Docker for developers who want the same build on every machine

Docker's promise is simple: build once, run the same thing everywhere. In practice images grow to a gigabyte, builds break when a base image updates underneath you, and someone bakes an API key into a layer that ends up in a public registry.

This section treats Docker as a deployment tool rather than a novelty. We look at writing Dockerfiles that build fast and small, multi-stage builds, pinning base image versions, Docker Compose for local and single-server setups, health checks, and passing configuration at runtime instead of at build time. We'll also cover what to scan for before an image goes anywhere near production.

We assume you can run docker build. We don't assume you've had to debug why a container works locally and exits instantly on the server. That's usually where the useful lessons are.