Docker's various caches, unused images, and stopped containers can consume substantial disk space over time. This guide covers systematically cleaning up without accidentally removing something you actually need.
Checking Docker's Current Disk Usage
docker system dfBreaks down disk usage by images, containers, volumes, and build cache — identifies which category is consuming the most space before deciding what to clean up.
Removing Stopped Containers
docker container pruneRemoves all stopped containers — safe for containers you're genuinely done with, though verify nothing important is stopped-but-intended-to-restart before running this.
Removing Unused Images
docker image pruneRemoves "dangling" images (untagged layers, typically leftover from rebuilds) — the safest image cleanup, unlikely to remove anything you're actually using.
Removing ALL Unused Images (More Aggressive)
docker image prune -aRemoves any image not currently used by a running container, not just dangling ones — more aggressive; verify you don't need any currently-stopped-container's image before running this.
Removing Unused Volumes
docker volume pruneCaution: removes volumes not referenced by any container — if a volume contains important data from a container you've temporarily stopped, this would delete that data; always verify carefully before running.
Clearing Build Cache
docker builder pruneRemoves cached build layers — safe to clear, since it just means future builds are slightly slower (rebuilding cache) rather than any actual data loss.
The Comprehensive Cleanup Command
docker system prune -a --volumesRemoves stopped containers, unused networks, all unused images, and unused volumes in one command — the most aggressive option; use with real caution and understanding of exactly what it removes.
Always Verify Before Aggressive Cleanup
docker system df -vShows detailed breakdown including specific image/volume names and sizes — review this before running aggressive prune commands, especially --volumes, to confirm you're not about to remove something important.
Setting Up Regular, Safe Automated Cleanup
0 3 * * 0 docker container prune -f && docker image prune -f && docker builder prune -fA weekly cron job for the safer cleanup operations (excluding volumes, which carry genuine data-loss risk if automated without careful review) — keeps routine buildup in check without risking accidental data loss.
Identifying Specific Large Images
docker images --format "table {{.Repository}}\t{{.Tag}}\t{{.Size}}" | sort -k3 -h -rSorted by size, helping identify specific large images worth investigating — sometimes a single overly-large image (not optimized with multi-stage builds, see Docker Multi-Stage Builds: Reducing Image Size) is the primary disk usage culprit.
Checking for Orphaned Compose Project Resources
docker compose down --volumes --remove-orphansCleans up a specific compose project's resources including volumes — useful when decommissioning an entire application stack, though again, verify volume removal is genuinely intended.
Common Errors
Accidentally deleted needed volume data — if caught immediately and the underlying storage hasn't been overwritten, data recovery tools might help, but this isn't reliable; the real lesson is always reviewing carefully before volume-affecting prune commands, and maintaining proper backups regardless.
Continue Reading
- Docker Volumes: How to Persist Data Beyond a Container's Lifecycle
- Docker Multi-Stage Builds: Reducing Image Size
- How to Clean Up Disk Space on a Linux VPS (Safe Practices)
Browse more articles in Docker & Containers.
