Kubernetes takes in the mixin that most clusters already run

SIG Instrumentation has adopted kubernetes-mixin, a project with more than eight years of history. Stephen Lang of Grafana Labs wrote up the move on 12 October 2026. The repository transfer into the Kubernetes SIGs organization was completed on 8 September, tracked as KEP-5905.

The mixin is the practical monitoring kit a lot of clusters already use. Kubernetes components, kube-state-metrics, cAdvisor, and node_exporter expose a large set of metrics. Knowing which ones matter, which PromQL stays cheap on a big fleet, and which numbers mean “page someone” is a separate problem. kubernetes-mixin ships recording rules, alert rules such as KubePodCrashLooping, and Grafana dashboards that fit together. People install them directly or through bundles such as kube-prometheus-stack.

The awkward part was status. Operators treated the alerts as standard, but they were not an official Kubernetes project, so official docs had a hard time pointing at them. A documentation request inside SIG Instrumentation made that gap obvious. Maintainers agreed to move the existing repository rather than fork it, so current Git remotes and runbook URLs keep working.

Maintenance does not stop because the org changed. The mixin still tracks the Kubernetes release cycle, including new metrics and removals that show up in real clusters. What changes is that a cluster’s default dashboards and alerts can now be discussed as part of the project, not as a popular downstream extra.

If your Prometheus stack already loads these rules, the immediate action is small: know which copy you run, and follow the SIG repository for the next Kubernetes release instead of an unofficial mirror.