Check manual page of kube_agent_health_node

Kubernetes: Monitoring agent node-scraper health

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

This optional cluster-host service checks one node's kubelet statistics, kubelet health and system-agent payloads independently. Receipt age is WARN at 90 seconds and CRIT at 120 seconds. An expected payload that is missing from the cache is CRIT. These states have no startup grace period.

The service reports the receipt age of each payload independently. Scrape durations, node-scraper versions and Git revisions are available in the service details and metrics.

Missing cached data may never have arrived or may have expired. The defaults accommodate a 60-second collection interval and 120-second cache retention. After expiry the missing-payload state takes precedence over age thresholds.

The rule Kubernetes agent node health configures receipt-age thresholds, missing-payload state, and version mismatch state (OK by default). Item conditions allow different settings for different nodes. Version and Git revision are shown per payload. Missing optional metadata is informational. Each payload with available measurements has receipt-age and scrape-duration metrics. Disabling age thresholds does not disable missing-data detection.

The cluster overview continues to evaluate the node independently using its own rule. Both services can alert for the same fault. Nodes removed from the section use the standard vanished-service behavior.

Item

Kubernetes node name.

Discovery

One service per reported node-scraper node when node services are enabled in Kubernetes agent health service discovery. Also discovers nodes whose payloads are absent from the cache. Services reside on the cluster host.