Core Concepts & Architecture

The control plane is the cluster brain. It makes global decisions and constantly works to make the cluster’s real state match the state you asked for.

The kube-apiserver is the front door to the control plane. Every interaction — yours via kubectl, or between components — flows through this API.

etcd is the consistent key-value store holding the entire cluster state. It is your single source of truth — if etcd dies, the cluster loses its memory.

The kube-scheduler spots Pods with no home and picks the best node for each, weighing resources, constraints, and affinity rules.

The controller manager runs control loops that keep reality matching your desired state — noticing dead nodes, ensuring the right Pod count, and more.

Worker nodes are where your containers actually run. Each one carries the services needed to run Pods, handle networking, and talk to the control plane.

The kubelet is the agent on every worker node. It takes Pod specs from the API server and makes sure those containers are running and healthy.

kube-proxy runs on each node and handles networking for Services — routing traffic to healthy Pods, whether it comes from inside or outside the cluster.

Solid work! You now know the master-worker model, the control plane (API server, etcd, scheduler, controller manager), and the worker pieces (kubelet, kube-proxy).