Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-18939

The apiservers shouldn't expose both SLO and SLI latency metrics to Prometheus

XMLWordPrintable

    • No
    • False
    • Hide

      None

      Show
      None
    • Developer Preview
    • In Progress

      All the apiservers:

      • kube-apiserver
      • openshift-apiserver
      • oauth-apiserver

      Expose both `apiserver_request_slo_duration_seconds` and `apiserver_request_sli_duration_seconds`. The SLI metric was introduced in Kubernetes 1.26 as a replacement of `apiserver_request_slo_duration_seconds` which was deprecated in Kubernetes 1.27. This change is only a renaming so both metrics expose the same data. To avoid storing duplicated data in Prometheus, we need to drop the `apiserver_request_slo_duration_seconds` in favor of `apiserver_request_sli_duration_seconds`.

              dgrisonn@redhat.com Damien Grisonnet
              dgrisonn@redhat.com Damien Grisonnet
              Ke Wang Ke Wang
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

                Created:
                Updated:
                Resolved: