Key Concepts and Terminologies

Your first kubectl output is full of words — node, pod, namespace, kubelet. Without a map, every error message reads like a riddle. This lesson is that map.

The rule

A cluster is a brain and some hands: the control plane decides, nodes do.

The cluster is the whole system. It has two halves.

The control plane is the brain:

• kube-apiserver — the front door. Every command, from you or from inside, goes through it.
• etcd — a key-value store holding the desired and recorded state of everything.
• kube-scheduler — picks which node each new pod should run on.
• kube-controller-manager — runs the control loops that close gaps between wish and reality.

The nodes are the hands — the machines that actually run your containers. Each has:

• kubelet — an agent that starts the containers the control plane assigned and reports back.
• container runtime — the engine that really runs containers, such as containerd.
• kube-proxy — maintains network rules so Service traffic reaches the right pods.

The objects you create:

• Pod — the smallest unit you deploy: one or more containers that share a network address.
• Namespace — a named folder that groups objects, like notes for our app.
• Labels — key-value tags such as app: notes-web; selectors find objects by label.
• Manifest — a YAML file describing an object; kubectl is the command-line tool that sends it to the API server.

The shortcut

For any new word, ask "brain, hands, or thing?" Brain parts decide, hand parts run, and things are what you describe in YAML.

A common mix-up: the scheduler only chooses a node. It never starts a container itself — that is the kubelet's job.

Drills

Answer fast — trust the rule.

Remember

• The control plane (API server, etcd, scheduler, controllers) decides; nodes (kubelet, runtime, kube-proxy) do.
• For any new word ask: brain, hands, or thing?
• Trap: the scheduler picks the node but the kubelet runs the container.