failed to register layer: no space left on device

TL;DR: The pull downloaded fine but there is nowhere to unpack it. Run docker system prune -a to delete unused images (the layers of images you are not running are the usual hog), then pull again. Verify with docker system df before and after.

The error

failed to register layer: Error processing tar file(exit status 1): write /usr/lib/x86_64-linux-gnu/libc.so.6: no space left on device

Fix it

1. Check docker disk usage:

docker system df Expected: Images row large with high reclaimable.

1. Remove unused images:

docker image prune -a Expected: deletes images not referenced by any container.

1. Retry the pull:

docker pull [image] Expected: layers register cleanly.

When this applies

• docker pull failing at the extract step
• Hosts that pull many image versions over time

When this does NOT apply

• Build failures (build temp space, separate skill)
• "no space" on the HOST root from non-docker files

Versions

All Docker versions.

Why it happens

Pulls download compressed layers then extract them, transiently needing roughly 2x the image size. Registration writes into /var/lib/docker; when it is full the tar extraction aborts mid-layer.

Edge cases

• Dangling build cache (docker builder prune) can dwarf image usage on build hosts.
• docker pull of a huge image on a small disk may never fit; use a slimmer tag (alpine/slim variants).
• Thin-provisioned VM disks: the guest sees free space the host already reclaimed; expand the virtual disk.