user diagnostics dashboard issues - with data sources?

[consideRatio]

• The Home directory usage works if viewed from for example https://grafana.smithsonian.2i2c.cloud/ which only has a single prometheus data source, but it shows "No data" when viewed from https://grafana.pilot.2i2c.cloud/ that is having multiple data sources.
• The CPU and Memory usage panels consistently broken showing No data.
• The CPU and Memory requests panels work for the single datasource instance, but seem to misbehave when used with multiple even if a dedicated datasource is picked out. I don't think the data is getting filtered by datasource correctly.

Single datasource	Multiple, and the single is selected

[ateucher]

Hi @consideRatio! A 👍 from me on this - would love to have the per-user, per-server memory and CPU usage queries working again in the Openscapes hubs, as we try and correlate usage with costs to generate cost estimates for certain workflows in the hub.

[ateucher]

I've been poking around at this - it looks like container_cpu_usage_seconds_total metric does not have any data for user pods. The hub_jupyter_org_node_purpose label has only one value:"core", and the pod label does not have any jupyter- prefixed values

[jnywong]

Hey @ateucher ! I just took a look at the user diagnostics dashboard on the openscapes grafana and they look okay to me. Is there still a problem from your POV?

[ateucher]

Thanks @jnywong! It looks like it's been fixed since March 25 - if I set the timeframe to the last 30 days there is no data before then:

Very glad to see it back working, thank you!