go-docker-deploy

When to Use

• Containerizing a Go microservice
• Setting up local development with docker-compose
• Deploying to Cloud Run, ECS, or Kubernetes
• Optimizing Docker image size

Critical Patterns

Pattern	Rule
Multi-stage builds	Build in Go image, run in distroless/scratch
One service, one Dockerfile	Each service has its own Dockerfile
docker-compose for dev	Full local stack: services + Postgres + NATS + MinIO
Cloud agnostic deploy	Same image deploys to Cloud Run, ECS, or k8s
Health checks	Every service exposes /health

Dockerfile (Multi-Stage)

Reference: assets/Dockerfile.example

.dockerignore

Reference: assets/dockerignore

docker-compose (Full Local Stack)

Reference: assets/docker-compose.example.yml

Init Script (Multiple Databases)

Reference: assets/init-databases.sql

Makefile (Service Level)

Reference: assets/service.Makefile

Root Makefile

Reference: assets/root.Makefile

Cloud-Agnostic Deployment

The SAME Docker image works on any platform:

Cloud Run (GCP)

Reference: assets/deploy-cloud-run.sh

ECS (AWS)

Reference: assets/ecs-task.json

Kubernetes

Reference: assets/k8s-deployment.yaml

Commands

# Full stack up
docker-compose up -d

# Rebuild single service
docker-compose build booking && docker-compose up -d booking

# View logs
docker-compose logs -f booking

# Run migrations inside container
docker-compose exec booking /migrations/run.sh

# Shell into container (distroless has no shell — use debug variant if needed)
docker-compose exec postgres psql -U postgres -d bastet_booking

Anti-Patterns

Don't	Do
Full Go image in production	Distroless or scratch for minimal attack surface
Cloud-specific entrypoint scripts	Same Dockerfile everywhere, config via env vars
docker-compose in production	docker-compose for dev only, use orchestrator in prod
Shared database containers between services	Each service gets its own database
Hardcoded ports	Configure via environment variables