Deploying JupyterHub for 2000 students on a private cluster

HI,
I’m looking for a good documentation for creating a jupyterhub able to dispatch the notebooks for 2000 students on our local infrastructure (VMs managed by a vCenter)
I tried a lot of tutorials without sucess.
Some of them use AWS or external cloud.
Others use the SwarmSpawner (deprecated).
Other ones are not precise enough (for me) regarding the internal network configuration, and the way the jupyterhub is using the underlying “migration” layer.
Thanks for any reference or hint.

Have you tried the Zero to JupyterHub for Kubernetes guide?

http://z2jh.jupyter.org/en/latest/

This is the base for what we use at Berkeley for student sizes similar to this amount. Maybe that’d be helpful?

Hello,
Thanks for your help, and for your suggestion.
Actually, I tried to follow this tutorial, before following other ones.
I may be wrong, but it is based on external cloud solutions.
That leads it to hide some important details about it, such as the internal network to define, and its configuration for jupyterHub.
Sincerely Yours
H. Delebecque

Ah yeah, I didn’t catch the “managed VMs” part of your original post. You’re right that z2jh assumes you have kuberneres running. For something that large that doesn’t use kubernetes I’m not sure :thinking:

Z2JH will work on private Kubernetes clusters, though obviously it means taking on the burden of maintaining it. See:

1 Like

Thanks a lot.
I will take a look at it.