Closed yuvipanda closed 7 months ago
Hi @yuvipanda, I might be able to help with the deploy on non-2i2c infrastructure. Can you point me to where I can start?
I had a nice conversation with Sima a few days ago, and we're working on getting this all set up on GCP. I've given her access to the earlier GCP project we used for testing (binderhub-service-development
I think), and are going to take it slow :)
Would love for you to try it out too, @rgaiacs
We probably should also add documentation, for users and admins. Having documentation is also relevant for the NFDI and CESSDA to allow them to deploy this.
Some next steps from chatting with @arnim:
Thank you @yuvipanda . These next step are perfect 👍
As an update, Sima has spent some time getting acquainted with GCP and set up a z2jh there to test (https://2i2c.slack.com/archives/C03RLNFM43F/p1699262794125269).
Next step is to try deploy binderhub-service there alongside.
Just had a meeting with @simaattar2003 on the deployment.
We've done a lot this quarter!
I'm going to close this out, and create a new issue for Q1 2024. Thanks everyone.
Technical Progress
These are broad outcomes of technical progress that need to happen for smoother UX. We currently have code that is decent demo quality, but needs to get closer to production quality
@jupyterhub/binderhub-client
JS library and release on npmjs, so we can use the same code as binderhub for talking to the binderhub APIOut of scope items
Since work will continue in Q1, the following items are explicitly marked as out of scope for Q4.
binderhub-service
andjupyter-fancy-profiles
repo live. There is ongoing conversation in the jupyterhub community about ajupyterhub-contrib
org, so let's wait on that.Deploy this on non-2i2c infrastructure
We need to make sure that all this work is not 2i2c specific. While the tech work has proceeded in a non-2i2c specific way, we currently don't have any deployment of this outside 2i2c infrastructure. This means there isn't clear step by step documentation on tying together the three projects (z2jh, binderhub-service, jupyterhub-fancy-profiles) that make this workflow possible.
We work potentially with Sima Attar to get a deployment running elsewhere - either on GCP (we have funding to setup a spare project) or on other infrastructure they have access to. The outcome of this is documentation that can be followed by anyone to set this up.
Write a blog post about this complex yet 'upstream first' process
There's been a lot of work done in making this goal possible, spread out across multiple projects -
kubespawner
,binderhub
,zero-to-jupyterhub-on-k8s
,repo2docker
,jupyterhub-fancy-profiles
, etc. For organizations approaching contributing major features to an open source ecosystem, this can be quite daunting! So we write a blog post describing how we did this, including the exporation paths that didn't go anywhere.