[07:57:08] <akosiaris>	 ottomata: yes, I think so. Not because different people would answer this specific question, but because it would help raise awareness and knowledge more
[07:57:21] <akosiaris>	 also no bots here, better SNR
[09:05:27] <jayme>	 So, I was reading the backscroll on this and the change at https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/905260/
[09:06:33] <elukey>	 https://grafana.wikimedia.org/d/hyl18XgMk/kubernetes-container-details?orgId=1&var-datasource=eqiad%20prometheus%2Fk8s-dse&var-namespace=rdf-streaming-updater&var-pod=flink-app-wdqs-d4cb85fcf-bsnk7&var-container=All&from=now-30m&to=now
[09:06:35] <jayme>	 what puzzles me is that the prometheus infrastructure user has the roles "view,api-metrics,system:monitoring" - that should include get/list/watch access to the resources
[09:06:37] <elukey>	 ottomata: --^
[09:06:39] <elukey>	 really nice
[09:08:05] <elukey>	 also dcausse --^ (IIRC you asked in another chan the same question, now we have metrics for containers in DSE)
[09:08:49] <jayme>	 akosiaris: see my message here
[09:08:58] <jayme>	 I think something is hiding here
[09:10:30] <akosiaris>	 jayme: view doesn't have nodes listing, system:monitoring doesn't have nodes either and api-metrics is just /metrics
[09:10:45] <akosiaris>	 I think we can remove view btw
[09:11:18] <akosiaris>	 in the past we did a ClusterRoleBinding for system:heapster
[09:11:29] <akosiaris>	 and it has those rights, but it's deprecated anyway
[09:11:40] <akosiaris>	 and ok, it makes more sense to actually ship what upstream ships
[09:12:18] <jayme>	 ah, okay. view does not have list on nodes is the part I'm missing
[09:12:54] <jayme>	 but we should def. clean up the infrastructure user then as it will absolutely confuse our future selves again
[09:13:06] <akosiaris>	 yeah, +1 on that 
[09:13:50] <jayme>	 the change also reminds me that I still have to clean up helmfile.d/admin_ng from the k8s 1.16 stuff :-(
[09:16:50] <dcausse>	 elukey: thanks!
[12:39:00] <ottomata>	 <3 <3 <3 thank you all!
[14:49:38] <inflatador>	 do we use persistent volumes anywhere in our k8s clusters?
[14:56:48] <jhathaway>	 inflatador: no, not yet
[14:57:13] <jhathaway>	 we have discussed them a bit in the k8s sig meetings, but there hasn't been a compelling use case yet
[15:03:30] <inflatador>	 jhathaway thanks, that's what I thought. Was just talking about storage needs for some of upcoming project w ottomata and some others
[15:04:31] <jhathaway>	 yeah, I think everyone agrees that having persistent volumes would be great, but also supporting such a feature would require quite a bit of engineering and ongoing maintenance work
[15:05:09] <inflatador>	 100% agree, just wanted to make sure there were no new developments on that front
[15:05:27] <akosiaris>	 I think that https://en.wikipedia.org/wiki/The_Chicken_and_the_Pig fits here
[15:05:49] <akosiaris>	 all? of us are the chicken in this fable. Nobody is committed enough to be the pig 
[15:06:25] <akosiaris>	 we 'd all love it, nobody has pressing enough use cases to pay the price for that
[15:06:51] <akosiaris>	 and all is with a question mark, cause I think DE are the only exception
[15:07:04] <inflatador>	 I'll go on the record saying I would NOT love it ;)
[15:07:31] <akosiaris>	 I would love the use cases it allows, but I wouldn't love it on wikikube