Docker-to-Production Starter Kit

The gap between “it runs in Docker on my laptop” and “it is a deployable service” is about nine specific things. This kit walks all nine, in order, with working files you can paste and adapt.

What you build

#	Section	Time
1	A production-grade multi-stage image — small, cached well, and not running as root	40 min
2	A local stack with a database, healthchecks and dependency ordering that actually works	25 min
3	Configuration and secrets so the same image runs in every environment	20 min
4	A hardening pass: non-root, read-only filesystem, dropped capabilities, resource limits	30 min
5	Signals, graceful shutdown and healthchecks so orchestrators can manage the container	25 min
6	A CI pipeline that builds, scans, tags and pushes	30 min
7	A deploy path and a rollback plan that survives migrations	40 min
8	Logs, metrics, and the debugging moves that work when there is no shell	20 min
9	The 15-point pre-flight checklist	10 min

Real files, not prose

Complete multi-stage Dockerfiles for Node and Python. A compose stack with service_healthy conditions and a failing-fast password check. The hardening block. A graceful-shutdown implementation that fails readiness before it stops accepting connections. A GitHub Actions workflow with Trivy scanning and cache mounts. The expand / migrate / contract migration sequence.

The mistake that causes cascading outages: pointing your liveness probe at the database. The database hiccups, every replica fails liveness, the platform restarts all of them at once, and now you have a cold-start stampede against a database that was already struggling. Liveness checks the process. Readiness checks dependencies. The kit shows both.