We are using docker to build and push images to repositories.
Is there a 'smart' way to clean/remove docker objects (images, layers, volumes etc.) but not remove cache files (we need to preserve cache so our builds takes as low time as possible)?
We are already:
removing images that were pushed to repo (couse we do not need them on building machine) - docker rmi <already_pushed_image>
Is there something more we can do? We could also have scheduled job with docker rmi $(docker images -q -f dangling=true), but that would also remove cache from our builds?
Related
I have a dockerfile which when I want to build results in the error
E: You don't have enough free space in /var/cache/apt/archives/
Note that the image sets up a somewhat complex project with several dependencies that require quite a lot of space. For example, the list includes Qt. This is only a thing during the construction of the image, and in the end, I expect it to have a size of maybe 300 MB.
Now I found this: https://unix.stackexchange.com/questions/578536/how-to-fix-e-you-dont-have-enough-free-space-in-var-cache-apt-archives
Given that, what I tried so far is:
Freeing the space used by docker images so far by calling docker system prune
Removing unneeded installation files by calling sudo apt autoremove and sudo apt autoclean
There was also the suggestion to remove data in var/log, which has currently a size of 3 GB. However, I am not the system administrator and thus wary to do such a thing.
Is there any other way to increase that space?
And, preferably, is there a more sustainable solution, allowing me to build several images without having to search for spots where I can clean up the system?
Try this suggestion. You might have a lot of unused images that need to be deleted.
https://github.com/onyx-platform/onyx-starter/issues/5#issuecomment-276562225
Converting a #Dre suggestion into the code, you might want to use Docker prune command for containers, images & volumes
docker container prune
docker image prune
docker volume prune
You can use these commands in sequence:
docker container prune; docker image prune; docker volume prune
Free Space without removing your latest images
Use the following command to see the different types of reclaimable storage (the -v verbose option provides more detail):
docker system df
docker system df -v
Clear the build cache (the -a option will remove unused build cache):
docker builder prune -a
Remove dangling images ( tagged images, old and previous image builds):
docker rmi -f $(docker images -f "dangling=true" -q)
Increase Disk image size using Docker UI
Docker > Preferences > Resources > Advanced > adjust Disk image size > Apply & Restart
TLDR;
run
docker system prune -a --volumes
I tried to increase the disk space and prune the images, containers and volumes manually but was facing the issue again and again. When I tried to check the memory consumption on my machine, I found a lot of memory consumed by ~/Library/Containers/com.docker.docker location. Did a system prune which cleaned up a lot of space and docker builds started working again.
On docker build from jenkins 1st docker images are created tagged properly as LATEST on remote server.
On rebuild it supposed to overwrite docker image on server. But it didn't.
Virtually it creates new docker images with no-tag repository and no-tag tag. And avoid using predefined domain name for image.
So creates a new because it's supposed create NEW image because they are totally different.
Is there any way to avoid just deleting image directly from the remote server? But to update image docker on same tag(domain) name?
Any ideas on workaround? How to avoid making new docker images with static unchanged TAG name from jenkins build.
Because it eats a lot of memory space on the moment when run on cron.
How can I overwrite docker image or make a facade that it's overwrited?
You can delete no-tag images after build image from jenkins by: docker rmi $(docker images -f "dangling=true" -q)
here is my input and output:
shshenhx#shshenhx:~/Desktop/Docker$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
python latest 336d482502ab 4 days ago 692 MB
shshenhx#shshenhx:~/Desktop/Docker$ docker rmi 336
Untagged: python:latest
Untagged: python#sha256:bf0718e2882cabc144c9c07d758b09c65efc104a6ddc72a9a976f8b26f67c2ee
Deleted: sha256:336d482502ab564b0b2964b2ed037529ba40c7b4ade2117ca11d74edbf11f99e
Deleted: sha256:1e2f72b0bf844de7bfe57a32115b660770355843ef8078fb55629e720147e6a9
Deleted: sha256:b5818ba96f33835318f3e9d7b4094e1007be247f04ab931ea9ef83b45b451f90
Deleted: sha256:0c2d7cafdab1084ebbd0112b3bedf76458ae6809686fb0ad9bae8022f11c3a84
shshenhx#shshenhx:~/Desktop/Docker$ docker pull python
Using default tag: latest
latest: Pulling from library/python
4176fe04cefe: Already exists
851356ecf618: Already exists
6115379c7b49: Already exists
aaf7d781d601: Already exists
40cf661a3cc4: Already exists
975fe2fd635f: Pull complete
bf4db784e7fd: Pull complete
0491f7e9426b: Pull complete
Digest: sha256:bf0718e2882cabc144c9c07d758b09c65efc104a6ddc72a9a976f8b26f67c2ee
Status: Downloaded newer image for python:latest
My question is, I have already rm python image, why it still shows already exist for some of layers? How can I totally delete all the python layers.
Thanks.
What helped for me was to run docker-system prune after removing all containers and images. So the whole process I got it to work was:
Remove all containers docker rm -vf $(docker ps -a -q)
Remove all images docker rmi -f $(docker images -a -q)
Prune system with docker system prune
From reference of docker images command
Docker images have intermediate layers that increase reusability, decrease
disk usage, and speed up docker build by allowing each step to be cached.
These intermediate layers are not shown by default.
Maybe those Already exists are intermediate layers. By default, they are hided when you run docker images, please try docker images --all.
Docker has a solution for this on docker dashboard, just press the button below and presto it works! (This is in the troubleshooting section. or the bug icon)
There is a docker command that deletes images that do not have an associated running container:
docker image prune
allows you to clean up dangling images (a dangling image is not tagged and not referenced by any container)
docker image prune -a
will remove all images that are not used by any existing containers
docker image prune -a --filter "until=48h"
you can add a filter (--filter) to limit the images to be pruned, as in this example, only images that are older than 48 hours
You can add the -f or --force flag to prune the images and not prompt for confirmation
You could use this in a scheduled bash script to keep your drive tidy
docker system prune -a helped me.It will remove both unused and dangling images.
You can use docker system prune but it will clear everything in docker.
Its not an image problem its actually cache problem so deleting your images wont work because the problem relies in layer caching.
There is a --no-cache flag in docker build but I don't know if it works with docker pull too.
For someone interested in: (1) using the command line; (2) removing cache only for images that have been removed with docker rmi (i.e., not visible with docker images --all), but for which caches are still present, and for which the Already exists shows for layers when pulling image; you can try this command:
docker builder prune
(It may take up to 15 minutes to run this command, so be patient.)
This command seems to have solved the problem for me. I did not want to use the general, "system-wide" prune / purge / clean docker commands and functionalities, because I had some images and containers running (not related to the image that I have already removed), that I wished to have left as they were.
Source of the solution: https://forums.docker.com/t/how-to-delete-cache/5753/7
I think that the situation of "invisible" cached layers for images removed with docker rmi may happen due to the fact of using docker compose (as suggested also by John Doe in his comment to one of the answers in the current question: docker pull always show "Already exists" for layers during pull even after deleting all images).
So after a lot of searching, I have a solution although I don't think it's necessarily a great one.
The "sledge hammer solution" would be to do:
rm -rf /var/lib/docker
This basically factory resets docker, so you would lose everything. This was fine for my case, but be very careful and think this through before using this one.
Please try this command docker rmi 336d482502ab and then try to pull it again.
This is probably a duplicate, but all answers that I saw didn't work for me.
I'm using docker (17.06.2-ce), docker-compose (1.16.1).
I have an image of solr which I use for development and testing purposes (and on CI too).
When making changes to the image I need to rebuild the image and recreate containers, so that the containers use the latest possible image, which, in turn, takes the latest possible code from the local repo.
I've created my own image which is based on official solr-docker image. The repo is a folder with additional steps that I'm applying to the image, such as copying files and making changes to existing configs using sed.
I'm working in the repo and have the containers running in the background.
When I need to refresh the containers, I usually do these commands
sudo docker-compose stop
sudo docker rm $(sudo docker ps -a -q)
sudo docker rmi $(sudo docker images -q)
sudo docker-compose up
The above 4 commands is the only way it works for me. All other approaches that I've tried din't rebuild the images and didn't create the containers based off the new, rebuilt images. In other words, the code in the image would be stale.
Questions:
Is it possible to refresh the image + rebuild the container using fewer commands?
Every time I'm running above 4 commands, docker would download ~500MB of dependencies. Is it possible to not to download them and just rebuild the image using updated local code and existing cached dependencies?
I usually do docker-compose rm && docker-compose build && docker-compose up for recreating docker containers: it won't download 500mb.
You can use docker-compose down which does the following:
down Stop and remove containers, networks, images, and volumes
Therefore the command to use will be: docker-compose down --rmi local && docker-compose up
The --rmi local option will remove the built image, and thus forcing a rebuild on up
I created my own image and pushed it to my repo on docker hub. I deleted all the images on my local box with docker rmi -f ...... Now docker images shows an empty list.
But when I do docker run xxxx/yyyy:zzzz it doesn't pull from my remote repo and starts a container right away.
Is there any cache or something else? If so, what is the way to clean it all?
Thank you
I know this is old now but thought I'd share still.
Docker will contain all those old images in a Cache unless you specifically build them with --no-cache, to clear the cache down you can simply run docker system prune -a -f and it should clear everything down including the cache.
Note: this will clear everything down including containers.
You forced removal of the image with -f. Since you used -f I'm assuming that the normal rmi failed because containers based on that image already existed. What this does is just untag the image. The data still exists as a diff for the container.
If you do a docker ps -a you should see containers based on that image. If you start more containers based on that same previous ID, the image diff still exists so you don't need to download anything. But once you remove all those containers, the diff will disappear and the image will be gone.