Richard Hartmann

Thanks; some comments/questions: * The label cmdline can get into high cardinality. Do you know how many total values a user should normally expect? * scaph_process_power_consumption_microwatts - is is possible...

That's a fair point; in a call now, do you want to put something in?

@scoren-gl reminder-poke :)

Minor nits/comments: * It would be more customary to use task**s**. * Can `celery_task_received_total` be calculated from the other metrics? * Is `celery_worker_up` about one single worker or a group...

Did you close this on purpose? Both the metric and the label names would need some work; I would be happy to help you polish them.

No problem! You should prefix them with a name, in your case I think it's `webscraper_` and suffix them with the unit, so: Metrics should be in base units, so...

Also, can you please sign your commits? See e.g. https://github.com/src-d/guide/blob/master/developer-community/fix-DCO.md

Thanks for this. Could you paste sample output of your integration, please?

These metrics don't follow Prometheus best practices; how much chance would you say is there to get them improved a bit? As an aside, UUIDv7 is not officially specified yet;...

Thanks for the cross-ref, I left a comment there as well. A few specific suggestions: * A prefix is recommended, i.e. `triton_*` for all metrics * `nv_inference_request_success` -> `nv_inference_request_success_total` *...