[01:46:10] gehel / ebernhardson: with `sdcquery01` killed, I updated https://gerrit.wikimedia.org/r/c/operations/puppet/+/713946 to no longer accept an `Optional` for `oauth_settings`, and I changed `oauth` to true in wcqs hiera instead of false (it was being set as true in horizon for wcqs-beta-01 BTW). However I think the patch still isn't ready to merge because we need to actually set up the `oauth_settings` array now [01:46:24] Could use some guidance on that tomorrow (perhaps in the weds meeting) [09:21:43] zpapierski: you might be able to help ryankemper on the oauth config [09:59:31] lunch [10:03:57] lunch [10:03:59] break [10:43:37] ebernhardson (for when you're here): it seems we are somehow doing the same thing? [10:44:26] (I forgot to put my notes in standup doc, I've done that today) [10:49:35] aand you actually had reviews for that I didn't notice yesterday [10:49:59] fyi I poked some folks toward that wqcs ui deploy think etc, you already see Amir looking at it :) [10:50:11] thanks addshore ! [10:50:18] :party-emoji: [10:57:35] ebernhardson: I follow up with a test on yarn with your changes, but please assign yourself to the ticket in the future, so we won't duplicate work. Fortunately, between meetings lately it didn't take me long to do those changes [11:00:48] anyway, your changes are way nicer than mine, I'm glad you had the chance to look at the code :) [11:13:21] ryankemper: we have an office hour, but I'll gladly assist after [12:03:04] hmm, this is weird -I'm getting a spotbugs error on rdf master [12:03:36] and on the line that has nothing to do with the error itself, no less [12:10:09] ah, known issue on JDK11 [12:34:17] zpapierski: which spotbug error? do we need to upgrade spotbug version? [12:34:31] I just did it, but there's another false positive there [12:35:09] basically, I have this - https://github.com/spotbugs/spotbugs/issues/756 (it's an old one but apparently resurfaced later) [12:35:58] after I update spotbugs to 4.3.0 (we don't have anything newer - which would be 4.4.0 - on our mirror, maybe plugin itself doesn't have this version, it's very new) [12:36:10] :/ I remember that one biting us before [12:36:21] I get EI_EXPOSE_REP [12:36:22] I wonder why this code isn't tagged as generate [12:36:25] generated [12:36:38] it's not I think? [12:36:50] EI_EXPOSE_REP is also a false positive? [12:37:01] it depends how you treat that [12:37:34] ah, it's hard to explain like that - a meet? [12:38:02] meet.google.com/nxz-bpje-tjs [12:52:28] dcausse: are you around? Any chance you could jump in ^ to talk hiring? [13:04:33] lunch and relocation [13:05:10] ryankemper: I missed that I already have a meeting after office hours, but I can help a bit later on ( like 30min after unmeeting, I will need to relocate back, first) [13:58:16] gehel: sorry was away [13:58:47] dcausse: no problem, I'll ping you in 30' (meeting) [13:59:06] will be in a meeting too [14:36:56] then tomorrow [14:37:10] actually, that's better, let's debrief with both you and Zbyszko [14:37:19] * gehel is too impatient to move this hiring forward [14:56:09] gehel, ryankemper: I'm working on config for Toolhub and hit a question that y'all may be able to advise me on. In https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/714867/5..7/helmfile.d/services/toolhub/values.yaml#35 I need to pick an envoy proxy to reach the elasticsearch cluster. Would one of the {chi,psi,omega} variants be correct to use or should I get the generic "search-https_eqiad" exposed to the helm configuration? [14:57:12] Naively they all seem to be the same endpoint, but I'm not sure if there is something subtle I'm missing [14:58:13] mostly the same endpoint, with different ports [14:58:17] bd808: https://search.svc.eqiad.wmnet:9243/ is the main cluster, also called chi [14:58:32] Did we decide already on which of those clusters we want to host your index? [14:58:42] bd808: chi kinda existed before they had names, when there was just one, so it's kinda grandfathered in to not having chi in the dns [14:59:07] I assume we have those small indices on psi or omega, but I haven't checked [14:59:21] yes, chi is the ~200 largest wikis, the rest are evenly split between psi and omega [15:00:00] we probably want toolhub to be on the same cluster as phab, translations and those other not cirrus related indices? [15:00:09] those are all on chi too [15:00:39] then chi it is! [15:01:17] awesome! thanks ebernhardson and gehel [15:01:41] at some point, we'll spin up a dedicated cluster for those non cirrus use cases [15:58:31] zpapierski: RE the oauth stuff from earlier, if you do have a bit of time after the unmeeting that works great, but I know it'll be quite late there for you so no worries either way [16:30:01] that's ok, but I might need to relocate first [18:46:11] ebernhardson: https://phabricator.wikimedia.org/T280006#7325872 I might have butchered the explanation, but it's a start [19:15:53] looks good. Hopefully i'm not completely wrong on how that routing works :)