Check manual page of kube_agent_health_reflector

Kubernetes: Monitoring agent API reflector health

Included in	All Checkmk editions
Source Code License	Open Source
Supported Agents	Kubernetes

This optional cluster-host service checks the API reflector for one Kubernetes resource kind. Before its first list completes it is WARN. An ongoing list or relist is WARN at 120 seconds and CRIT at 300 seconds. Recent watch-error alerts are disabled by default. States, duration thresholds and the recent-error window are configurable in Kubernetes agent reflector health.

The summary shows initialization state, the duration of any ongoing list and the time since the last resource event. The age and duration of the last completed list, and recent watch errors, are available in the service details. Resource-event age thresholds are disabled by default because a quiet resource kind may legitimately receive no changes for a long time.

Last completed-list age and duration, and watch errors since cluster-aggregator start, are diagnostic. Completed-list age is not used for alerting because a healthy watch can run indefinitely. If watch-error alerts are enabled, inspect the cluster-aggregator logs for the underlying Kubernetes API error. The alert expires with time and does not establish recovery. The lifetime error total resets when the cluster aggregator restarts and is graphed as a total, not as an error rate. Other metrics are the duration of an ongoing list (while present), the duration of the last completed list and the time since the last resource event.

The cluster overview evaluates reflectors independently using its own rule; both services can alert for the same fault. Removed reflectors use Checkmk's standard vanished-service behavior.

Item

Kubernetes resource kind, for example Pod or DaemonSet.

Discovery

One service per reported reflector when reflector services are enabled in Kubernetes agent health service discovery. Services reside on the cluster host.