12.2
      NAME                                      DISPLAY                       VERSION   REPLACES                                  PHASE
      openshift-pipelines-operator-rh.v1.12.2   Red Hat OpenShift Pipelines   1.12.2    openshift-pipelines-operator-rh.v1.12.1   Succeeded
      
      $ oc get hpa tekton-pipelines-webhook -o json | jq '.spec.minReplicas'
      1
      

      Attachments

        Issue Links

          Activity

            People

              jkandasa-rh Jeeva Kandasamy
              rhn-support-sreber Simon Reber
              Shubham Minglani Shubham Minglani
              Votes:
              0 Vote for this issue
              Watchers:
              4 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: