Docker Compose for Local Spring Boot + PostgreSQL + Redis Dev

Docker Compose lets you define a multi-container application as a single YAML file and manage the entire stack with one command. For local development, it replaces running three separate Docker commands and manually creating networks.

Service definition and networking

• Each service in docker-compose.yml maps to one container; Compose creates a project-specific bridge network automatically — services resolve each other by name, so your Spring Boot app connects to jdbc:postgresql://postgres:5432/mydb by the service name postgres, not by IP
• The depends_on directive controls startup order but not readiness — depends_on: [postgres] ensures the Postgres container starts before Spring Boot, but Postgres is not necessarily accepting connections yet; add a healthcheck to the postgres service and use condition: service_healthy to wait for actual readiness
• Without health check conditions, Spring Boot starts before Postgres is ready and fails with connection refused — the most common first-time Docker Compose debugging problem

Environment variables and secrets

• Define credentials in a .env file and reference them as ${POSTGRES_PASSWORD} in the YAML — Compose reads .env automatically and substitutes values; never hardcode passwords directly in the YAML
• Add .env to .gitignore immediately — it is the most common way credentials leak into public repositories; commit a .env.example with placeholder values instead
• Production does not use Docker Compose; secrets in production go through the orchestrator's secret management (Kubernetes Secrets, AWS Secrets Manager) — Docker Compose is for local development and integration testing

Volume persistence

• Named volumes (volumes: postgres_data: {}) survive docker compose down; bind mounts do not — use named volumes for database data directories so a development database is not lost on every stack restart
• docker compose down -v removes named volumes explicitly — useful when you need a clean database state, but easy to confuse with bare docker compose down which keeps the data