ローカルKubernetesの構築

Create a local Kubernetes development environment for fast iteration and testing.

使用タイミング

• Need local Kubernetes environment for application development
• Want to test Kubernetes manifests and Helm charts before deploying to production
• Require fast inner-loop development with automatic rebuild and redeploy
• Testing multi-service applications with service dependencies
• Learning Kubernetes without cloud costs
• CI/CD pipeline testing locally before pushing changes
• Need isolated environment for experimentation and debugging

入力

• 必須: Docker Desktop or Docker Engine installed
• 必須: At least 4GB RAM available for cluster
• 必須: Choice of local cluster tool (kind, k3d, or minikube)
• 任意: Application source code to deploy
• 任意: Kubernetes version preference
• 任意: Development tool preference (Skaffold, Tilt, or manual)
• 任意: Number of worker nodes needed

手順

See Extended Examples for complete configuration files and templates.

ステップ1: Install Local Kubernetes Cluster Tool

Choose and install kind, k3d, or minikube based on your requirements.

Install kind (Kubernetes in Docker):

# Linux example
curl -Lo ./kind https://kind.sigs.k8s.io/dl/v0.20.0/kind-linux-amd64
chmod +x ./kind
sudo mv ./kind /usr/local/bin/kind

# Verify installation
kind version

Install k3d (k3s in Docker):

# Linux/macOS
curl -s https://raw.githubusercontent.com/k3d-io/k3d/main/install.sh | bash

# Verify installation
k3d version

Install minikube:

# Linux example
curl -LO https://storage.googleapis.com/minikube/releases/latest/minikube-linux-amd64
sudo install minikube-linux-amd64 /usr/local/bin/minikube

# Verify installation
minikube version

Install kubectl if not already present:

# Linux example
curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
chmod +x kubectl
sudo mv kubectl /usr/local/bin/
kubectl version --client

See references/EXAMPLES.md for macOS and Windows installation commands.

期待結果: Tool binary installed and in PATH. Version command returns expected version. kubectl available for cluster interaction.

失敗時:

• Ensure Docker is running: docker ps
• Check system PATH includes installation directory
• For permission issues, verify sudo/admin rights
• On macOS, may need to allow binary in Security & Privacy settings
• Windows users: ensure running terminal as Administrator

ステップ2: Create Local Cluster with Configuration

Create a multi-node cluster with ingress and local registry support.

Create kind cluster:

# kind-config.yaml (abbreviated)
kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
name: dev-cluster
nodes:
- role: control-plane
  extraPortMappings:
  - containerPort: 80
    hostPort: 80
  - containerPort: 443
    hostPort: 443
- role: worker
- role: worker

# Create cluster
kind create cluster --config kind-config.yaml

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

# Create local registry
docker run -d --restart=always -p 5000:5000 --name kind-registry registry:2
docker network connect kind kind-registry

See references/EXAMPLES.md for complete kind-config.yaml with registry mirrors and ingress configuration.

Create k3d cluster:

# Create cluster with ingress and registry
k3d cluster create dev-cluster \
  --port "80:80@loadbalancer" \
  --port "443:443@loadbalancer" \
  --agents 2 \
  --registry-create k3d-registry:5000

Create minikube cluster:

# Create cluster with multiple nodes
minikube start \
  --nodes=3 \
  --cpus=2 \
  --memory=4096 \
  --driver=docker \
  --addons=ingress,registry,metrics-server

Test cluster:

# Deploy test application
kubectl create deployment hello --image=k8s.gcr.io/echoserver:1.4
kubectl expose deployment hello --type=NodePort --port=8080
kubectl port-forward service/hello 8080:8080

# Clean up test
kubectl delete deployment,service hello

期待結果: Multi-node cluster running with control plane and worker nodes. Ingress controller installed and ready. Local registry accessible at localhost:5000. kubectl context set to new cluster. Test deployment successful.

失敗時:

• Check Docker has sufficient resources (4GB+ memory recommended)
• Verify no port conflicts: lsof -i :80,443,5000,6550
• For kind: ensure Docker desktop Kubernetes is disabled (conflicts)
• For k3d: check Docker network connectivity
• For minikube: try different driver (virtualbox, hyperv, kvm2)
• Review cluster creation logs: kind get clusters, k3d cluster list, minikube logs

ステップ3: Configure Development Workflow Tools

Set up Skaffold or Tilt for automated rebuild and redeploy.

Install Skaffold:

# Linux example
curl -Lo skaffold https://storage.googleapis.com/skaffold/releases/latest/skaffold-linux-amd64
chmod +x skaffold
sudo mv skaffold /usr/local/bin
skaffold version

Create Skaffold configuration:

# skaffold.yaml (abbreviated)
apiVersion: skaffold/v4beta7
kind: Config
metadata:
  name: my-app
build:
# ... (see EXAMPLES.md for complete configuration)

See references/EXAMPLES.md for complete skaffold.yaml with profiles, file sync, and port forwarding.

Install Tilt:

# Linux/macOS
curl -fsSL https://raw.githubusercontent.com/tilt-dev/tilt/master/scripts/install.sh | bash
tilt version

Create Tiltfile:

# Tiltfile (abbreviated)
allow_k8s_contexts('kind-dev-cluster')

docker_build(
  'localhost:5000/my-app',
  '.',
  live_update=[
    sync('./src', '/app/src'),
  ]
)

k8s_yaml(['k8s/deployment.yaml', 'k8s/service.yaml'])
k8s_resource('my-app', port_forwards='8080:8080')

See references/EXAMPLES.md for complete Tiltfile with live updates, Helm charts, and custom buttons.

Create sample Kubernetes manifests:

# k8s/deployment.yaml (abbreviated)
apiVersion: apps/v1
kind: Deployment
metadata:
  name: my-app
spec:
  replicas: 1
  template:
    spec:
      containers:
      - name: app
        image: localhost:5000/my-app
        ports:
        - containerPort: 8080

See references/EXAMPLES.md for complete manifests with service, ingress, and resource limits.

Test development workflow:

# Using Skaffold
skaffold dev --port-forward

# Using Tilt
tilt up

# Add entry to /etc/hosts for ingress
echo "127.0.0.1 my-app.local" | sudo tee -a /etc/hosts
curl http://my-app.local

期待結果: Skaffold or Tilt watching for file changes. Code changes trigger automatic rebuild and redeploy. Hot reload working for supported languages. Port forwarding allows local access. Logs streaming in terminal/UI. Build caching makes rebuilds fast.

失敗時:

• Verify Docker daemon accessible: docker ps
• Check if local registry reachable: curl http://localhost:5000/v2/_catalog
• For file sync issues, ensure paths in config match actual structure
• Review Skaffold/Tilt logs for build errors
• Ensure Dockerfile has proper base image and builds successfully: docker build .
• Check resource limits not causing OOMKills: kubectl describe pod -l app=my-app

ステップ4: Set Up Local Storage and Databases

Configure persistent storage and deploy database services for testing.

Create local storage class:

# local-storage.yaml (abbreviated)
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: local-path
provisioner: rancher.io/local-path
# ... (see EXAMPLES.md for complete configuration)

See references/EXAMPLES.md for complete storage configuration with PVC templates.

Deploy PostgreSQL for development:

# postgres-dev.yaml (abbreviated)
apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: postgres
spec:
  serviceName: postgres
  template:
    spec:
      containers:
      - name: postgres
        image: postgres:15-alpine
        envFrom:
        - secretRef:
            name: postgres-secret

See references/EXAMPLES.md for complete PostgreSQL StatefulSet with secrets and volume templates.

Deploy Redis for caching:

# Using Helm
helm install redis bitnami/redis \
  --set auth.enabled=false \
  --set replica.replicaCount=0

See references/EXAMPLES.md for kubectl-based Redis deployment.

Test database connectivity:

# Apply manifests
kubectl apply -f local-storage.yaml
kubectl apply -f postgres-dev.yaml

# Wait for PostgreSQL
kubectl wait --for=condition=ready pod -l app=postgres --timeout=60s

# Test connection
kubectl exec -it postgres-0 -- psql -U devuser -d devdb -c "SELECT version();"

期待結果: Storage class configured for dynamic provisioning. Database pods running and ready. Services accessible via port-forward or from other pods. Data persists across pod restarts. Resource usage appropriate for development (small limits).

失敗時:

• Check if storage provisioner installed: kubectl get storageclass
• Verify PVC bound to PV: kubectl get pvc,pv
• Review pod events for mounting errors: kubectl describe pod postgres-0
• For permission issues, check if hostPath directory exists and is writable
• Test database startup: kubectl logs postgres-0 for PostgreSQL errors
• Ensure no port conflicts for port-forwarding

ステップ5: Configure Observability for Local Development

Add minimal monitoring and logging for debugging.

Deploy lightweight monitoring stack:

# Install metrics-server
kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml

# For local clusters, disable TLS verification
kubectl patch deployment metrics-server -n kube-system --type='json' -p='[
  {"op": "add", "path": "/spec/template/spec/containers/0/args/-", "value": "--kubelet-insecure-tls"}
]'

# Verify metrics
kubectl top nodes
kubectl top pods -A

Set up local logging:

# Install stern (multi-pod log tailing)
curl -Lo stern https://github.com/stern/stern/releases/download/v1.26.0/stern_1.26.0_linux_amd64.tar.gz
tar -xzf stern_1.26.0_linux_amd64.tar.gz
sudo mv stern /usr/local/bin/

# Usage
stern my-app --since 1m

See references/EXAMPLES.md for development dashboard ConfigMaps and useful aliases.

期待結果: Metrics-server providing resource usage data. kubectl top commands working. k9s or dashboard showing cluster status. Logs accessible via stern or kubectl logs. Low overhead monitoring suitable for development.

失敗時:

• For metrics-server TLS errors, apply insecure TLS flag patch
• Check if metrics-server pod running: kubectl get pods -n kube-system -l k8s-app=metrics-server
• Verify heapster API available: kubectl get apiservices | grep metrics
• For stern, ensure kubectl context is set correctly
• Test basic kubectl access before debugging observability tools

ステップ6: Document Workflow and Create Helpers

Create scripts and documentation for team onboarding.

Create setup script:

#!/bin/bash
# setup-local-cluster.sh (abbreviated)
set -e

echo "=== Local Kubernetes Cluster Setup ==="

# ... (see EXAMPLES.md for complete configuration)

See references/EXAMPLES.md for complete setup script with service deployment and verification.

Create teardown script:

#!/bin/bash
# teardown-local-cluster.sh (abbreviated)
echo "=== Tearing Down Local Cluster ==="

if kind get clusters 2>/dev/null | grep -q dev-cluster; then
  kind delete cluster --name dev-cluster
  docker stop kind-registry && docker rm kind-registry
fi

docker system prune -f

See references/EXAMPLES.md for complete teardown script and README template.

期待結果: Setup script creates cluster in one command. Teardown script cleans everything up. README provides clear instructions for common tasks. Team members can get productive quickly.

失敗時:

• Test scripts manually before distributing
• Add error handling for each step
• Provide troubleshooting section in README
• Create video walkthrough for complex setups
• Maintain scripts as cluster tool versions update

バリデーション

• Local cluster created with multiple nodes
• Ingress controller installed and responding
• Local registry accessible and accepting pushes
• Sample application deploys successfully
• File sync working (changes reflected without full rebuild)
• Port forwarding allows local access to services
• Database services running and accessible
• Metrics server providing resource usage
• Logs accessible via kubectl/stern/Tilt
• Setup/teardown scripts work reliably
• Documentation clear and up-to-date
• Team members can onboard in <30 minutes

よくある落とし穴

• Insufficient Resources: Local clusters need 4GB+ RAM, 2+ CPU cores. Check Docker Desktop settings. Reduce replicas and resource requests for development.

• Port Conflicts: Ports 80, 443, 5000 commonly used. Check with lsof -i :<port> before cluster creation. Adjust port mappings if needed.

• Slow Rebuilds: Without proper caching, Docker rebuilds are slow. Use multi-stage builds, .dockerignore, and BuildKit. Enable Skaffold/Tilt caching.

• Context Confusion: Multiple kubectl contexts cause confusion. Use kubectl config current-context and kubectx tool to switch clearly.

• File Sync Not Working: Path mismatches between host and container break sync. Verify paths in skaffold.yaml/Tiltfile match Dockerfile WORKDIR.

• Ingress Not Resolving: Forgot to add entry to /etc/hosts. Or ingress controller not ready. Wait for controller pods before testing.

• Database Data Loss: Default storage ephemeral. Use PersistentVolumes for data that should survive restarts. Be explicit about storage class.

• Resource Limits Too High: Don't copy production resource specs to local. Reduce limits significantly for local development to fit in Docker Desktop.

• Network Isolation: Local cluster can't always reach host services. Use host.docker.internal (Docker Desktop) or ngrok for reverse proxying.

• Version Skew: Local cluster version differs from production. Explicitly set Kubernetes version during creation to match production.

関連スキル

• deploy-to-kubernetes - Application deployment patterns tested locally first
• write-helm-chart - Helm charts tested in local cluster
• setup-prometheus-monitoring - Monitoring setup tested locally
• configure-ingress-networking - Ingress configuration validated locally
• implement-gitops-workflow - GitOps tested with local cluster
• optimize-cloud-costs - Cost optimization strategies developed locally