[13:41:06] jayme: hi I have a question for you. Gerrit got switched to use Envoy this morning and I found your Grafana dashboard for Envoy telemetry at https://grafana-rw.wikimedia.org/d/VTCkm29Wz/envoy-telemetry [13:41:06] If I search for gerrit2003 in `Origin instance` it does not show up. It turns out the reason is the datasource is set to `eqiad prometheus/ops`. I am wondering whether that can be changed to `Thanos` instead which will make it easy to search for an host which is not in eqiad ;) [13:41:37] (else I gotta teach me brain to always check the `datasource` field) [14:10:10] hashar: it can be, but the panels/queries would have to be adapted to that change ofc (filtering by site and prometheus label) [14:53:02] jayme: I imagine your primary case is for Wikikube and you thus use the datasource to switch between eqiad/codfw? [14:53:18] or have it default to whatever DC MW is running from [14:56:39] not really since this is for non-k8s envoys only. for k8s there is a separate dashboard [14:56:55] the main reason for it not using thanos is: there was no thanos back then [14:58:58] ah that makes sense [14:59:41] so if I get I understood probably the dashboard would need: 1) a variable for $site and 2) update all metrics to have site="$site" ? [14:59:50] (I am rephrasing what you said earlier to make sure I got it right :) ) [15:04:06] and similar topic, Envoy has statistics for downstream / http connection manager, which I guess are requests received by Envoy from ATS. And I am wondering whether those would be worth adding to that dashboard [15:35:02] sounds about right. you need to filter by site and prometheus labels then. there should be plenty of dashboards that to something like that [15:35:34] and if you feel like something is missing, feel free to add it [16:16:21] jayme: I'll look at it tomorrow, that sounds a nice friday task :] Thank you!