For the records, it looks like there's a third option: - honor_timestamps -> false (fake ilusion of real time metric)
On Wednesday, 1 April 2020 21:11:29 UTC+2, Albert Serrallé wrote: > > Hello, > > I'm trying to figure out how to federate my cloudwatch_exporter metrics. > > The default metrics are 10m old, I understand that this is needed because > cloudwatch consolidate metrics over time. My prometheus is scraping the > exporter and saving the values with the original timestamp. Federation > queries (as they are instant queries) cannot retrieve those metrics. > > So, I think that my only options are: > > - set_timestamp -> false (fake illusion of real time metric, which are > really 10m old) > - delay_seconds -> 1s-5m (metric might not be consolidated in > cloudwatch) > > There's anything else I can try? I don't like any of those two methods for > the exposed reasons. I've been looking at the changes in stale logic for > 2.0, but I don't know if this applies to me or how to do it: > > If you're federating instance-level metrics, switch to only aggregated >> metrics >> > > So, what's the recommended way to overcome this? > > Many thanks. > -- 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/3ca678a8-9ffb-4a47-9372-76297f56e136%40googlegroups.com.
