kubespray-monitoring
Installation
SKILL.md
Kubespray Cluster Monitoring
Overview
Production monitoring for Kubespray-deployed clusters uses a three-layer stack: NFS provisioner for persistent storage, kube-prometheus-stack for metrics collection and alerting, and etcd metrics exposure for cluster health visibility.
Core principle: Monitoring must cover all cluster layers - infrastructure (node-exporter), Kubernetes components (API server, scheduler, controller-manager), etcd (leader, WAL, peer latency), and workloads (pod metrics via kube-state-metrics).
When to Use
- Setting up Prometheus, Grafana, and Alertmanager on Kubespray clusters
- Deploying NFS storage provisioner for monitoring persistence
- Importing Grafana dashboards (community or custom)
- Enabling etcd metrics collection
- Writing PromQL queries for cluster health
Not for: Initial cluster deployment (use kubespray-deployment), HA configuration (use kubespray-ha-configuration), cluster upgrades (use kubespray-operations), troubleshooting failures (use kubespray-troubleshooting)