Docker size analysis

I often see posts automatically assuming user error in not binding logs etc. to persistent storage as the reason for a large docker volume.

While that can be true, there are many cases when the docker images might be large adding up to the large docker volume. This post is to help definitively answer that.

1. Images, Containers, Volumes...
   If Images are taking up most of the space, then you are running containers with large images, and it is likely not a case of incorrect mapping. Run the following command:

   docker system df

   This is the output from my system:

   TYPE            TOTAL     ACTIVE    SIZE      RECLAIMABLE
   Images          61        57        41.74GB   2.207GB (5%)
   Containers      65        57        4.406GB   82.56kB (0%)
   Local Volumes   525       7         6.082GB   6.081GB (99%)
   Build Cache     119       0         938.1MB   938.1MB

   As you can see, I do have containers that have large images. Lets see what they are

2. List all containers in descending order of image size:

   docker ps --format "{{.ID}}\t{{.Image}}" | while read id image; do   size=$(docker image inspect $image --format='{{.Size}}');   echo -e "$size\t$id\t$image"; done | sort -nr | awk '{printf "%.2f MB\t%s\t%s\n", $1/1024/1024, $2, $3}'

   Here is a sample of output in my system:

   4650.39 MB      8a3dea858fc2    ghcr.io/blakeblackshear/frigate:stable-tensorrt
   3891.40 MB      dfe0cd71485d    sameersbn/gitlab
   3125.77 MB      0ecd52d154a3    ollama/ollama
   2295.23 MB      6479f874c647    ghcr.io/ollama-webui/ollama-webui:main
   2240.48 MB      5e76761352ec    binhex/arch-qbittorrentvpn
   2081.17 MB      62c422a06451    photoprism/photoprism
   2081.17 MB      3e4eca35def8    photoprism/photoprism
   2081.17 MB      3167b1b03e5e    photoprism/photoprism
   1695.60 MB      b8fdf429129b    binhex/arch-lidarr
   1655.56 MB      2af5a08e546e    binhex/arch-delugevpn:latest
   1089.15 MB      a59e198c391b    bokker/unraidapi-re
   931.06 MB       dd9d56275356    linuxserver/nextcloud
   695.54 MB       fca5f682288e    sctx/overseerr:latest
   669.66 MB       a74e31a4cbed    metabase/metabase:latest
   603.31 MB       e7272f88f2fb    linuxserver/sonarr:0.6.1385
   562.54 MB       47f8de7efd63    ghcr.io/esphome/esphome:stable
   532.10 MB       644abfadeb96    rhasspy/wyoming-whisper
   490.14 MB       799d4db50a9e    golift/telegraf
   486.44 MB       73281bc7ba55    ghcr.io/music-assistant/server:latest
   462.53 MB       8d897f0f2dce    grafana/grafana:latest

   Based on this, I now can make an informed decision on do I want to remove/delete any of my containers, or if I have enough resources to not be bothered by it.
   Maybe I can explore if I should bind persistent volume for the tensor models for frigate... I did do that for openvscodeserver, where I was able to bind all binaries to persistent storage

3. Tell me more about the container - do I have containers with volumes unintentionally bound to the docker volume?
   While you can also see this in Unraid UI, if you are someone like me running way too many containers, it might be easier to analyze an output from CLI.
   This is what I used:
    

   docker inspect --format="{{ .Name }} {{ .Mounts }}" $(docker ps -aq)

   Sample output in my system:
    

   /prometheus [{bind  /mnt/cache_appdata/appdata/prometheus/etc /etc/prometheus  rw true rprivate} {bind  /mnt/cache_appdata/appdata/prometheus/data /prometheus  rw true rprivate}]
   /jackettvpn [{volume 9fbfc9969581ff252cfc8d18626e600ae4359baf52834b1be0e5d34d42276234 /var/lib/docker/volumes/9fbfc9969581ff252cfc8d18626e600ae4359baf52834b1be0e5d34d42276234/_data /blackhole local  true } {bind  /mnt/cache_appdata/appdata/jackettvpn /config  rw true rprivate}]
   /QDirStat [{bind  /tmp /ram  rw true rprivate} {bind  /var /var_dir  rw true rslave} {bind  /mnt/cache_appdata/appdata/QDirStat /config  rw true rprivate} {bind  /mnt /storage  ro false rprivate}]
   /crowdsec [{bind  /mnt/cache_appdata/appdata/traefik/crowdsec /etc/crowdsec  rw true rprivate} {bind  /mnt/cache_appdata/appdata/traefik/log /var/log/traefik  ro false rprivate} {bind  /mnt/cache_appdata/appdata/nextcloud/log /var/log/nextcloud  ro false rprivate} {bind  /mnt/cache_appdata/appdata/traefik/crowdsec/log /var/log  rw true rprivate} {bind  /mnt/cache_appdata/appdata/bitwarden/bitwarden.log /var/log/bitwarden/bitwarden.log  ro false rprivate} {bind  /mnt/cache_appdata/appdata/traefik/crowdsec/data /var/lib/crowdsec/data  rw true rprivate}]
   /nextcloud [{bind  /mnt/cache_appdata/appdata/nextcloud /config  rw true rprivate} {bind  /mnt/user/nextcloud /data  rw true rprivate}]

4. Lastly, some basic housekeeping:
   1. Prune dangling images ie. images that are no longer needed because you removed the container:

      docker image prune

      If you want to prune images of any container that is currently stopped/paused:
       

      docker image prune -a

Hope this is helpful!