kubernetes/kube-state-metrics
6,141 stars·2,187 forks·Go·Apache-2.0·2 viewskubernetes.io/docs/concepts/cluster-administration/kube-state-metrics↗
Kube State Metrics
kube-state-metrics is a Kubernetes metrics exporter that generates Prometheus-compatible metrics from the current state of cluster objects such as pods, deployments, and nodes. It operates by watching the Kubernetes API server and transforming resource snapshots into metric families, which are then exposed over an HTTP endpoint in the Prometheus text-based exposition format for direct scraping.
The project distinguishes itself through horizontal scaling capabilities, distributing metric collection across multiple instances using object UID hashing to reduce per-instance memory consumption. It also supports per-node sharding for pod metrics using field selectors, and includes self-monitoring metrics for operational observability, along with health check endpoints for Kubernetes startup, liveness, and readiness probes.
The tool provides capabilities for filtering metrics by resource type, querying object state through labels and annotations, and restricting access to specific namespaces. It also includes label-name conflict resolution for Prometheus compatibility and supports configuration of admission webhooks for request filtering and mutation.
