[07:52:25] <brouberol>	 nice elukey thanks!
[09:52:21] <jayme>	 heads up: I've added a new node.kubernetes.io/link-speed-mbps label to k8s nodes https://gerrit.wikimedia.org/r/c/operations/puppet/+/1353053
[10:22:23] <btullis>	 jayme: Nice. Have you still got some 1 Gbps NICs on wikikube, then? Or are you moving some to 25 Gbps?
[10:42:43] <jayme>	 the majority of wikikube in eqiad is still 1G
[10:43:35] <jayme>	 in codfw it's better, but far from all on 10G
[11:52:59] <_joe_>	 which explains easily why codfw sees less memcached errors, btw :)
[12:04:13] <btullis>	 Right, yep Makes perfect sense. We had some issues with postgres running on hosts with 1Gbps NICs, but we ended up replacing all of the cards to make everything 10 Gbps. T414787
[12:04:13] <btullis>	 Our labels were hand-curated, but your method looks better.
[13:32:09] <jayme>	 btullis: interesting...but I don't have accesss to T414787 for whatever reason. Can you grant that? We will have to look into adding 10G NICs too I suppose
[14:11:31] <btullis>	 Ah yes, it was set to subscribers-only. Maybe we were discussing costs or something. I've added you.
[14:18:55] <jayme>	 <3