Add info how to persist user data

[mathbunnyru]

Describe your changes

Issue ticket if applicable

Fix: #1959

Checklist (especially for first-time contributors)

• I have performed a self-review of my code
• If it is a core feature, I have added thorough tests
• I will try not to use force-push to make the review process easier for reviewers
• I have updated the documentation for significant changes

[mathbunnyru]

@laundmo @bjornjorgensen please, take a look here.

[mathbunnyru]

This makes it possible to address it in other parts of docs.

[laundmo]

It's good to have this documented, but it doesnt address the core issue of how to persist packages installed by a user. it only documents how to persist packages installed by the admin of the container host.

[mathbunnyru]

but it doesnt address the core issue of how to persist packages installed by a user

Could you please explain, what you mean by this?
Which packages and installed how exactly?

[laundmo]

a user of jupyter lab, who only sees the webinterface, may run !pip install package to install a package. These wont be persisted with the methods documented here.

[mathbunnyru]

Ok, I added a note that a package installed inside a running container will not be preserved.
Please, take a look.

[laundmo]

the core issue is that users cant interact with the docker image - by design, and because my users dont have the knowhow. knowing that it wont be persisted is good, but it shouldnt close #1959

[mathbunnyru]

the core issue is that users cant interact with the docker image - by design

Why can't they? They can, but as it always works with docker images, some changes are lost after a container exists, and this is indeed by design.

and because my users dont have the knowhow.

That's why we're adding these docs.

knowing that it wont be persisted is good, but it shouldnt close #1959

Could you make a suggestion what to add here?
You told this is a documentation issue and I believe I added all the neccessary information to the docs.

[laundmo]

Why can't they? They can, but as it always works with docker images

sorry, this was unclear: they cant access the docker host or start/stop/rebuild images.

You told this is a documentation issue

initially i thought it was. maybe its actually a feature request.

[mathbunnyru]

Why can't they? They can, but as it always works with docker images

sorry, this was unclear: they cant access the docker host or start/stop/rebuild images.

You told this is a documentation issue

initially i thought it was. maybe its actually a feature request.

Ok, then please create a new issue with a feature request, which describes what you want users to be able to do (and then we will discuss if it's possible).
Also, if you want to have the whole state saved, you can use docker commit command (you need to have an access to the host machine obviously), but this is not the best practice: https://stackoverflow.com/a/44480870/4881441

For now, I'm merging this PR (and closing the issue) as I believe there is nothing more to document on this issue.

[benz0li]

@laundmo You might be interested in my/b-data's JupyterLab docker stacks.

Those images

1. allow to mount¹ the whole user directory and thus preserve all user data.
2. use Python (no Conda/Mamba) and install packages in the home directory.

Demo environment: https://demo.jupyter.b-data.ch

Cross references:

1. Populate home dir if it is empty #1478 (comment)
2. [ENH] - /home/jovyan/work is confusing (documentation) #1969

Footnotes

1. Either volume or bind mount ↩