When a container will not stay up, check the same things in the same order every time.

1. docker ps -a              state and exit code
2. docker logs NAME          what the program said before it exited
3. docker ps -a --format     the command and restart policy it was created with
4. docker run --rm IMAGE ... look inside the image with a throwaway container
5. docker rm -f NAME         remove the broken container
6. docker run ...            run a correct one with the same name and options
7. docker ps / docker logs   confirm it stays up

1. State and exit code

2. Logs

docker logs --tail 20 NAME shows the last attempt. In a restart loop, the same error repeats once per attempt.

3. How it was created

docker ps -a --format 'table {{.Names}}\t{{.Image}}\t{{.Command}}'
docker inspect -f '{{.Path}} {{.Args}}' worker
docker inspect -f '{{.HostConfig.RestartPolicy.Name}} {{.RestartCount}}' worker

docker ps truncates the COMMAND column. The inspect templates print it in full. Chapter 3 covers docker inspect.

4. Look inside the image

docker exec needs a running container, and a crashing one is stopped most of the time. Start a throwaway container from the same image with a command you choose:

docker run --rm IMAGE ls /app
docker run --rm IMAGE cat /app/README.txt

--rm keeps these from piling up as stopped containers.

5-7. Replace it

A container's command, image and options are fixed at creation, so restart cannot fix a wrong command. Remove the container and run a new one with the same name, image and restart policy and the corrected command:

docker rm -f NAME
docker run -d --name NAME --restart POLICY IMAGE [correct command]

Leaving out the command uses the image's default command.

Confirm: docker ps shows Up and the time keeps growing, and docker logs shows a clean start. Clean up anything you created while investigating.