DockerMetricsMissingFE

This alert is triggered when docker metrics are missing at least one node of the cluster for more than 10 minutes.

Troubleshooting hints

• Check if the exporter is healthy

  Caascad cloud and Infra zonesCaascad client zonesNGOT zones

  export NAMESPACE="docker-exporter"
  kubectl get pods -l app.kubernetes.io/name=caascad-docker-exporter -n $NAMESPACE
  kubectl logs -l app.kubernetes.io/name=caascad-docker-exporter -n $NAMESPACE
  

  export NAMESPACE="caascad-monitoring"
  kubectl get pods -l app.kubernetes.io/name=caascad-docker-exporter -n $NAMESPACE
  kubectl logs -l app.kubernetes.io/name=caascad-docker-exporter -n $NAMESPACE
  

  export NAMESPACE="docker-exporter"
  kubectl get pods -l app.kubernetes.io/name=caascad-docker-exporter -n $NAMESPACE
  kubectl logs -l app.kubernetes.io/name=caascad-docker-exporter -n $NAMESPACE
  

• Check that docker metrics (ex: docker_data_space_available_bytes) are available on the related Grafana. Choose either the Thanos datasource or Thanos-client if it is about a Caascad cloud or client zone. For NGOT zone, choose Thanos datasource.

• Check if there is a problem with the docker daemon on the cluster nodes

  kubectl node-shell <NODE> -- systemctl status docker.service -l