Disk usage, pruning, log rotation, reclaiming space.

What’s using disk

• Summary: docker system df

• Per-resource breakdown: docker system df -v

Sample output of docker system df:

TYPE            TOTAL     ACTIVE    SIZE      RECLAIMABLE
Images          42        8         12.3GB    9.1GB (74%)
Containers      15        4         287MB     201MB (70%)
Local Volumes   23        6         4.5GB     2.1GB (46%)
Build Cache     128       0         3.2GB     3.2GB

docker system prune

Removes unused containers, networks, dangling images, build cache. By default leaves named volumes alone and only deletes dangling images.

• Interactive: docker system prune

• No prompt: docker system prune -f

• Also all unused images: docker system prune -a

• Also unused volumes (destructive!): docker system prune -a --volumes

Warning: --volumes deletes named volumes’ data. Backup first (see volumes).

Targeted prunes

Containers

• All stopped: docker container prune

• Older than 24h: docker container prune --filter "until=24h"

Images

• Dangling: docker image prune

• All unused: docker image prune -a

• By age (older than 30 days): docker image prune -a --filter "until=720h"

• By label: docker image prune --filter "label!=keep"

Volumes

• Unused: docker volume prune

• Include anonymous (Docker 23+): docker volume prune -a

• By label: docker volume prune --filter "label!=keep"

Networks

• Unused user-defined (built-ins never touched): docker network prune

Build cache

• Default (keep recent): docker builder prune

• Everything: docker builder prune -a

• Limit by size: docker builder prune --keep-storage 5GB

Container logs

json-file driver (default) grows without limit.

Per-container

docker run -d \
  --log-opt max-size=10m \
  --log-opt max-file=3 \
  --name web \
  nginx

Global via daemon.json

/etc/docker/daemon.json:

{
  "log-driver": "json-file",
  "log-opts": {
    "max-size": "10m",
    "max-file": "3"
  }
}

Apply with sudo systemctl restart docker. Existing containers keep old settings until recreated.

Find / truncate log file

• Locate: docker inspect --format '{{.LogPath}}' web

• View size: sudo ls -lh "$(docker inspect --format '{{.LogPath}}' web)"

• Truncate without restarting (don’t rm — daemon keeps FD open): sudo truncate -s 0 "$(docker inspect --format '{{.LogPath}}' web)"

Remove specific resources

• All stopped: docker rm $(docker ps -aq -f status=exited)

• All (force-stops running): docker rm -f $(docker ps -aq)

• All images: docker rmi $(docker images -q)

• All volumes (skips in-use): docker volume rm $(docker volume ls -q)

Reclaim from a specific image

• Containers using image: docker ps -a --filter "ancestor=postgres:15"

• Remove every tag of one repo: docker images "myapp" -q | xargs -r docker rmi -f

Where disk goes

/var/lib/docker/
├── containers/    # per-container metadata, logs, configs
├── image/         # image metadata
├── overlay2/      # image layers + writable container layers (big)
├── volumes/       # named volumes' data
├── network/       # network state
└── buildkit/      # build cache

• Check size: sudo du -sh /var/lib/docker/*

Never delete files directly — go through docker commands. State files in containers/ and image/ get out of sync otherwise.

Scheduled cleanup

Weekly cron entry, prune images older than 168h:

0 4 * * 0 docker system prune -af --filter "until=168h" >/dev/null 2>&1

Quick reference