Werk #19862: Kubernetes Cluster collector service: Show and alert on cache metrics
| Component | Checks & agents | ||||
| Title | Kubernetes Cluster collector service: Show and alert on cache metrics | ||||
| Date | May 1, 2026 | ||||
| Level | Trivial Change | ||||
| Class | New Feature | ||||
| Compatibility | Incompatible - Manual interaction might be required | ||||
| Checkmk versions & editions |
|
The Checkmk Kubernetes cluster collector now provides metrics about its internal caches, which Checkmk monitors in the Cluster collector service on the Kubernetes source host.
As the cluster collector stores information about nodes and containers, the cache size increases naturally over time. When the maximum cache size is reached, gaps might appear in your graphs for Kubernetes related services.
The maximum cache size is configurable in the cluster collector Helm chart
(clusterCollector.cacheMaxsize).
You can now set alerts when the cache grows beyond a particular absolute size or crosses a percentage threshold. As alerting is enabled by default, this Werk is marked incompatible. The default thresholds for both cache types (container metrics and machine sections) are to warn at 80% full and go critical at 95%. These can be configured or disabled in the Kubernetes Collector info ruleset.