CNCF Atlas

English

日本語

English

日本語

Appearance

Getting Started ​

Aimed at a local cluster for learning, matching the v1.36 line current at the pinned commit. Commands assume macOS or Linux with Docker running.

Prerequisites ​

  • Docker (or another container runtime that kind supports)
  • kubectl, the Kubernetes CLI
  • kind, which runs a cluster inside Docker containers

Install ​

bash
# kubectl (Linux amd64; see the docs for other platforms)
curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
sudo install -o root -g root -m 0755 kubectl /usr/local/bin/kubectl

# kind
go install sigs.k8s.io/kind@latest

A first working setup ​

The shortest path to a running cluster with a workload on it.

  1. Create a local cluster.

    bash
    kind create cluster --name demo

  2. Run a Deployment and expose it.

    bash
    kubectl create deployment hello --image=registry.k8s.io/echoserver:1.4
kubectl expose deployment hello --port=8080

  3. Watch the scheduler place the Pod and the kubelet start it.

    bash
    kubectl get pods -o wide

Expected output shows the Pod moving to Running with a node assigned in the NODE column.

Verify it works ​

Confirm the control plane is reachable and the node is ready.

bash
kubectl get nodes
kubectl get deployment hello

A healthy cluster reports the control-plane node as Ready and the hello Deployment with 1/1 ready replicas. Port-forwarding the Service to a local port and curling it confirms the workload serves traffic.

bash
kubectl port-forward service/hello 8080:8080 &
curl http://localhost:8080

Where to go next ​

For production concerns such as high availability of the control plane, RBAC and admission hardening, and scaling, follow the official documentation at https://kubernetes.io/docs/. kubeadm (under cmd/kubeadm) is the supported path for bootstrapping real clusters rather than kind.