[06:53:35] <moritzm>	 not sure if https://gerrit.wikimedia.org/r/c/operations/puppet/+/1273441 needs some kind of user announcement? probably not given it's not really user-visible and applies in the background?
[07:33:58] <taavi>	 moritzm: I don't think so\
[07:38:28] <moritzm>	 ok, thanks. I'll simply merge that on Monday then
[14:07:25] <andrewbogott>	 taavi, the linter hates hiera lookups in classes, were you thinking of something else? https://gerrit.wikimedia.org/r/c/operations/puppet/+/1272837/4/modules/openstack/manifests/serverpackages/epoxy/bookworm.pp
[14:08:44] <taavi>	 it'll need to be plumbed through the profile where we call those classes
[14:09:17] <andrewbogott>	 messy. ok.
[14:10:16] <taavi>	 I don't think you could've avoided that even with the original implementation
[14:10:24] <taavi>	 also, typing will need to be Optional (and can be made more specific like Optional[Wmflib::HTTPUrl])
[14:13:51] <andrewbogott>	 I was planning to avoid it in /most/ profiles by relying on the default. I can still do that but then those classes will still need '"http://webproxy.${::site}.wmnet:8080"' hardcoded as the default
[14:15:25] <andrewbogott>	 well... I try just altering all the profiles and see if I get sick of it
[14:57:02] <andrewbogott>	 godog, if still around, do you have any concerns about this change?  https://gerrit.wikimedia.org/r/c/operations/puppet/+/1273833/2/hieradata/pontoon.yaml
[14:59:17] <dhinus>	 I'm about to send this announcement to cloud-announce, it was reviewed by manuel but can I get a +1 from WMCS as well? https://etherpad.wikimedia.org/p/FJfXKQLwuHX49XPPBRY7
[15:02:05] <andrewbogott>	 looks good
[15:02:35] <andrewbogott>	 is 'labswiki' still backing wikitech, or is it defunct?
[15:04:42] <dhinus>	 probably defunct, but the db is still there
[15:05:11] <dhinus>	 maybe we should check if it can be dropped completely
[15:06:15] <andrewbogott>	 definitely check in case it's still backing wikitech for some reason :)
[15:06:26] <dhinus>	 if that's the case, we should rename it instead :)
[15:06:34] <dhinus>	 I'm not gonna touch it anyway
[15:09:19] <taavi>	 andrewbogott: dhinus: labswiki is wikitech
[15:12:17] <dhinus>	 taavi: ack!
[15:13:15] <taavi>	 andrewbogott: and I just realized what I said earlier is wrong.. the type is Stdlib::HTTPUrl not Wmflib::HTTPUrl, sorry :/
[15:19:40] <andrewbogott>	 np
[16:01:49] <dhinus>	 we have a volunteer maintainer for superset.wmcloud: https://phabricator.wikimedia.org/T416373#11833942
[16:09:36] <bliviero>	 dhinus: if wikitech is affected by this work, and people are trying to access a wikitech article, would it mean that they'll get a 404 during that 12 hour period?   I am wondering whether the announcement could have said about wikitech also not being accessible 
[16:10:04] <taavi>	 bliviero: no
[16:10:41] <andrewbogott>	 bliviero: the maintenance is for toolforge-facing replicas, not the live production db
[16:10:41] <taavi>	 the wiki replicas have nothing to do with wiki page views
[16:13:32] <bliviero>	 taavi: andrewbogott thank you for setting me straight
[16:13:46] <andrewbogott>	 It's confusing!
[16:14:35] <bliviero>	 taavi: thank you both
[16:25:35] <andrewbogott>	 taavi: a bit of preliminary work :)  https://gerrit.wikimedia.org/r/c/operations/puppet/+/1273861
[16:25:48] <andrewbogott>	 Letting the pcc test that everywhere while I go to lunch
[16:26:12] <taavi>	 andrewbogott: do we actually set profile::openstack::base::version? I think ~all uses of that get overridden in a deployment-specific profile
[16:26:22] <andrewbogott>	 hm, good point
[16:26:31] <andrewbogott>	 so I need deployment-specific profiles
[16:26:34] <taavi>	 also, don't we need that on cloudbackups at least?
[16:26:37] <andrewbogott>	 I guess pcc would've told me that in a hurry
[16:26:50] <taavi>	 anyway, +1 on the idea
[16:27:08] <andrewbogott>	 We don't, cloudbackups just talk directly to rbd. We used to run cinder services on them and there's probably some cleanup overdue
[16:27:16] <taavi>	 ah
[16:27:45] <andrewbogott>	 the cinder backup service was a bit janky, I fixed some upstream bugs but it was never fast enough to really do volume of backups that we wanted.