On 29.04.23 08:09, Adam Sydnor wrote: > > I was wondering if it is a bad idea to aggregate histograms that have > different bucket boundaries. If so, why? > > Or can Prometheus handle this and produce reliable metrics like quartiles?
Generally, Prometheus cannot handle this. You could play some tricks and only add or remove buckets while keeping enough buckets the same, and then carefully craft your queries so that only buckets that are shared by all involved histograms are taken into account. But that's a lot hassle and not really recommended. The new Native Histograms (still an experimental feature, and the only instrumentation client with full support is client_golang) have everything you need. You might want to explore them. We still need to write a good "getting started" guide, but there are a number of talks available on Youtube. Maybe start here: - https://promcon.io/2022-munich/talks/native-histograms-in-prometheus/ - https://promcon.io/2022-munich/talks/promql-for-native-histograms/ -- Björn Rabenstein [PGP-ID] 0x851C3DA17D748D03 [email] bjo...@rabenste.in -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-users+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/ZGO6NAQ/EtqThCXs%40mail.rabenste.in.
