docker-compose — Multi-Container Development

Docker solved our “works on my machine” problem. But running five containers manually with twenty-line docker run commands? docker-compose (then still called Fig) brought an elegant solution — one YAML file, one command, the entire stack.

Our Development Stack¶

version: '2'
services:
  app:
    build: .
    ports:
      - "8080:8080"
    depends_on:
      - db
      - redis
    environment:
      - DB_HOST=db
  db:
    image: postgres:9.4
    environment:
      - POSTGRES_DB=myapp
      - POSTGRES_PASSWORD=dev
    volumes:
      - pgdata:/var/lib/postgresql/data
  redis:
    image: redis:2.8
  rabbitmq:
    image: rabbitmq:3-management
    ports:
      - "15672:15672"
volumes:
  pgdata:

Advantages Over Vagrant¶

Service isolation, speed (seconds vs. minutes), less memory, closer to production. docker-compose creates an isolated network — services see each other by the names in the YAML.

Configuration and Tips¶

.env file for variables. docker-compose.override.yml for local customization (in .gitignore). Use depends_on, but implement retry logic in the application. Named volumes for persistent data.

docker-compose Is a Game Changer¶

One file defines the entire infrastructure. Versioned, reviewed, shared in Git. If you’re using Docker, docker-compose is the logical next step.