Docker Commands: A Practical Reference for Containers, Images, and Compose (2026)

Plain docker ps lists running containers only, so a stopped one can be absent from the output. I’m interested in how a missing name can point to container state rather than prove no container exists.

Follow the command’s target to the object you need to work with. Let that object’s state guide your next operation.

Docker commands operate on different objects and states

An image is the package Docker uses to create a container. A container is one instance created from that image, while Compose groups services into a project.

Choose the object first, then choose the action that matches its state.

The Docker CLI reference groups commands by resource and lists the shorter forms. Match an action to whether its container exists and is running.

Check the CLI, context, and Docker Engine

The Docker CLI is the client, and Docker Engine handles its requests. A client version alone does not prove the selected Engine is reachable.

docker --version && docker context ls

The asterisk marks the active context, which selects the Engine endpoint for later commands. Run docker version when you need both client and server details, then resolve any connection error before trying container operations.

docker version

If the response names a Unix socket and says permission denied, the CLI cannot access that Engine endpoint. Check the context and the account’s access to the socket instead of reinstalling the client.

Choose Docker commands by the job

The command groups follow the resource they change. An image can produce a container, while Compose manages service containers as a project.

Build, pull, and tag images

Pull an image when a registry holds the starting point, or build one from a Dockerfile in a build context. Docker’s build reference explains that the final dot sends the current directory as context.

The tag names an image reference rather than copying its layers. A build fails if the selected context does not contain the Dockerfile or files the build expects.

Create a container or start one that exists

Docker’s run reference says the command creates and starts a new container, pulling its image if needed. The start reference covers an existing stopped container.

docker run -d --name web -p 8080:80 nginx:alpine
docker ps
docker ps -a

The run flags assign a name and publish host port 8080 to container port 80. Plain docker ps lists running containers, while docker ps -a includes stopped ones so you can find a container before starting it.

The container removal reference limits its volume flag to anonymous volumes, so removing a container does not delete its named volume. If you’re packaging a Node.js service, the Dockerize Node.js application walkthrough applies this image-to-container flow to an app.

Read logs, inspect settings, or run a process

Docker logs reads output captured from a container’s main process, while docker inspect returns detailed configuration and state. The exec reference says it starts a process only while the container’s primary process is running.

docker logs --tail 50 web
docker inspect web
docker exec web nginx -v
docker exec -it web sh

The final command opens an interactive shell because -i keeps input open and -t allocates a terminal. For a shell expression with operators, pass a shell such as sh -c explicitly because exec expects an executable command.

Keep data and services in their own lifecycle

A container’s writable layer belongs to that container, so removing it removes changes stored there. A Docker volume stores data separately and can be mounted by another container.

services:
  web:
    image: nginx:alpine
    ports:
      - "8080:80"
    volumes:
      - web-data:/usr/share/nginx/html
volumes:
  web-data:

Compose treats the file as a project, and the config reference says docker compose config prints resolved service settings without starting containers.

docker compose config

Docker compose up -d creates and starts the project services, while docker compose ps lists their state and docker compose logs reads their output. Docker compose exec adds another process inside a running service.

Diagnose connection and cleanup mistakes

When a command fails, first separate an Engine connection problem from a container-state problem. An inaccessible context affects every Engine operation, while a stopped container prevents exec from starting a process inside it.

Docker system df reports disk use without removing resources. The system prune reference says this command removes unused containers and networks, along with unused images and build cache.

Inspect the confirmation prompt before accepting a prune command.

docker system df
docker system prune

Docker system prune does not remove volumes by default. Add –volumes to prune unused anonymous volumes. The Compose down reference says its volumes option removes the project’s named volumes and attached anonymous volumes, so check for data you need before using it.

Let the container’s state choose the next command

A stopped container still exists, which is why start can reuse it and run cannot. Check the container list before creating another instance from the same image.

docker ps -a

Docker command FAQ

Docker run creates a new container, and docker start reuses a stopped one. Docker exec starts another process only while the container’s primary process is running.