[WIP] deployment at SDSC

[zonca]

Following https://github.com/jupyterhub/batchspawner/issues/12 I prepared an example based on our configuration at SDSC.

I chose a folder name that is a summary of the configuration details.

Documentation is part in the README.md and part in the docstrings.

@mbmilligan, @minrk some suggestions? either on the setup itself or on documentation?

For convenience, here direct link to the rendered README

[mbmilligan]

Nice, thanks for contributing this! Now I'm both inspired to finish documenting my setup and have a reasonable template for it too.

In terms of suggestions, I guess things I would like to see added would be:

• mention explicitly what software needs to be installed on the Hub and compute hosts (looks like you're using conda packages, but also system packages like supervisord) -- this probably merits its own heading in the README
• note what, if any, deployment automation or central configuration management you are using -- a point that keeps coming up when I talk to devops types about this is the question of whether Jupyterhub can play nicely with such systems (YES!)

[zonca]

thanks @mbmilligan !

added Requirements to the README.

No deployment automation for this, I feel like the setup is quite simple on the system side. Also I don't plan to deploy many of this HPC-connected instances. So I think an ansible playbook would be too much.

Instead for teaching workshops or classes, where you want to deploy many times, Ansible works great, see: https://github.com/jupyterhub/jupyterhub-deploy-teaching

[mbmilligan]

I like it, thanks!

[willingc]

@zonca and @mbmilligan 🍰 🍪 Thanks!