prometheus internal metrics missing

#1

Certain Prometheus metrics are missing, like scrape_duration_seconds. A lot of these help with diagnosing metrics issues. Recently I discovered that one of my apps was taking 20-30s to scrape and was throwing out rate calculations.

I imagine these aren’t currently exposed due to single instances of some internal services being used to scrape apps from different organizations but I’m only guessing.

I thought I saw that some of the built-in Prometheus metrics were exposed in a doc page but I can’t now find it.

#2

We actually use Vector to scrape app metrics locally on our hosts, which then ships them off to VictoriaMetrics via the prometheus remote write protocol. Vector doesn’t include metrics like scrape duration unfortunately, although I can definitely see how it might be useful!