k8s.io/perf-tests/clusterloader2@v0.0.0-20240304094227-64bdb12da87e/testing/experiments/ignore_known_gce_container_restarts.yaml (about)

     1  RESTART_COUNT_THRESHOLD_OVERRIDES: |
     2    # To be investigated in https://github.com/kubernetes/perf-tests/issues/872.
     3    fluentd-gcp: 999
     4  
     5    # Main purpose of this check is detection crashlooping pods.
     6    # It was extended to check whether master pods were restarted even once.
     7    # For components that we run multiple instances of we should be less aggressive and tolerate restarts e.g. due to node restarts.
     8    kube-proxy: 2
     9    metadata-proxy: 2
    10    prometheus-to-sd-exporter: 2
    11    coredns: 2
    12    konnectivity-agent: 2
    13    node-problem-detector: 2
    14  
    15    # Allow for a single restart of master components.
    16    # As long as tests are passing, a single restart shouldn't be a problem
    17    kube-scheduler: 1
    18    kube-controller-manager: 1
    19    l7-lb-controller: 1