[11:22:06] <jayme>	 have not seen this before: https://helm-playground.com
[16:02:49] <brouberol>	 oh this is nifty
[16:03:33] <brouberol>	 I will never get back the hours of my life that were spent in debugging helm templates, but this might helm me save some  
[17:20:11] <kamila_>	 is there also a helm-playground for debugging helmfile weirdness that behaves differently in prod and CI?
[17:20:17] <kamila_>	 or is that one called "production"?
[17:33:43] <jayme>	 that's called staging usually :)
[17:33:57] <jayme>	 kamila_: what do you see?
[17:34:47] <kamila_>	 jayme: in CI I have the correct image version: https://integration.wikimedia.org/ci/job/helm-lint/14024/console @ 17:49:00
[17:35:37] <kamila_>	 but when I try to deploy the diff says `image: docker-registry.discovery.wmnet/kube-state-metrics:v2.9.2`
[17:36:10] <kamila_>	 it's as if it were using the upstream chart without my changes, but the commit with my changes is there
[17:36:38] <kamila_>	 (and yes, it's called staging, you're correct :D)
[17:37:08] <jayme>	 could be that the new chart version has not yet made it's way to the helm chart repo
[17:37:13] <jayme>	 it takes a couple of minutes
[17:37:21] <jayme>	 ah, dammit
[17:37:39] <jayme>	 kamila_: my bad. In https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/972400 you did change the chart but you did not bump it's version
[17:37:48] <kamila_>	 oh crap -.-
[17:37:54] <kamila_>	 sorry
[17:38:06] <jayme>	 np, could have spotted that during review :)
[17:38:13] <kamila_>	 and thank you
[17:38:25] <kamila_>	 I could have not forgotten :D
[17:39:16] <kamila_>	 but... how do we actually deal with versioning of upstream-imported charts anyway?
[17:39:32] <jayme>	 yes! :D
[17:40:28] <kamila_>	 :D
[17:40:50] <jayme>	 kamila_: there is some discussion about that at https://wikitech.wikimedia.org/wiki/Talk:Kubernetes/Upstream_Helm_charts_policy
[17:42:14] <jayme>	 seems like my last recent idea was to add a annotation which carries the latest upstream version imported and just bump the actual version like you would with in house charts
[17:42:31] <kamila_>	 yeah
[17:43:15] <kamila_>	 ah, because of the "Helm considers these 'development' versions` thing? makes sense
[17:43:24] <kamila_>	 ok, thanks!
[18:21:39] <kamila_>	 (huh, it actually works! that's weird :D)
[18:26:19] <kamila_>	 question: how do we handle port allocation for system stuff? I know about https://wikitech.wikimedia.org/wiki/Kubernetes/Service_ports , but I'm asking about things in admin_ng, specifically which port to use for kube-state-metrics
[18:42:51] <kamila_>	 (the default is 8080, which seems like a bad idea, so sticking with the default isn't an option here)