JupyterHub Operations With Respect to User Data and Uptime Availability

[julienchastang]

I would like to have some policies in place concerning this JupyterHub server. My goal here is for this project to be useful to the end user but also to keep cognitive load to a minimum for admins of this server:

User Data

I don't want this JupyterHub server to be a place where users can park their data and notebooks indefinitely. In short, I don't want to have to manage that especially in the long-term. I would like to have a simple policy that is easy to programatically enforce. Perhaps user directories will be deleted after 30 days of inactivity by a purge script that runs on cron. Moreover, I want users to understand this server is not backed by a Google data center with mega-redundancy behind the scenes. In other words, save your important work locally (File -> Download As).

JupyterHub Server Uptime

How do we communicate to users about downtime, maintenance, etc? We currently do not collect emails, only XSEDE user names. Again, as before, I want to keep cognitive load at a minimum here. Perhaps we should require users communicate critical uptime to us, for example, if they are running a class. Apart from that, they can learn about downtime and availability through an RSS feed or a mail list.