add Grafana dashboard

[HarHarLinks]

Fixes #147

There isn't too much going on in the pushes section, in particular I don't have jira and figma set up and i'm not sure if there are metrics for generic hooks.

I'm not a pro Grafana user, but I think it's a solid start.

A possible improvement would be to limit the y axis legends to integers which I don't know how to do yet. I used templating to easily select the correct prometheus globally, however to my knowledge the use of templating prevents one from using Grafana alerts (for now).

Signed-off-by: Kim Brose 2803622+HarHarLinks@users.noreply.github.com

[Half-Shot]

Wow, this is really cool! It's certainly much more expanded than the one I use at home :)

[HarHarLinks]

Wow, this is really cool! It's certainly much more expanded than the one I use at home :)

Mayhaps you can check the parts I noted in OP against a bigger instance that uses those features?

[Half-Shot]

Looking great!

[HarHarLinks]

This is giving me some trouble. E.g. I'm using rate(process_cpu_seconds_total[$__rate_interval]).

Ever since I restarted the bridge, resetting the counter, I only get "No data" from that point onward. Even though the rate() function is supposed to handle such resets:

Breaks in monotonicity (such as counter resets due to target restarts) are automatically adjusted for.

Similarly for the many panels using similar functions.

[HarHarLinks]

I also don't understand what the CI issue is, so please advise if necessary.

[HarHarLinks]

So the issue above appears related to how often my prometheus polls metrics vs the range the derivation functions ((i)rate, (i)delta, etc) are applied on. I'm still not sure why it would work before the first counter reset and not after.

I am able to work around this issue by telling Grafana these 2 global variables:

Interval is my sampling interval, 2x Interval is simply twice that, since I was unable to do this programmatically (it appears these don't support arithmetics).

[HarHarLinks]

Still some kind of trouble I don't yet understand.

I gotta investigate if half my trouble originates from the fact that I originally polled metrics at a much higher rate and then decreased that to 5m...

[HarHarLinks]

This actual data

when put through idelta results in this weird dip

instead of being thrown out as a reset and idk why it would do that.

Otherwise works well as far as I can see.

[Half-Shot]

Have now tried it on EMS and it looks great, one nit!