Cluster architecture

Learn the architecture of control plane and nodes.

Cluster architecture

A Kubernetes cluster consists of two kinds of machines:

Control plane: the brain of the cluster. The API server, scheduler, controllers and the etcd database run here.
Worker nodes: the machines where application pods actually run.

You and all the components talk to the cluster only through the API server.

$ kubectl get nodes
NAME            STATUS   ROLES           AGE   VERSION
control-plane   Ready    control-plane   12d   v1.31.2
worker-1        Ready    <none>          12d   v1.31.2
worker-2        Ready    <none>          12d   v1.31.2

Where is a pod running?

The scheduler picks a suitable node for each pod. -o wide shows which node a pod runs on and with which IP:

$ kubectl get pods -o wide
NAME                  READY   STATUS    IP           NODE
web-7c5b9f6d8-abcde   1/1     Running   10.244.1.5   worker-1
web-7c5b9f6d8-fghij   1/1     Running   10.244.2.7   worker-2

In managed services (EKS, GKE, AKS), the provider runs the control plane; you only see the worker nodes.

Exercise Look at the cluster

Write two commands: list the nodes in wide format (-o wide); then list the pods together with the nodes they run on.

Show example solution

kubectl get nodes -o wide
kubectl get pods -o wide

Test yourself

Which component of the cluster do kubectl commands go to?
- Directly to etcd
- To the worker nodes
- To the Docker daemon
- kube-apiserver
Show answer

kube-apiserver — The API server authenticates, validates the request and writes to etcd.
List the machines in the cluster.
```
kubectl get ____
```
Show answer
```
kubectl get nodes
```
Match.

control plane

The cluster's brain
worker node

Runs pods
kube-apiserver

Single entry point
-o wide

Node and IP columns

Next lesson: Control plane

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies