Docker Engine Containers Runtime

Container reports 'Up' but the app inside crashed and is no longer serving HTTP. Load balancer keeps sending traffic.

Add HEALTHCHECK --interval=30s --timeout=5s --start-period=10s --retries=3 CMD curl -fsS http://localhost:8080/health || exit 1 in the Dockerfile (or healthcheck in compose). 'docker ps' status becomes 'healthy' / 'unhealthy'. Load balancer and depends_on consumers gate on health.

DB container uses '-v /home/user/db:/var/lib/postgres' — host path. After host upgrade, the host path changes and DB data appears empty.

Use named volume: '-v pgdata:/var/lib/postgres' with 'docker volume create pgdata'. Engine manages storage path. Stable across host upgrades. Bind mounts are appropriate for source-code dev where host editability matters.

Long-running service crashes on transient downstream errors. Currently started with 'docker run' and no --restart.

Use --restart unless-stopped (auto-restart on crash; respects explicit stop) or --restart on-failure:5 (cap retries). --restart always restarts even after explicit stop on next daemon boot. Pair with HEALTHCHECK so the engine restart isn't masked by a hung process.