Your first docker run

What docker run actually does

docker run is the docker command working engineers type most. It is how you get a Postgres for local development in one line, try a tool without installing it, or reproduce a bug in a clean environment. The classic first command:

docker run hello-world

Step by step, Docker:

1. Looks for an image called hello-world on your machine
2. Doesn't find it, so it pulls (downloads) it from Docker Hub, the default public image registry
3. Creates a new container from the image
4. Starts the container, which prints a greeting
5. The program ends, so the container exits

That last step matters: a container lives exactly as long as the one program it was started to run. When the program stops, the container stops. hello-world prints and quits, so its container is done in under a second.

Anatomy of the command

docker run -it python:3.12 python

Reading it piece by piece:

• docker run: create and start a container
• -it: interactive mode with a terminal attached, so you can type into it
• python:3.12: the image, as name:tag. The tag after the colon picks a version. Omit it and Docker assumes :latest
• python (the last word): the command to run inside the container, overriding the image's default

That one line drops you into a real Python 3.12 prompt inside an isolated container, even if your laptop has no Python installed at all. Type exit() and the container stops.

Pin your tags. python:3.12 today is python:3.12 next year, but latest silently moves, which is the works-on-my-machine bug from lesson 1-1 sneaking back in.