JupyterHub and BinderHub Team Meeting#
Date: Thursday 16th September 2021
Time: 5pm UTC
GitHub issue: jupyterhub/team-compass#442
Calendar for future meetings: https://jupyterhub-team-compass.readthedocs.io/en/latest/meetings.html
Welcome to the Team Meeting#
If you are joining the team video meeting, sign in below so we know who was here. Roll call:
tim / Binder / @betatim
Min / Simula / @minrk
Sarah / 2i2c / @sgibson91
Simon / University of Dundee / @manics
Erik / Sundell Open Source / @consideRatio
Rollin / NERSC / @rcthomas
Isabela / Quansight Labs / @isabela-pf
Mridul / GESIS / @Mriduls
60 second updates on things you have been up to, questions you have, or developments you think people should know about. Please add yourself, and if you do not have an update to share, you can pass.
Erik about jupyterhub/docker-image-cleaner extracted out from jupyterhub/binderhub
Reports and celebrations#
This is a place to make announcements (without a need for discussion). This is also a great place to give shout-outs to contributors! We’ll read through these at the beginning of the meeting.
Isabela is here to give a heads-up that there will be a Jupyter accessibility workshop in the next month. Date and time to be announced soon.
Let’s collect all potential agenda items here before the start of the meeting. We will then attempt to create a coherent agenda that fits in the 60m meeting slot. If there are similar items try and group them together.
Simon (10m): Running BinderHub without Kubernetes
Katacoda example https://katacoda.com/manics/scenarios/binderhub-in-a-box
Collaborating with Yuvi for this work. Simon gives a demo showing how it runs and how to set it up.
This creates a single-server JupyterHub with minimal requirements in about five minutes.
Motivation: this could make it easier to test PRs
Question: where is repo2docker being run? It doesn’t really need to be run in a container, it’s run directly on the VM.
would be interesting to make a conda-forge package to make installing BinderHub easier (no compiler stuff needed)
Question: is this related to discussions about wanting to be able to build JupyterHub from images on the fly? That wasn’t the motivation, but it may make it more possible.
If we brought Binder as a a service to JupyterHub, what is the UI/UX a user would want?
figuring out this would be super useful, Tim thinks lots of people have lots of different ideas about this.
Question: where does the image end up? Locally on your docker.
Does this need support or review? It has already been reviewed, so it’s just implementing feedback.
Tim (10m): JupyterLab as default UI
old man Tim isn’t bothered by the switch but not so happy with how it turned out
is there a plan to update the documentation and examples?
should we show a link to a “how to update your repo” tutorial on the 404 page? Lots of people who do not pay attention to mybinder.org development will see a 404 now and (I guess) are kind of stranded.
A process question: why did the initial switch to lab not involve work to update the docs?
What is the strategy for dealing with the “dead” links that now exist around the internet?
some cases I’ve seen:
repo uses custom Dockerfile without lab -> now shows 404s
could argue that “custom Dockerfile means you are on your own”
teaching repo that links directly to notebooks with custom extensions -> now get “retro lab” but the extensions are gone/not there/UI looks different
filepath=with old version of JupyterLab which doesn’t support
Question: Have these consequences been discussed somewhere? Have the people involved intentionally not made these changes? Min says it may have gotten overlooked because this documentation is not in the same repo as the PR that made this change.
Chris: in the future this might be better served as a Jupyter Enhancement Proposal to notify the wider community because this work probably has influenced even more than we are currently aware of.
Min mentioned there was a Discourse poll https://discourse.jupyter.org/t/poll-should-zero-to-jupyterhub-switch-to-jupyterlab-as-the-default-user-interface/8001
Maybe we have to expect that things like this may have larger influence than we can easily anticipate and just prepare to be on the look out for broken things for a period of time after.
Sarah says perhaps there’s a way to better document this process too since it was a bigger change than typical.
Simon (10m): General plan for using DockerHub vs quay.io
We moved repo2docker to quay.io due to limited number of API tokens in the
jupyterhub) Docker Hub
What should we do in general?
Current inclination: don’t break the existing docker-focused repos, but update the dev builds. Encourage switching but don’t force to move it at the moment.
Min: this does seem to be the popular process amongst projects right now.
Chris: notes lacking documentation of process related to credentials on docker hub / quay / etc.
Tim: what do we need to update in our docs once we use quay.io?
maybe look at the docs to find where we refer to Docker hub in the context of setting up a BinderHub
If we have docs recommending docker credentials to pull the images from docker hub to avoid rate limiting, that needs to be updated ([name=Min] I’m not sure we have this, but mybinder-sre or z2jh)
Erik: diataxis.fr for doc structure
Min (5m): JupyterHub 2.0 beta. Ready to publish after
Changelog PR: jupyterhub/jupyterhub#3602
Erik: after jupyterhub beta, we can make z2jh alpha with jupyterhub beta
Choice: main branch w/ prerelease and ‘stable’ branch, vs main=stable, ‘dev’ preview
Focus discussion here: jupyterhub/zero-to-jupyterhub-k8s#2386
General discussion: explore A/B demo of new features; bring back ‘beta.mybinder.org’? (open issue on mybinder)