Issues with the NRPE dashboard

[przemeklal]

Bug Description

1. NRPE Command Status is compressed into the space of the panel making it unreadable (screenshot attached).
   
   

Even when the panel is expanded the content is still constrained to the height of the panel. This becomes especially problematic with a high number of nrpe units (the attached screenshot was only 3 NRPE units, in real-life scenarios we'd deal with 100+ units).

Nagios/Thruk UI for comparison:

All service statuses are visible, sorted alphabetically, and there are multiple pages.

2. juju_unit label is hardcoded. In real-life scenarios the unit name is not always nrpe, but it can be something like nrpe-host, nrpe-lxd, etc. Usually, there are also multiple NRPE applications deployed in the same model.

3. Prometheus datasource is hardcoded.

4. The only way to use this dashboard is manual import from the nrpe.json file.

To Reproduce

Import nrpe.json into COS Grafana after relating nrpe <-> cos-proxy <-> prometheus.

Environment

cos-proxy latest/edge

Relevant log output

Screenshots attached.

Additional context

No response

[lucabello]

After looking into this for quite a bit, this is what I found:

That panel is using the State timeline visualization: apparently, this bug has been discussed in the grafana/grafana repo since 2021, and then in August 2023 it was moved to an issue, which is still open and doesn't seem to be getting a lot of activity besides for people +1-ing it.

The other points I'm still looking into :)

[lucabello]

About your points 2. and 3., they are being fixed by the linked PR :)

I think 4. is also unfortunately not changing, as cos-proxy is not sending those dashboards at all to avoid overcrowding Grafana with unused dashboards; thinking about it, they should probably not even live in that repo (since they are not sent to Grafana).