adopt-a-street/deploy

Adopt-a-Street Deployment

This directory contains deployment configurations for the Adopt-a-Street application on Kubernetes (Raspberry Pi cluster).

Directory Structure

deploy/
├── k8s/                              # Kubernetes manifests
│   ├── namespace.yaml                # Namespace definition
│   ├── configmap.yaml                # Environment configuration
│   ├── secrets.yaml.example          # Secret template (COPY TO secrets.yaml)
│   ├── mongodb-statefulset.yaml      # MongoDB StatefulSet with PVC
│   ├── backend-deployment.yaml       # Backend Deployment + Service
│   ├── frontend-deployment.yaml      # Frontend Deployment + Service
│   └── ingress.yaml                  # Ingress for routing
├── README.md                         # This file
└── scripts/                          # Deployment helper scripts

Prerequisites

Cluster Requirements

• Kubernetes cluster with 3 nodes:
  • 2x Raspberry Pi 5 (8GB RAM) - ARM64
  • 1x Raspberry Pi 3B+ (1GB RAM) - ARMv7
• kubectl configured to access your cluster
• Container registry accessible from cluster
• Ingress controller installed (Traefik or NGINX Ingress)
• Persistent storage provisioner (local-path, NFS, or Longhorn)

Local Requirements

• Docker with buildx for multi-arch builds
• kubectl CLI tool
• Access to container registry (Docker Hub, GitHub Container Registry, or private registry)

Quick Start

1. Build Multi-Arch Docker Images

Build images for both ARM64 (Pi 5) and ARMv7 (Pi 3B+):

# From project root
cd /home/will/Code/adopt-a-street

# Create buildx builder (one-time setup)
docker buildx create --use --name multiarch-builder

# Build and push backend
docker buildx build --platform linux/arm64,linux/arm/v7 \
  -t your-registry/adopt-a-street-backend:latest \
  --push ./backend

# Build and push frontend
docker buildx build --platform linux/arm64,linux/arm/v7 \
  -t your-registry/adopt-a-street-frontend:latest \
  --push ./frontend

Note: Replace your-registry with your actual registry (e.g., docker.io/username or ghcr.io/username)

2. Configure Secrets

# Copy secrets template
cp deploy/k8s/secrets.yaml.example deploy/k8s/secrets.yaml

# Edit secrets with your actual values
nano deploy/k8s/secrets.yaml

# IMPORTANT: Add secrets.yaml to .gitignore if not already there
echo "deploy/k8s/secrets.yaml" >> .gitignore

Required Secrets:

• JWT_SECRET - Strong random string for JWT signing
• CLOUDINARY_CLOUD_NAME - Your Cloudinary cloud name
• CLOUDINARY_API_KEY - Your Cloudinary API key
• CLOUDINARY_API_SECRET - Your Cloudinary API secret

3. Update Image References

Update the image references in deployment files:

# Update backend image reference
nano deploy/k8s/backend-deployment.yaml
# Change: image: your-registry/adopt-a-street-backend:latest

# Update frontend image reference
nano deploy/k8s/frontend-deployment.yaml
# Change: image: your-registry/adopt-a-street-frontend:latest

4. Update Domain Name

Update the ingress host:

nano deploy/k8s/ingress.yaml
# Change: host: adopt-a-street.local
# To your actual domain or IP

5. Deploy to Kubernetes

# Create namespace
kubectl apply -f deploy/k8s/namespace.yaml

# Create secrets (IMPORTANT: Make sure you've edited secrets.yaml!)
kubectl apply -f deploy/k8s/secrets.yaml

# Create ConfigMap
kubectl apply -f deploy/k8s/configmap.yaml

# Deploy MongoDB
kubectl apply -f deploy/k8s/mongodb-statefulset.yaml

# Wait for MongoDB to be ready (this may take 1-2 minutes)
kubectl wait --for=condition=ready pod -l app=mongodb -n adopt-a-street --timeout=120s

# Deploy backend
kubectl apply -f deploy/k8s/backend-deployment.yaml

# Wait for backend to be ready
kubectl wait --for=condition=ready pod -l app=backend -n adopt-a-street --timeout=120s

# Deploy frontend
kubectl apply -f deploy/k8s/frontend-deployment.yaml

# Deploy ingress
kubectl apply -f deploy/k8s/ingress.yaml

# Check deployment status
kubectl get all -n adopt-a-street

Verification

Check Pod Status

# View all resources
kubectl get all -n adopt-a-street

# Check pod status
kubectl get pods -n adopt-a-street

# Expected output:
# NAME                                      READY   STATUS    RESTARTS   AGE
# adopt-a-street-backend-xxxxxxxxxx-xxxxx   1/1     Running   0          5m
# adopt-a-street-backend-xxxxxxxxxx-xxxxx   1/1     Running   0          5m
# adopt-a-street-frontend-xxxxxxxxx-xxxxx   1/1     Running   0          5m
# adopt-a-street-frontend-xxxxxxxxx-xxxxx   1/1     Running   0          5m
# adopt-a-street-mongodb-0                  1/1     Running   0          10m

Check Logs

# Backend logs
kubectl logs -f deployment/adopt-a-street-backend -n adopt-a-street

# Frontend logs
kubectl logs -f deployment/adopt-a-street-frontend -n adopt-a-street

# MongoDB logs
kubectl logs -f adopt-a-street-mongodb-0 -n adopt-a-street

Check Services

kubectl get svc -n adopt-a-street

# Expected output:
# NAME                        TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)     AGE
# adopt-a-street-backend      ClusterIP   10.43.x.x       <none>        5000/TCP    5m
# adopt-a-street-frontend     ClusterIP   10.43.x.x       <none>        80/TCP      5m
# adopt-a-street-mongodb      ClusterIP   None            <none>        27017/TCP   10m

Check Ingress

kubectl get ingress -n adopt-a-street

# Get ingress details
kubectl describe ingress adopt-a-street-ingress -n adopt-a-street

Access the Application

# Port forward for testing (if ingress not working)
kubectl port-forward svc/adopt-a-street-frontend 3000:80 -n adopt-a-street

# Then open http://localhost:3000 in your browser

Resource Allocation

The deployment is optimized for Raspberry Pi hardware:

MongoDB (Pi 5 nodes only)

• Requests: 512Mi RAM, 250m CPU
• Limits: 2Gi RAM, 1000m CPU
• Storage: 10Gi persistent volume

Backend (prefers Pi 5 nodes)

• Requests: 256Mi RAM, 100m CPU
• Limits: 512Mi RAM, 500m CPU
• Replicas: 2 pods

Frontend (any node)

• Requests: 64Mi RAM, 50m CPU
• Limits: 128Mi RAM, 200m CPU
• Replicas: 2 pods

Total Cluster Requirements

• Minimum RAM: ~3.5 GB (1.5GB MongoDB + 1GB backend + 200MB frontend + 800MB system)
• Recommended: 2x Pi 5 (8GB each) handles this comfortably

Scaling

Scale Deployments

# Scale backend
kubectl scale deployment adopt-a-street-backend --replicas=3 -n adopt-a-street

# Scale frontend
kubectl scale deployment adopt-a-street-frontend --replicas=3 -n adopt-a-street

Note: MongoDB is a StatefulSet with 1 replica. Scaling MongoDB requires configuring replication.

Updating

Update Images

# Build and push new version
docker buildx build --platform linux/arm64,linux/arm/v7 \
  -t your-registry/adopt-a-street-backend:v2.0 \
  --push ./backend

# Update deployment
kubectl set image deployment/adopt-a-street-backend \
  backend=your-registry/adopt-a-street-backend:v2.0 \
  -n adopt-a-street

# Check rollout status
kubectl rollout status deployment/adopt-a-street-backend -n adopt-a-street

Rollback

# Rollback to previous version
kubectl rollout undo deployment/adopt-a-street-backend -n adopt-a-street

# Rollback to specific revision
kubectl rollout undo deployment/adopt-a-street-backend --to-revision=2 -n adopt-a-street

Monitoring

Resource Usage

# Node resource usage
kubectl top nodes

# Pod resource usage
kubectl top pods -n adopt-a-street

Events

# View recent events
kubectl get events -n adopt-a-street --sort-by='.lastTimestamp'

Describe Resources

# Describe pod (useful for troubleshooting)
kubectl describe pod <pod-name> -n adopt-a-street

# Describe deployment
kubectl describe deployment adopt-a-street-backend -n adopt-a-street

Troubleshooting

Pod Not Starting

# Check pod events
kubectl describe pod <pod-name> -n adopt-a-street

# Check logs
kubectl logs <pod-name> -n adopt-a-street

# Check previous logs (if pod crashed)
kubectl logs <pod-name> -n adopt-a-street --previous

Image Pull Errors

• Verify image exists in registry
• Check image name and tag in deployment
• Verify cluster can access registry
• Check if imagePullSecrets are needed

MongoDB Connection Issues

# Shell into backend pod
kubectl exec -it <backend-pod-name> -n adopt-a-street -- sh

# Test MongoDB connection
wget -qO- http://adopt-a-street-mongodb:27017

Persistent Volume Issues

# Check PVCs
kubectl get pvc -n adopt-a-street

# Check PVs
kubectl get pv

# Describe PVC
kubectl describe pvc mongodb-data-adopt-a-street-mongodb-0 -n adopt-a-street

Cleanup

Delete Everything

# Delete all resources in namespace
kubectl delete namespace adopt-a-street

# Or delete resources individually
kubectl delete -f deploy/k8s/ingress.yaml
kubectl delete -f deploy/k8s/frontend-deployment.yaml
kubectl delete -f deploy/k8s/backend-deployment.yaml
kubectl delete -f deploy/k8s/mongodb-statefulset.yaml
kubectl delete -f deploy/k8s/configmap.yaml
kubectl delete -f deploy/k8s/secrets.yaml
kubectl delete -f deploy/k8s/namespace.yaml

# Note: This will also delete the persistent volume data!

Security Best Practices

1. Never commit secrets.yaml - Always use secrets.yaml.example
2. Use strong JWT_SECRET - Generate with: openssl rand -base64 32
3. Enable TLS/HTTPS - Uncomment TLS section in ingress.yaml and use cert-manager
4. Restrict ingress - Use network policies to limit pod communication
5. Use image digests - Pin images to specific SHA256 digests for production
6. Enable RBAC - Create service accounts with minimal permissions
7. Scan images - Use tools like Trivy to scan for vulnerabilities

Performance Optimization

1. Use imagePullPolicy: IfNotPresent - After initial deployment to save bandwidth
2. Implement HPA - Horizontal Pod Autoscaler for dynamic scaling
3. Add Redis - For caching to reduce MongoDB load
4. Use CDN - For frontend static assets
5. Enable compression - Nginx already configured with gzip
6. Monitor resources - Use Prometheus + Grafana for metrics

Additional Resources

• Kubernetes Documentation
• Raspberry Pi Kubernetes Guide
• Helm Charts - Consider migrating to Helm for easier management
• ArgoCD - GitOps continuous delivery for Kubernetes

Support

For issues or questions:

1. Check pod logs: kubectl logs <pod-name> -n adopt-a-street
2. Check events: kubectl get events -n adopt-a-street
3. Describe resources: kubectl describe <resource> -n adopt-a-street
4. Review application logs in the backend