updated
No space left. Do not nuke volumes.
When dc-up fails with ENOSPC, run dc-df then dc-prune --yes. Cache/images/nets are engine-wide. Never docker system prune -af --volumes.
When dc-up fails with no space left / ENOSPC:
dc-df # images / cache / volumes + Colima guest df
dc-prune # dry-run
dc-prune --yes # safe set only (same as TUI P when CRITICAL)
dc-up # retry
| Flag | Risk |
|---|---|
dc-prune --yes |
Engine-wide build cache, dangling images, unused nets; owned-only orphan dc-forward sidecars (target proven absent) |
dc-prune --all --yes |
Engine-wide unused tagged images; parked stacks rebuild on next dc-up |
dc-prune --volume NAME --yes |
High / owned-only — named volume data (DBs). One name only; never bulk; mount inventory must succeed |
docker system prune -af --volumes |
Do not — not wrapped; destroys named volumes |
If Colima guest / is still ~100% after prune, the VM disk cap is full: dc-prune --colima-hint. Prune does not grow the qemu/VZ image.
Named volumes are listed by dc-df --volumes but not auto-deleted.
See stuck if prune is not enough.
Broke something?report it