Closed aqw closed 3 years ago
So would this mean ups_output_percent_load_in_percent
should rather be ups_output_percent_load_ratio
?
Correct. And the value itself should be (for example) 0.82
instead of 82
.
Ah okay, I've just exposed the data as it is given by the API. Is it a common thing to change the format of the scraped data?
Done
For something as simple as this, it definitely makes sense. If we normalize the data flowing into Prometheus, it makes it much easier to reuse graphs and compare across systems.
However, in other situations, we ignore it. For example, Prometheus suggests using joules rather than watts. IMO, it's not worth it to adjust our watt metrics into joules, so we'll leave them as they are. :-)
Sorry to bug again on this issue, but it's better to prefix metrics with eaton_ups_*
instead of ups_*
, because this exporter only works for eaton UPS's.
Alright I will make the changes
Alex has addressed this in PR #15 already. If you can validate it and merge it to master, then we can run it and I'll set up the accompanying Eaton UPS Dashboard.
I just saw it.
Merged
Prometheus best practice is to use "ratio": rather than percent. https://prometheus.io/docs/practices/naming/#base-units
The field names should be updated to end in
_ratio
rather than_percent
. And the values should be changed accordingly.