Kill all those damn left over docker-compose deployments

Have you ever booted your machine and discovered some left over docker containers?

You can easily stop them all with:

docker stop `docker ps -a -q`

But then, you check again… and they are back up again!

That’s probably because they are part of docker-compose deployments with restart: always .

Of course in production, you are happy when a docker-compose deployment starts automatically on boot. However on your dev machine, it’s pretty much annoying.

Luckily I have a one liner for you to try out:

docker inspect `docker ps -a -q` | grep working_dir | awk '{ print $2 }' |  sed -r 's/[",]+//g' | sort | uniq | while read p; do pushd $p && docker-compose down && popd; done

Let’s see what we have here:

docker inspect `docker ps -a -q`

Runs docker inspect for each container id.

grep working_dir | awk '{ print $2 }'

Filters out the working-dir key and prints out it’s value.

sed -r 's/[",]+//g'

Cleans the value, because it shows up with “ and , .

At this point, we have a list of working directories for each container. So, we might have some directories showing up more than once.

sort | uniq 

This prints out only the unique directories (every directory shows once).

And finally:

while read p; do pushd $p && docker-compose down && popd; done

Iterate over the directories, go in to each one and run docker-compose down .

So that’s it!

Now your containers aren’t using up all your precious CPU and RAM.