[Request]: Storage - Add `container system prune` command to remove unused resources

[realrajaryan]

Feature or enhancement request details

Add a container system prune subcommand to remove unused images, stopped containers, and dangling volumes in a single operation. This provides a convenient way for users to reclaim disk space.

Some basic requirements:

• Remove all stopped containers
• Remove dangling images (untagged and not referenced by any container)
• Remove unused volumes (not mounted by any container)
• Support --all flag to remove all unused images (not just dangling)
• Support --volumes flag to include volume cleanup
• Show what will be deleted and prompt for confirmation (unless --force is used)
• Automatically trigger fstrim after cleanup to actually reclaim disk space (related - [Request]: Storage - Automatically reclaim unused disk space from containers with minimal user interaction containerization#414)

Example usage:

container system prune     # Remove stopped containers and dangling images
container system prune --all  # Remove all unused images
container system prune --volumes --force # Include volumes, skip confirmation

Example output:

Pruned containers:
abc123def456
789ghi012jkl

Pruned images:
sha256:abcd1234...
sha256:efgh5678...

Reclaimed 8.5GB in disk space

Some things to discuss

• Filter options (e.g., --filter "until=24h" to prune resources older than 24h)
• Whether to automatically run fstrim inline vs. async (See [Request]: Storage - Automatically reclaim unused disk space from containers with minimal user interaction containerization#414)
• Dry-run mode to preview what would be deleted

Code of Conduct

• I agree to follow this project's Code of Conduct

[PhantomInTheWire]

I can take a look at this as well. I recently had to manually clean up a lot on my own device to free up storage, so I’m pretty motivated to get this done.

[tico88612]

Hi @PhantomInTheWire, are you still working on this?

[PhantomInTheWire]

@tico88612 yes i am, i have a local branch ready but there are some edge cases that im unsure of, if you want to help out i'll open a draft pr soon and we can collaborate if you like(i'll make sure to ping you when that happens)

[tico88612]

Hi @realrajaryan

I'm thinking of extracting the core prune logic from each command (container, image, volume, network) into the ContainerAPIClient module as Client*+Prune.swift extensions, so that system prune can compose them without duplicating the filtering/deletion logic. The CLI commands would then become thin wrappers that only handle argument parsing and output formatting. Does this direction sound reasonable?

I have some design questions and wanted to ping to get your thoughts. cc @jglogan @saehejkang

Because I use container myself, I hope this feature can be implemented as soon as possible.

[saehejkang]

Does this direction sound reasonable?

Sounds reasonable to me and extracting logic that overlaps across resources into the client layer makes sense. Honestly, I see this growing a bit/morphing once the ManagedResource work is completed. All the resources would follow the same pattern and then the client logic could be even more simplified.

A draft PR implementing this for a single resource could be a great way to get the ball rolling. That being said, it still might be best to wait for feedback from others on the direction you’d like to take.

[tico88612]

I will try to implement this in 2 weeks.