Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-31037

Must-gather does not collect performance related information from tainted nodes

XMLWordPrintable

    • Icon: Bug Bug
    • Resolution: Unresolved
    • Icon: Minor Minor
    • None
    • 4.14
    • oc
    • None
    • No
    • False
    • Hide

      None

      Show
      None
    • 2024-04-17: Waiting for maintainer approval

      Description of problem:

          When running must-gather, a DaemonSet is created to collect performance related information for nodes in a cluster. If a node is tainted (for example with well defined OpenShift taints for Infra nodes, ODF nodes, master nodes etc), then the DaemonSet does not create Pods on these nodes and the information is not collected.

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

          4.14.z

      How reproducible:

          Reproducible

      Steps to Reproduce:

      1. Taint a node in a cluster with a custom taint i.e. "oc adm taint  node <node_name> node-role.kubernetes.io/infra=reserved:NoSchedule node-role.kubernetes.io/infra=reserved:NoExecute". Ensure at least one node is not tainted.    
      
      2.Run `oc adm must-gather` to generate report to local filesystem
        
          

      Actual results:

          The performance stats collected under directory <must_gather_dir>/nodes/ only contains results for nodes without taints.

      Expected results:

          The performance stats collected under directory <must_gather_dir>/nodes/ should contain entries for all nodes in the cluster.

      Additional info:

          This issue has been identified by using the Performance Profile Creator. This tool requires the output of must-gather as its input (as described in the instructions here: https://docs.openshift.com/container-platform/4.14/scalability_and_performance/cnf-create-performance-profiles.html#running-the-performance-profile-profile-cluster-using-podman_cnf-create-performance-profiles). When following this guide, the missing performance information for tainted nodes results in being returned the error "failed to load node's worker's GHW snapshot: can't obtain the path: <node_name>" when running the tool in discovery mode

       

       

            msivak@redhat.com Martin Sivak
            rh-ee-mperry Maximilian Perry
            Workloads Team Bot Account Workloads Team Bot Account
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

              Created:
              Updated: