Where do half baked containers live and how to delete them?

Product Support Self-hosting

AI-generated summary

User helmi is experiencing issues with a failed Docker bootstrap on a test installation. They’re trying to delete the leftover data, but Docker doesn’t recognize the container. They’ve found data in /var/lib/docker/aufs/, but can’t identify which hashed subdirectories belong to the failed container. cro suggests using docker images to view local Docker images and docker rmi to remove unwanted images.

1

Hi,

i’m currently just playing around on a test installation trying to warm up with docker, ruby, discourse et al.

My current problem is that when bootrapping fails and probably things getting really broken i’d like to complete kill all data that’s been there so far without killing other containers. Problem is that docker doesn’t recognize the container from the failed bootstrap yet but the data seems to be there and doesn’t get downloaded.

This may not happen in normal circumstances but i’m really trying to break things at the moment to find solutions for exactly these cases.

As far as i know so far there is data in /var/lib/docker/aufs/ (in mnt and diff) but i can’t tell from the launchers output which of the hashed subdirectories is responsible for the container i’m currently trying to build.

Any general chance to get rid of this data as long as no container really extists (at least docker ps and docker ps -a doesn’t list it).

Thanks for helping,
Frank

2

Use ‘docker images’ to see what your local docker install thinks is there. ‘Docker rmi’ will remove unwanted images (as opposed to containers).

–cro