Sharing kernels along with personal kernels

[Sean Smith]

I'm using Jupyterhub and I've been in the habit of creating my own kernels by dropping json files into this directory:

~/.local/share/jupyter/kernels

where the kernel looks like

~/.local/share/jupyter/kernels/env0
                             kernel.json
                             logo-32x32.png
                             logo-64x64.png

And the `kernel.json` looks like:

{

 "argv": [

  "~/.conda/envs/env0/bin/python",

  "-m",

  "ipykernel_launcher",

  "-f",

  "{connection_file}"

 ],

 "language": "python",

 "display_name": "env0"

}

This works just fine for just me.  However, I'm working with my admin to set this up for other users less inclined to mess with config files.

My desire is to establish a set of kernels that everyone will have access to when they kick off their own notebook server via jupyterhub.  But I also want them to be able to add their own kernels if they wanted.

My first plan would be to create symbolic links in each users `~/.local/share/jupyter/kernels/` directory to some globally accessible directory with the appropriately specified kernels.  They could then create additional kernels on their own in the same directory.

However, this seems like there should be a configuration option that points jupyter to a directory to search for kernels and that we could build upon available kernels by looking in additional directories.  So that in some final state I could have a global config that points to where the globally accessible kernels are and users can have additional ones in their own directory.

Is my current plan sufficient?

Is my current plan good?

Are there configuration options that help me accomplish my goal?

Thanks,

Sean

[Roland Weber]

Hello Sean,

there's an environment variable JUPYTER_PATH that you can populate with additional directories to be searched.

https://jupyter.readthedocs.io/en/latest/projects/jupyter-directories.html#data-files

Cheers,

  Roland