Uploaded image for project: 'OpenShift Service Mesh'
  1. OpenShift Service Mesh
  2. OSSM-39

Default ServiceMeshControlPlane Breaks OCP Cluster Upgrades


    • Sprint:
      Service Mesh 1.0.2


      Actual Behavior
      The default ServiceMeshControlPlane installed by following https://access.redhat.com/documentation/en-us/openshift_container_platform/4.1/html/service_mesh/service-mesh-installation#ossm-control-plane-deploy-operatorhub_installing-ossm and just creating a new SMCP from the default template breaks the OCP cluster from being able to upgrade.

      Expected Behavior
      Service Mesh should prevent users from configuring a SMCP that blocks cluster upgrades. And, the default SMCP yaml example shipped with the operator should definitely not cause this problem.

      Additional Details
      Cluster upgrades are blocked because Service Mesh is created some Deployments with PodDisruptionBudgets with a minAvailable of 1 but those Deployments also only have their number of replicas set to 1. That means that when OCP tries to drain and restart cluster nodes, any nodes containing these Service Mesh Deployments can never drain. This breaks being able to update the cluster as new OCP releases come out.

      This was initially reported as a bug against OpenShift Serverless at https://bugzilla.redhat.com/show_bug.cgi?id=1747472. The Openshift Serverless Operator will attempt to workaround this bug for now, but it's ultimately a bug in Service Mesh itself.

        Gliffy Diagrams


            Issue Links



                • Assignee:
                  kconner Kevin Conner
                  benbrowning Ben Browning
                • Votes:
                  0 Vote for this issue
                  6 Start watching this issue


                  • Created: