[docker] How to remove old and unused Docker images

When running Docker for a long time, there are a lot of images in system. How can I remove all unused Docker images at once safety to free up the storage?

In addition, I also want to remove images pulled months ago, which have the correct TAG.

So, I'm not asking for removing untagged images only. I'm searching for a way to remove general unused images, which includes both untagged and other images such as pulled months ago with correct TAG.

Update the second (2017-07-08):

Refer (again) to VonC, using the even more recent system prune. The impatient can skip the prompt with the -f, --force option:

docker system prune -f

The impatient and reckless can additionally remove "unused images not just the dangling ones" with the -a, --all option:

docker system prune -af

https://docs.docker.com/engine/reference/commandline/system_prune/

Update:

Refer to VonC's answer which uses the recently added prune commands. Here is the corresponding shell alias convenience:

alias docker-clean=' \
  docker container prune -f ; \
  docker image prune -f ; \
  docker network prune -f ; \
  docker volume prune -f '

Old answer:

Delete stopped (exited) containers:

$ docker ps --no-trunc -aqf "status=exited" | xargs docker rm

Delete unused (dangling) images:

$ docker images --no-trunc -aqf "dangling=true" | xargs docker rmi

If you have exercised extreme caution with regard to irrevocable data loss, then you can delete unused (dangling) volumes (v1.9 and up):

$ docker volume ls -qf "dangling=true" | xargs docker volume rm

Here they are in a convenient shell alias:

alias docker-clean=' \
  docker ps --no-trunc -aqf "status=exited" | xargs docker rm ; \
  docker images --no-trunc -aqf "dangling=true" | xargs docker rmi ; \
  docker volume ls -qf "dangling=true" | xargs docker volume rm'

References:

• docker ps -f
• docker rm
• docker images -f
• docker rmi
• Docker v1.9.0 release notes
• docker volume ls
• docker volume rm

I recently wrote a script to solve this on one of my servers:

#!/bin/bash

# Remove all the dangling images
DANGLING_IMAGES=$(docker images -qf "dangling=true")
if [[ -n $DANGLING_IMAGES ]]; then
    docker rmi "$DANGLING_IMAGES"
fi

# Get all the images currently in use
USED_IMAGES=($( \
    docker ps -a --format '{{.Image}}' | \
    sort -u | \
    uniq | \
    awk -F ':' '$2{print $1":"$2}!$2{print $1":latest"}' \
))

# Get all the images currently available
ALL_IMAGES=($( \
    docker images --format '{{.Repository}}:{{.Tag}}' | \
    sort -u \
))

# Remove the unused images
for i in "${ALL_IMAGES[@]}"; do
    UNUSED=true
    for j in "${USED_IMAGES[@]}"; do
        if [[ "$i" == "$j" ]]; then
            UNUSED=false
        fi
    done
    if [[ "$UNUSED" == true ]]; then
        docker rmi "$i"
    fi
done

If you want to remove images pulled X months ago, you can try the below example which remove images created three months ago:

three_months_old_images=`docker images | grep -vi "<none>" | tr -s ' ' | cut -d" " -f3,4,5,6 | grep "3 months ago" | cut -d" " -f1`
docker rmi $three_months_old_images

I usually do docker rm -f $(docker ps -a -q) and docker system prune to purge all dangling containers.

The other answers are great, specifically:

docker system prune # doesn't clean out old images
docker system prune --all # cleans out too much

But I needed something in the middle of the two commands so the filter option was what I needed:

docker image prune --all --filter "until=4320h" # delete images older than 6 months ago; 4320h = 24 hour/day * 30 days/month * 6 months

Hope that helps :)

For reference: https://docs.docker.com/config/pruning/#prune-images

If you wish to automatically/periodically clean up exited containers and remove images and volumes that aren't in use by a running container you can download the image meltwater/docker-cleanup.

Just run:

docker run -d -v /var/run/docker.sock:/var/run/docker.sock:rw  -v /var/lib/docker:/var/lib/docker:rw --restart=unless-stopped meltwater/docker-cleanup:latest

It runs every 30 minutes by default. You can however set the delay time by using this flag in seconds (DELAY_TIME=1800 option).

More details: https://github.com/meltwater/docker-cleanup/blob/master/README.md

To prune all images and volumes as well
docker system prune -af --volumes

Assuming you have Docker 1.13 or higher you can just use the prune commands. For your question specifically for removing old images, you want the first one.

# Remove unused images
docker image prune

# Remove stopped containers.
docker container prune

# Remove unused volumes
docker volume prune

# Remove unused networks
docker network prune

# Command to run all prunes:
docker system prune

I would recommend not getting used to using the docker system prune command. I reckon users will accidentally remove things they don't mean to. Personally, I'm going to mainly be using the docker image prune and docker container prune commands.

@VonC already gave a very nice answer, but for completeness here is a little script I have been using---and which also nukes any errand Docker processes should you have some:

#!/bin/bash

imgs=$(docker images | awk '/<none>/ { print $3 }')
if [ "${imgs}" != "" ]; then
   echo docker rmi ${imgs}
   docker rmi ${imgs}
else
   echo "No images to remove"
fi

procs=$(docker ps -a -q --no-trunc)
if [ "${procs}" != "" ]; then
   echo docker rm ${procs}
   docker rm ${procs}
else
   echo "No processes to purge"
fi

There is sparrow plugin docker-remove-dangling-images you can use to clean up stopped containers and unused (dangling) images:

$ sparrow plg run docker-remove-dangling-images

It works both for Linux and Windows OS.

docker system prune -a

(You'll be asked to confirm the command. Use -f to force run, if you know what you're doing.)

To remove tagged images which have not container running, you will have to use a little script:

#!/bin/bash

# remove not running containers
docker rm $(docker ps -f "status=exited" -q)

declare -A used_images

# collect images which has running container
for image in $(docker ps | awk 'NR>1 {print $2;}'); do
    id=$(docker inspect --format="{{.Id}}" $image);
    used_images[$id]=$image;
done

# loop over images, delete those without a container
for id in $(docker images --no-trunc -q); do
    if [ -z ${used_images[$id]} ]; then
        echo "images is NOT in use: $id"
        docker rmi $id
    else
        echo "images is in use:     ${used_images[$id]}"
    fi
done

Here is a script to clean up Docker images and reclaim the space.

#!/bin/bash -x
## Removing stopped container
docker ps -a | grep Exited | awk '{print $1}' | xargs docker rm

## If you do not want to remove all container you can have filter for days and weeks old like below
#docker ps -a | grep Exited | grep "days ago" | awk '{print $1}' | xargs docker rm
#docker ps -a | grep Exited | grep "weeks ago" | awk '{print $1}' | xargs docker rm

## Removing Dangling images
## There are the layers images which are being created during building a Docker image. This is a great way to recover the spaces used by old and unused layers.

docker rmi $(docker images -f "dangling=true" -q)

## Removing images of perticular pattern For example
## Here I am removing images which has a SNAPSHOT with it.

docker rmi $(docker images | grep SNAPSHOT | awk '{print $3}')

## Removing weeks old images

docker images | grep "weeks ago" | awk '{print $3}' | xargs docker rmi

## Similarly you can remove days, months old images too.

Original script

https://github.com/vishalvsh1/docker-image-cleanup

Usually Docker keeps all temporary files related to image building and layers at

/var/lib/docker

This path is local to the system, usually at THE root partition, "/".

You can mount a bigger disk space and move the content of /var/lib/docker to the new mount location and make a symbolic link.

This way, even if Docker images occupy space, it will not affect your system as it will be using some other mount location.

Original post: Manage Docker images on local disk

Until now (Docker version 1.12) we are using the following command to delete all the running containers. Also, if we want to delete the volumes, we can do that manually using its respective tag -v in the following command.

Delete all Exited Containers

docker rm $(docker ps -q -f status=exited)

Delete all Stopped Containers

docker rm $(docker ps -a -q)

Delete All Running and Stopped Containers

docker stop $(docker ps -a -q)
docker rm $(docker ps -a -q)

Remove all containers, without any criteria

docker container rm $(docker container ps -aq)

But, in version 1.13 and above, for complete system and cleanup, we can directly user the following command:

docker system prune

All unused containers, images, networks and volumes will get deleted. We can also do this using the following commands that clean up the individual components:

docker container prune
docker image prune
docker network prune
docker volume prune

If you build these pruned images yourself (from some other, older base images) please be careful with the accepted solutions above based on docker image prune, as the command is blunt and will try to remove also all dependencies required by your latest images (the command should be probably renamed to docker image*s* prune).

The solution I came up for my docker image build pipelines (where there are daily builds and tags=dates are in the YYYYMMDD format) is this:

# carefully narrow down the image to be deleted (to avoid removing useful static stuff like base images)
my_deleted_image=mirekphd/ml-cpu-py37-vsc-cust

# define the monitored image (tested for obsolescence), which will be usually the same as deleted one, unless deleting some very infrequently built image which requires a separate "clock"
monitored_image=mirekphd/ml-cache

# calculate the oldest acceptable tag (date)
date_week_ago=$(date -d "last week" '+%Y%m%d')

# get the IDs of obsolete tags of our deleted image
# note we use monitored_image to test for obsolescence
my_deleted_image_obsolete_tag_ids=$(docker images --filter="before=$monitored_image:$date_week_ago" | grep $my_deleted_image | awk '{print $3}')

# remove the obsolete tags of the deleted image
# (note it typically has to be forced using -f switch)
docker rmi -f $my_deleted_image_obsolete_tag_ids

How to remove a tagged image

1. docker rmi the tag first

2. docker rmi the image.

   # that can be done in one docker rmi call e.g.: # docker rmi <repo:tag> <imageid>

(this works Nov 2016, Docker version 1.12.2)

e.g.

$ docker images 
REPOSITORY              TAG                 IMAGE ID            CREATED             SIZE
usrxx/the-application   16112805            011fd5bf45a2        12 hours ago        5.753 GB
usryy/the-application   vx.xx.xx            5af809583b9c        3 days ago          5.743 GB
usrzz/the-application   vx.xx.xx            eef00ce9b81f        10 days ago         5.747 GB
usrAA/the-application   vx.xx.xx            422ba91c71bb        3 weeks ago         5.722 GB
usrBB/the-application   v1.00.18            a877aec95006        3 months ago        5.589 GB

$ docker rmi usrxx/the-application:16112805 && docker rmi 011fd5bf45a2
$ docker rmi usryy/the-application:vx.xx.xx && docker rmi 5af809583b9c
$ docker rmi usrzz/the-application:vx.xx.xx eef00ce9b81f
$ docker rmi usrAA/the-application:vx.xx.xx 422ba91c71bb
$ docker rmi usrBB/the-application:v1.00.18 a877aec95006

e.g. Scripted remove anything older than 2 weeks.

IMAGESINFO=$(docker images --no-trunc --format '{{.ID}} {{.Repository}} {{.Tag}} {{.CreatedSince}}' |grep -E " (weeks|months|years)")
TAGS=$(echo "$IMAGESINFO" | awk '{ print $2 ":" $3 }' )
IDS=$(echo "$IMAGESINFO" | awk '{ print $1 }' )
echo remove old images TAGS=$TAGS IDS=$IDS
for t in $TAGS; do docker rmi $t; done
for i in $IDS; do docker rmi $i; done

The following command will delete images older than 48 hours.

$ docker image prune --all --filter until=48h

I'm using this command:

export BEFORE_DATETIME=$(date --date='10 weeks ago' +"%Y-%m-%dT%H:%M:%S.%NZ")
docker images -q | while read IMAGE_ID; do
    export IMAGE_CTIME=$(docker inspect --format='{{.Created}}' --type=image ${IMAGE_ID})
    if [[ "${BEFORE_DATETIME}" > "${IMAGE_CTIME}" ]]; then
        echo "Removing ${IMAGE_ID}, ${BEFORE_DATETIME} is earlier then ${IMAGE_CTIME}"
        docker rmi -f ${IMAGE_ID};
    fi;
done

This will remove all images whose creation time is greater than 10 weeks ago.

This worked for me:

docker rmi $(docker images | grep "^<none>" | awk "{print $3}")

If you have a lot of them, it can be really tedious to remove them, but lucky for us Docker has a few commands to help us eliminate dangling images. In older versions of Docker (and this still works today), you can delete dangling images on their own by running docker rmi -f $(docker images -f "dangling=true" -q) .

docker rm `docker ps -aq`

or

docker rm $(docker ps -q -f status=exited)

First, run docker images to see list of images and copy IMAGE HASH ID into clipboard.

Run docker rmi -f <Image>

Remember option -f is force deleting.

To remove old tagged images that are more than a month old:

$ docker images --no-trunc --format '{{.ID}} {{.CreatedSince}}' \
    | grep ' months' | awk '{ print $1 }' \
    | xargs --no-run-if-empty docker rmi

Note that it'll fail to remove images that are used by a container, referenced in a repository, has dependent child images... which is probably what you want. Else just add -f flag.

Example of /etc/cron.daily/docker-gc script:

#!/bin/sh -e

# Delete all stopped containers (including data-only containers).
docker ps -a -q --no-trunc --filter "status=exited" | xargs --no-run-if-empty docker rm -v

# Delete all tagged images more than a month old
# (will fail to remove images still used).
docker images --no-trunc --format '{{.ID}} {{.CreatedSince}}' | grep ' months' | awk '{ print $1 }' | xargs --no-run-if-empty docker rmi || true

# Delete all 'untagged/dangling' (<none>) images
# Those are used for Docker caching mechanism.
docker images -q --no-trunc --filter dangling=true | xargs --no-run-if-empty docker rmi

# Delete all dangling volumes.
docker volume ls -qf dangling=true | xargs --no-run-if-empty docker volume rm

Occasionally I have run into issues where Docker will allocate and continue to use disk space, even when the space is not allocated to any particular image or existing container. The latest way I generated this issue accidentally was using "docker-engine" centos build instead of "docker" in RHEL 7.1. What seems to happen is sometimes the container clean-ups are not completed successfully and then the space is never reused. When the 80GB drive I allocated as / was filled with /var/lib/docker files I had to come up with a creative way to resolve the issue.

Here is what I came up with. First to resolve the disk full error:

1. Stop docker: systemctl stop docker

2. Allocated a new drive mounted as say /mnt/docker .

3. Move all the files in /var/lib/docker to /mnt/docker . I used the command:

   rsync -aPHSx --remove-source-files /var/lib/docker/ /mnt/docker/
   

4. Mount the new drive to /var/lib/docker.

At this point I no longer had a disk full error, but I was still wasting a huge amount of space. The next steps are to take care of that.

1. Start Docker: systemctl start docker

2. Save the all the images:

   docker save $(docker images |sed -e '/^<none>/d' -e '/^REPOSITORY/d' -e 's,[ ][ ]*,:,' -e 's,[ ].*,,') > /root/docker.img
   

3. Uninstall docker.

4. Erase everything in /var/lib/docker:

   rm -rf /var/lib/docker/[cdintv]*
   

5. Reinstall docker

6. Enable docker: systemctl enable docker

7. Start docker: systemctl start docker

8. Restore images:

   docker load < /root/docker.img
   

9. Start any persistent containers you need running.

This dropped my disk usage from 67 GB for docker to 6 GB for docker.

I do not recommend this for everyday use. But it is useful to run when it looks like docker has lost track of used disk space do to software errors, or unexpected reboots.

docker rm $(docker ps -faq)
docker rmi $(docker ps -faq)

-f force

-a all

-q in the mode

Remove old containers weeks ago.

docker rm $(docker ps -a | grep "weeks" | awk '{ print $1; }')

Remove old images weeks ago. Be careful. This will remove base images which was created weeks ago but which your new images might be using.

docker rmi $(docker images | grep 'weeks' | awk '{ print $3; }')