Finished proof of concept of JupyterHub running on Docker in a cluster

[isuftin]

Will add README and split out repositories tomorrow.

[rsignell-usgs]

@isuftin, who can access this JupyterHub?

[isuftin]

@rsignell-usgs No one at the moment. This is a proof of concept configuration that allows me to run the JupyterHub in a Docker Swarm cluster locally. I will spend tomorrow going through the configuration for JupyterHub a bit to perfect the process and hoping by the next day I will begin working on deploying that into CHS in a Docker Swarm cluster.

Is there any specific configuration that you'd like to see up front or is just having it run good enough to begin?

Also, through what means were you looking to authenticate into JupyterHub?

http://jupyterhub.readthedocs.io/en/latest/authenticators.html

For this proof of concept I am automating setting up user accounts on the Docker container that the JupyterHub instance runs. In production, that's not going to be viable.

[rsignell-usgs]

Are we allowed to choose authenticators? Personally, I think Github auth with a whitelist would be most flexible, but is it okay for our collaborators to use CHS services?

[isuftin]

@rsignell-usgs We're setting this up so we can do whatever we like :) A GitHub auth list can work. I'm not sure which CHS services you'd prefer users use? In the future, we can probably set up a load balancer with a well known fully qualified domain name which would allow DOI to link back to it as a SAML endpoint which would allow smartcard auth into the application via the JupyterHub Kerberos authorizer

[rsignell-usgs]

@isuftin , cool, let's use github auth for now. I don't care where we run, but doesn't the CHS service we pick impact who pays the cost (e.g. CMG or OWI)? Or are the costs already "covered" somehow?

[isuftin]

@rsignell-usgs I couldn't say. I'll email you in regards to that