The idea is that all of the generic, reusable config lives in the
Then all of the site-specific config, like SSL, users, and local Matlab install, goes in a separate
deploy-helper script and gets mounted into the Docker container at run time.
To do your own deployment, you'd copy the
deploy-helper, edit it with your own site-specific config, and run it. Or just use the
deploy-helper as documentation for how to work with
ninjaben/jupyter-jub-oauth-matlab. You do you!
Here are the site-specific configs you need:
- Google account email addresses for known users. At least one admin (you!).
- A domain name where you'll host your JupyterHub. You can use
lvh.meto point back at localhost, which is great for testing.
- OAuth web app Client Id and Client Secret obtained from your Google Developer Console.
- Your own SSL certificate and key. Or, the
deploy-helpercan self-sign one for you.
- The path to the Matlab installation on the Docker host.
- The path on the Docker host where to put the Matlab execution logs.
- The MAC address associated with your own Matlab license.
This image was inspired by the jupyterhub official oauthenticator example. Thanks!
Thanks also to Michael Perry at Stanford for info, inspiration, starter code!