It is an extremely common problem for Docker users to run out of disk space due to unused containers and images not being removed. In development, this can be an annoyance and confuse new users. In testing and production, it can be a critical problem. Anecdotally, we have heard of several users who have production systems go down due to Docker filling up their disks.

Docker should automatically remove images where it can, and provide easy-to-use tooling to remove images manually where users may not Docker to automatically delete data.

If we choose not have to any functionality in Docker itself, the common patterns should be clearly documented (e.g. docker rmi --help should suggest doing docker rmi $(docker images -aq)).

Possible solutions

• The concept of "anonymous images" could be removed, and Docker could remove danging images automatically or manually
• A command could be added to manually remove images which aren't used by any containers
• A command could be added to remove all stopped containers, or documentation could be improved to help you do this.
• Docker itself could automatically remove containers and images after a certain amount of time, ala https://github.com/rancher/sherdock

Related

• docker/docker: Implement a 'clean' command
• docker/compose: Clean orphan containers
• docker/compose: Epic: Use less disk space
• http://stackoverflow.com/questions/17236796/how-to-remove-old-docker-containers/24277052
• docker/docker: Image Graph operations are racey - race conditions in graph may be a blocker to doing this reliably

(Maintainers: Consider this a wiki to keep track of everything related to this issue... feel free to edit the description to add/remove/modify stuff.)