Installed a new debian server, installed docker, but then now i have a problem with permissions on passed directories.
On the previous server, the uid/gids inside the docker container match the uid/gid on the real server.
Root is 0, www-data is 33, and so on.
On this new server, instead, files owned by root (0) in the container are translated to 1000 on the server, www-data (33) is 100032, and so on (+1000 appended to the uid)
Is this normal or did I misconfigure something? On the previous server I was running everything as root (the interactive user was root), and i would like to avoid that
It’s actually a suggested configuration / best practice to NOT have container user IDs matching the host user IDs.
Ditch the idea of root and user in a docker container. For your containerized application use 10000:10001. You’ll have only one application and one “user” in the container anyways when doing it right.
To be even more on the secure side use a different random user ID and group ID for every container.
This is really dependent on whether or not you want to interact with mounted volumes. In a production setting, containers are ephemeral and should essentially never be touched. Data is abstracted into stores like a database or object storage. If you’re interacting with mounted volumes, it’s usually through a different layer of abstraction like Kibana reading Elastic indices. In a self-hosted setting, you might be sidestepping dependency hell on a local system by containerizing. Data is often tightly coupled to the local filesystem. It is much easier to match the container user to the desired local user to avoid constant sudo calls.
I had to check the community before responding. Since we’re talking self-hosted, your advice is largely overkill.
… basically anything. Yes. You will always find yourself in problems where the best practice isn’t the best solution for.
In your described use case an option would be having the application inside the container running with 10000:10001 but writing the data into another directory that is configured to use 1000:1001 (or whatever the user is you want to access the data with from your host) and just mount the volume there. This takes a bit more configuration effort than just running the application with 1000:1001 … but still :)
Yep! The names are basically just a convenient way for referencing a user or group ID.
Under normal circumstances you should let the system decide what IDs to use, but in the confined environment of a docker container you can do pretty much what you want.
If you really, really, really want to create a user and group just set the IDs manually:
FROM alpine:latest
COPY myscript.sh /app/myscript.sh
RUN addgroup -g 10001 mycoolgroup && adduser -D -u 10000 -G mycoolgroup mycooluser
USER mycooluser:mycoolgroup
CMD ["sh", "/app/myscript.sh"]
Just make sure to stay at or above 10000 so you won’t accidentally re-use IDs that are already defined on the host.
So add your user to the new docker group made on install of that package and you’ll be able to docker without sudo.
You may need to relogin or newgrp docker before it works tho
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !selfhosted@lemmy.world
A place to share alternatives to popular online services that can be self-hosted without giving up privacy or locking you into a service you don’t control.
Rules:
Be civil: we’re here to support and learn from one another. Insults won’t be tolerated. Flame wars are frowned upon.
No spam posting.
Posts have to be centered around self-hosting. There are other communities for discussing hardware or home computing. If it’s not obvious why your post topic revolves around selfhosting, please include details to make it clear.
Don’t duplicate the full text of your blog or github here. Just post the link for folks to click.
Submission headline should match the article title (don’t cherry-pick information from the title to fit your agenda).
Looks like you are running rootless.
I’m not very well versed on docker, but this sounds like a config issue. The behavior seems similar to “squash root” found in many other services.
It’s actually a suggested configuration / best practice to NOT have container user IDs matching the host user IDs.
Ditch the idea of root and user in a docker container. For your containerized application use 10000:10001. You’ll have only one application and one “user” in the container anyways when doing it right.
To be even more on the secure side use a different random user ID and group ID for every container.
This is really dependent on whether or not you want to interact with mounted volumes. In a production setting, containers are ephemeral and should essentially never be touched. Data is abstracted into stores like a database or object storage. If you’re interacting with mounted volumes, it’s usually through a different layer of abstraction like Kibana reading Elastic indices. In a self-hosted setting, you might be sidestepping dependency hell on a local system by containerizing. Data is often tightly coupled to the local filesystem. It is much easier to match the container user to the desired local user to avoid constant
sudocalls.I had to check the community before responding. Since we’re talking self-hosted, your advice is largely overkill.
… basically anything. Yes. You will always find yourself in problems where the best practice isn’t the best solution for.
In your described use case an option would be having the application inside the container running with
10000:10001but writing the data into another directory that is configured to use1000:1001(or whatever the user is you want to access the data with from your host) and just mount the volume there. This takes a bit more configuration effort than just running the application with1000:1001… but still :)Do I need to actually create the user in advance or can I just choose a string as I see fit?
You don’t need to create the user first. Here’s the simplest I can come up with:
This simply runs
/app/myscript.shwith UID 10000 and GID 10001.Wasnt aware that you can just think of IDs from fresh air.
Thought it was to create the user and ID manually amd then be able to use it.
Yep! The names are basically just a convenient way for referencing a user or group ID.
Under normal circumstances you should let the system decide what IDs to use, but in the confined environment of a docker container you can do pretty much what you want.
If you really, really, really want to create a user and group just set the IDs manually:
Just make sure to stay at or above 10000 so you won’t accidentally re-use IDs that are already defined on the host.
My go-to for user and group IDs is 1234:1234
checked .bash_history, looks like i installed docker in the new rootless mode
now i need to see how to restore it to work in the traditional way or i will become crazy with the permissions…
I fixed it:
for future reference:
dockerd-rootless-setuptool.sh uninstallWhy go through all of that complexity when you could just
sudo apt install docker?i don’t want to type
sudobefore each singledockercommandSo add your user to the new docker group made on install of that package and you’ll be able to docker without sudo. You may need to relogin or
newgrp dockerbefore it works thoYou can do that with regular docker. Just add your user to the docker group.
(don’t forget to log out and log in again after adding new groups to your user)
Niche use case, but you can also use
newgrpto run commands with a recently-added group to your user, without having to logout/login yet.Or start a new session by typing bash, when already in bash.