Request for early feedback on "JupyterHub Share Link"

rajiv.abraham · February 14, 2020, 9:20pm

we are using this in our demo right now for the sharing feature. So thank you very much.

Is it possible to customize the timeout instead of a fixed 1 hour value
It errored out if the second user who has the link does not have a notebook server up. So the user has to login to his account and start a server and then paste the link in the browser. Not a blocker, but I can see a lot of bug reports coming my way

rajiv.abraham · February 14, 2020, 9:25pm

Also, it would be nice to be able to apply your changes to the latest scipy/base jupyter stacks.
I’m looking at https://hub.docker.com/layers/danielballan/base-notebook-with-jupyterhub-share-labextension/latest/images/sha256-d08d45a6a14c6decb9912591889ee51c05fab46908d9f6ddcd5f80b5c1f8330a?context=explore to be inspired from but I see commands like
COPY file:be20131ba97563898c30998c19104b0d58d61a982ea5555a2002b0c2f0b49668 in /usr/local/bin/

So, I’m not sure what that means :). If you can share your Dockerfile that would be great.

danielballan · February 16, 2020, 3:28pm

@rajiv.abraham Great, glad you are getting use out of it. Thanks for the feedback, very helpful.

It is possible to customize the timeout (up to some hard-coded max, intended to ensure users don’t use this feature for very long term sharing, for which it is not suitable) by passing ?expiration_time=X into the request. The parameter is processed here:

github.com

danielballan/jupyterhub-share-link/blob/b248b5a9d49b5af8c6ace4d1d47addce738ad7db/jupyterhub_share_link/run.py#L39-L52


now = datetime.utcnow()
if data.get('expiration_time') is not None:
    expiration_time = datetime.fromtimestamp(data['expiration_time'])
else:
    # Default to one hour lifetime.
    expiration_time = now + timedelta(hours=1)
# Enforce a max of two days.
# This is not for long-term sharing, galleries, etc.
max_time = now + timedelta(days=2)
if expiration_time > max_time:
    raise HTTPError(
        403, (f"expiration_time must no more than two days "
              f"from now (current max: {max_time.timestamp()})")
    )

But this parameter is not currently exposed in the user interface of the JupyterLab extension. Do you think it would be sufficient to add a drop-down menu with a small selection of choices (1 hour, 1 day, 1 week)?

To your second point, the intended behavior is that the jupyterhub-share-link service should automatically start a server for the recipient (“second user who has the link”). Moreover, it ensures that it starts a server with the same spawner_args, e.g. the same Docker image in the case of dockerspawner. It first checks to see whether the recipient already has a compatible server already running and then starts one only if needed. It may require the user to be logged in, however. Could you verify what happens if the recipient is logged in but does not have any servers running when they paste the link into their browser? If that fails, you’ve found a bug that I haven’t reproduced myself.

Thanks for the ping about Docker stacks. No time to tackle that at this exact moment but I’ll circle back in the next couple days and update here.

rajiv.abraham · February 18, 2020, 2:02pm

hey @danielballan, thanks for the quick reply.

Expiration
Thanks for sharing that. I’ll check out the params option. The Drop down option looks good too. It would be great to have that!
Prerequisite: User 2 has to have a server up before the link can be shared.
I can’t reproduce it right now but we’ll be testing it and I’ll let you know.
Docker stacks. whenever you can, much appreciate it.

And this may be a vague question and perhaps just noise. But I had to install nbviewer to make it work. I can’t find the error right now. but just throwing it out there.

Topic		Replies	Views
Jupyterlab-hubshare: Jupyterlab extension to share notebooks JupyterHub jupyterhub	1	894	March 29, 2021
How to share notebooks with other users on jupyterhub JupyterHub	4	1602	April 12, 2024
Sharing of spawned notebook server discuss	3	871	June 10, 2020
Local notebook to be shared JupyterHub how-to , help-wanted	1	865	August 29, 2022
Collaborating on Jupyterhub Notebooks The Littlest JupyterHub how-to	1	967	August 9, 2021

Request for early feedback on "JupyterHub Share Link"

Related topics