API Server, Scheduler, and Controller Manager Debugging

The API server returns 500s on every request. Pods stop scheduling. Deployments freeze mid-rollout. All three symptoms point to different control plane components, and each one has its own health endpoints, log locations, and failure modes. If you don't know which port the scheduler listens on or what /livez versus /readyz actually means, you'll waste time guessing instead of diagnosing.

This article covers the three non-etcd control plane components: kube-apiserver, kube-scheduler, and kube-controller-manager. You'll learn how kubeadm configures them as static pods, what health endpoints each one exposes, how leader election works (and fails), and how to read audit logs when you need to trace what happened. This is CKA Troubleshooting domain territory (30% of the exam), and the exam expects you to move fast.