Suggested CPU/memory resource requests/limits?

I’m working to provide some suggestions on resource requests for the JupyterHub helm chart but am struggling on what makes sense for various situations and how to communicate this in a way that is useful.

What resource requests have you been setting on your deployments and why?

Open questions

  1. If we provide example resource requests, for what assumed usage examples do we provide examples for?

    I’m thinking that we don’t want to try maintain more than two explicit suggestions on resource requests/limits besides the suggestion for a trial installation that could have the suggestion “don’t set anything explicitly”.

  2. In prometheus/grafana one can collect metrics about memory RSS, is this useful to relate the resource requests of memory to or should one relate it to something else?