Hello,
I am looking at setting up a Littliest JuypterHub on a single powerful server which would serve ~30 people. However, we would want to allow users to create their own conda environments as it wouldn’t work to have everyone share a single environment.
Can this be done? Or do I need to setup JuypterHub using the Zero-to-Kubernetes even though we’ll only be running it on a single machine? Or can I setup JuypterHub another way?
I would rather not have to give everyone sudo rights to allow them to create their own environments.
I think so:
- Use nb-conda-kernel extension to help identify kernels in conda environments
- Help users create conda kernels in their home directories, maybe with configuration of conda/mamba
Can’t verify things further from my mobile atm, but something along those lines may very well work – you centralize some key plugins and config in the environment starting the jupyter server, ans they create stuff in their home dirs
Ah great, so If I install nb-conda-kernel in the server’s conda environment, then the users should be able to use conda kernels from their home directories? I’ll have to have a play around when I get access to the server.
Relevant config of nb-conda-kernels will be where to look for kernels - inte the users home dirs somewhere, which is also why you want to help users configure some default destination for the kernels to be within there home dirs
Hi, did any of you manage to actually test this out for yourselves?
I ended up using the GitHub - jupyterhub/jupyterhub-the-hard-way: Understanding JupyterHub by deploying it step by step · GitHub tutorial as then my jhub accounts match the Linux user accounts and I understand how to change things and set it to properly.