Uploaded image for project: 'OpenShift Monitoring'
  1. OpenShift Monitoring
  2. MON-3452

Configure trackTimestampsStaleness for kubelet ServiceMonitor


    • Icon: Epic Epic
    • Resolution: Done
    • Icon: Normal Normal
    • openshift-4.15
    • None
    • None
    • kubelet staleness
    • False
    • None
    • False
    • Not Selected
    • NEW
    • To Do
    • MON-3159Technical Debt
    • NEW
    • 0% To Do, 0% In Progress, 100% Done
    • 0

      In https://issues.redhat.com/browse/MON-1949 we added a feature to deploy an additional kubelet service monitor that adds cAdvisor metrics with the exposed timestamps. This improves consistency of some container metrics, at the cost of delayed staleness awareness.

      With https://github.com/prometheus/prometheus/pull/13060/ we can finally ask prometheus to ingest explicit timestamps but also use staleness markers for low lantency stale detection. I.e. we can deprecate and remove the dedicates ServiceMonitor feature when this Prometheus feature ships in OCP.

            jfajersk@redhat.com Jan Fajerski
            jfajersk@redhat.com Jan Fajerski
            Tai Gao Tai Gao
            0 Vote for this issue
            5 Start watching this issue