Setting Up Your Cluster

You do not need a data center, or a cloud account, to learn Kubernetes. A normal laptop is enough. In this lesson, you will see the options for running a cluster, and set up the exact one we use in this course.

• A laptop is enough
• Pick one local cluster tool
• Then install kubectl

Ways to run Kubernetes

There are several ways to get a cluster. Minikube starts a small cluster on your own computer, inside Docker or a virtual machine, and it is perfect for learning. Kind, which means Kubernetes in Docker, is similar, and popular for testing. Docker Desktop includes a single node cluster you can switch on. K three S is a very light version, great for small servers. And in production, most teams use a managed service, like E K S on Amazon, G K E on Google, or A K S on Azure. All of them run the same Kubernetes, so what you learn applies everywhere.

Install and start

Let's set up. We use minikube, and the command line tool, kube control. On a Mac, you can install both with homebrew, and on Windows and Linux, the Kubernetes website shows the matching steps. Make sure Docker is running, because minikube will use it. Then one command creates a cluster. We ask for three nodes, so we can see how pods spread across machines. After a minute, the cluster is ready. Finally, we ask for the list of nodes, to prove it works.

# install the tools (macOS)
brew install kubectl minikube

# start a cluster with three nodes
minikube start --nodes=3 --driver=docker

# check that it works
kubectl get nodes

Your cluster is alive

This is what success looks like. Kube control reads its settings, and asks the A P I server for the nodes. We see one control plane node, and two workers, all with the status ready. Let's also check the versions. The client version is the kube control program, on your computer. It is fine for the cluster to be a little older or newer, as long as they are close. If you see ready, you are ready to learn.

kubectl get nodes
NAME      STATUS   ROLES           AGE   VERSION
k8s       Ready    control-plane   14m   v1.34.0
k8s-m02   Ready    <none>          14m   v1.34.0
k8s-m03   Ready    <none>          14m   v1.34.0
kubectl version --client
Client Version: v1.34.1
Kustomize Version: v5.7.1
kubectl config current-context
k8s

Summary