Ignoring the restart policy for running on-off commands on the same docker image used for long-running services

Question

I am using the option to restart my docker instances in my docker-compose file like:

restart: always

The problem is that sometimes I run a single docker container for maintenance work like:

docker-compose run rails rake db:migrate 

The issue with this is when I do a 'docker ps' I can see those on-off commands are still running and constantly being restarted:

"rake db:migrate" 2 days ago Restarting (7) 18 seconds ago

Is there a way to run a docker image that is for on-off purposes, but still have the restart policy on it but somehow ignore it for this single instance usage?

Answer 1

Once the docker image is running (after docker-compose run), you could amend it, using docker update:

docker update --restart=no <MY-CONTAINER-ID>

That would prevent said container to restart when you stop it.
See restart policies.

Answer 2

Just create another service in your docker-compose.yml file that uses the same docker image, but sets the restart to restart: "no".

Dummy Example:

version: "2.1"

services:

  rake_web:
    image: rake-image-name
    restart: always
    networks:
      - rake_network

  rake_cli:
    image: rake-image-name
    restart: "no"
    networks:
      - rake_network

networks:
  rake_network:
    driver: "bridge"

Now instead of:

docker-compose run rake_web some-command

You use:

docker-compose run rake_cli some-command

Once rake_cli is using the same docker image, but with the restart policy disabled your container for rake_cli will not restart after you have run your command on it.