Uploaded image for project: 'OpenShift Service Mesh'
  1. OpenShift Service Mesh
  2. OSSM-11637

ServiceMeshMember reconcile times 40s in large clusters

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Minor Minor
    • None
    • None
    • Documentation
    • False
    • Hide

      None

      Show
      None
    • False
    • Moderate

      Description of problem:

      when creating a new namespace + new ServiceMeshMember CRD in it : reconcile can take 40 seconds    

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

      OSSM_2.6.11-3 on Openshift 4.16.51_1592

      How reproducible:

          hard to reproduce. its a timing issue

      Steps to Reproduce:

          1. you need an openshift cluster with thousands of pods (3000-4000) in the istio mesh 
          2. need to set istio sidecar injection to `auto`
          3. need to create namespace + ServiceMeshMember in it and watch the diff between createTimeStamp and status: ready      

      Actual results:

          in most cases : the time to reconcile that ServiceMeshMember object is less than 1sec. in some case its 6s,12s or even 40s. 

      Expected results:

          sub second ready state for ServiceMeshMember. just like with classic K8s and istio 1.21 under same pod count.

      Additional info:

          what other info is required to debug this slowness ?

              rhn-support-ssiddhar Shreya Siddhartha
              ofek3 t taitelman
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated: