Kind: Running Multi-Node Kubernetes Clusters Locally in Docker

Testing multi-node Kubernetes behaviors (such as node affinity, pod topology spread constraints, cordon/drain operations, and HA control plane failover) requires multi-node clusters. Spinning up multi-node clusters on cloud providers is slow and expensive, while Minikube defaults to a single VM.

Kind (Kubernetes IN Docker) is a tool maintained by the Kubernetes Special Interest Group (SIG Testing) that runs fully compliant multi-node Kubernetes clusters where each Kubernetes "node" is simply an isolated Docker container.

With Kind, you can spin up a complete 3-control-plane, 3-worker HA cluster on a developer laptop in under 60 seconds.

flowchart TD
    subgraph HostSystem["Developer Laptop / CI Runner (macOS / Linux)"]
        DockerDaemon["Docker Daemon (dockerd)"]
        KindCLI["Kind CLI (kind create cluster)"]
    end

    subgraph KindContainers["Docker Containers as Kubernetes Nodes"]
        C1["Container: kind-control-plane-1<br/>(kube-apiserver, etcd)"]
        C2["Container: kind-control-plane-2<br/>(kube-apiserver, etcd)"]
        C3["Container: kind-control-plane-3<br/>(kube-apiserver, etcd)"]
        LB["Container: kind-external-load-balancer<br/>(HAProxy for APIServer)"]
        
        W1["Container: kind-worker-1<br/>(containerd, kubelet)"]
        W2["Container: kind-worker-2<br/>(containerd, kubelet)"]
        W3["Container: kind-worker-3<br/>(containerd, kubelet)"]
    end

    KindCLI --> DockerDaemon
    DockerDaemon --> LB
    LB --> C1
    LB --> C2
    LB --> C3
    DockerDaemon --> W1
    DockerDaemon --> W2
    DockerDaemon --> W3

Provisioning a Multi-Node HA Cluster with Ingress Port Mapping

Create a cluster configuration file named kind-ha-config.yaml:

kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
nodes:
  # Highly Available Control Plane with HAProxy
  - role: control-plane
    kubeadmConfigPatches:
      - |
        kind: InitConfiguration
        nodeRegistration:
          kubeletExtraArgs:
            node-labels: "ingress-ready=true"
    extraPortMappings:
      - containerPort: 80
        hostPort: 80
        protocol: TCP
      - containerPort: 443
        hostPort: 443
        protocol: TCP
  - role: control-plane
  - role: control-plane
  # Worker Nodes
  - role: worker
  - role: worker
  - role: worker

Launch the cluster:

kind create cluster --name lab-cluster --config kind-ha-config.yaml

Verify that all 6 nodes are online:

kubectl get nodes -o wide

Sideloading Images Without Pushing to a Remote Registry

Waiting for CI to push large 2GB container images to Docker Hub or GHCR only to pull them right back down into a local cluster wastes time and bandwidth.

Kind allows you to sideload images directly from your host Docker daemon into the cluster nodes:

# 1. Build image locally on your workstation
docker build -t my-app:local .

# 2. Sideload image directly into Kind cluster nodes
kind load docker-image my-app:local --name lab-cluster

Now, set imagePullPolicy: IfNotPresent in your deployment manifest, and Kubernetes will launch the container immediately without touching the internet!

Deploying an Ingress Controller on Kind

Deploy the official NGINX Ingress controller optimized for Kind's host port mappings:

kubectl apply -f https://raw.githubusercontent.com/kubernetes/ingress-nginx/main/deploy/static/provider/kind/deploy.yaml

Wait for ingress pods to become ready:

kubectl wait --namespace ingress-nginx \
  --for=condition=ready pod \
  --selector=app.kubernetes.io/component=controller \
  --timeout=90s

You can now reach any in-cluster HTTP service directly on http://localhost:80 using standard Ingress manifests!

Kind in Automated CI/CD Pipelines (GitHub Actions)

Kind is the gold standard for testing Helm charts, operators, and end-to-end Kubernetes integration tests in GitHub Actions:

name: E2E Tests
on: [push, pull_request]

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Create Kind Cluster
        uses: helm/kind-action@v1.9.0
        with:
          cluster_name: ci-cluster
      - name: Run E2E Tests
        run: |
          kubectl get nodes
          helm install my-chart ./chart

Production Tips for Local Development

• Ephemeral Cluster Teardown: Clean up your environment anytime with a single command: kind delete cluster --name lab-cluster.
• Mount Local Workstation Folders: Use extraMounts in your Kind config to mount source code folders directly into container nodes for hot-reloading development loops.