[09:21:37] <_joe_> https://cloud.google.com/traffic-director/ [09:22:07] <_joe_> or, "the version of istio that actually works" [10:24:11] so me and gilles are planning to use mtail on thumbor servers [10:24:28] because the haproxy exporter is not producing that many useful metrics [10:25:28] it has a request time avg for the last 1024 successful connections [10:25:40] so we will use mtail [10:36:39] <_joe_> we really should explore using envoy instead [10:36:58] <_joe_> but sure, mtail seems sensible for now [10:37:18] <_joe_> I should look into the connection management methods envoy can use [10:41:59] _joe_: that's for the next level yes [13:14:40] envoy should be able to reploce the haproxy use case on thumbor [13:14:48] it can mimic that request protection [13:14:57] s/reploce/replace/g [14:16:33] ok to upgrade HHVM in mwmaint1002? there'll be 10-20 seconds of non-avail, but I guess every time is as good/bad as the other here... [14:27:50] <_joe_> moritzm: what will be unavailable? [14:28:02] <_joe_> just noc.w.org? [14:28:05] <_joe_> whatever [14:28:29] <_joe_> fsero: I'm pretty sure as well, and it has all the telemetry one might want [14:29:05] not sure whatever maintenance scripts on mwmaint1002 will be impacted or do they all use php7.2 by now? [14:30:42] <_joe_> no they all use hhvm [14:30:50] <_joe_> but if it's already running, that's not an issue [14:31:01] <_joe_> and once it's on disk, that is a brief moment [14:31:18] <_joe_> the time it usually takes are 1 - stopping the server 2 - unpacking the dbg package [14:34:40] yeah, it's only a few seconds, but that -dbg package is almost 500 MB, so takes a bit still [14:34:45] I'll do it in a few minutes [15:31:39] 10serviceops, 10Cassandra, 10Operations, 10Core Platform Team (Session Management Service (CDP2)), and 4 others: Credentials needed for session storage Cassandra cluster - https://phabricator.wikimedia.org/T219560 (10Eevans) >>! In T219560#5102752, @gerritbot wrote: > Change 502890 **merged** by Dzahn: > [... [23:07:21] 10serviceops, 10Release-Engineering-Team, 10Wikimedia-Site-requests, 10Continuous-Integration-Config: Consider creating a puppet-compiler equivalent for mediawiki-config.git - https://phabricator.wikimedia.org/T220775 (10Jdforrester-WMF) [23:24:20] 10serviceops, 10Cassandra, 10Operations, 10Core Platform Team (Session Management Service (CDP2)), and 4 others: Credentials needed for session storage Cassandra cluster - https://phabricator.wikimedia.org/T219560 (10Dzahn) >>! In T219560#5104271, @Eevans wrote: > FYI, this didn't fix the templating: Fixe... [23:24:30] 10serviceops, 10Cassandra, 10Operations, 10Core Platform Team (Session Management Service (CDP2)), and 4 others: Credentials needed for session storage Cassandra cluster - https://phabricator.wikimedia.org/T219560 (10Dzahn) p:05Triage→03High [23:25:43] 10serviceops, 10Operations, 10Core Platform Team (Session Management Service (CDP2)), 10Core Platform Team Kanban (Doing), and 4 others: Session storage Cassandra cluster configuration - https://phabricator.wikimedia.org/T215883 (10Dzahn) [23:25:50] 10serviceops, 10Cassandra, 10Operations, 10Core Platform Team (Session Management Service (CDP2)), and 4 others: Credentials needed for session storage Cassandra cluster - https://phabricator.wikimedia.org/T219560 (10Dzahn) 05Open→03Resolved [23:27:16] 10serviceops, 10Operations, 10Core Platform Team (Session Management Service (CDP2)), 10Core Platform Team Kanban (Doing), and 4 others: Session storage Cassandra cluster configuration - https://phabricator.wikimedia.org/T215883 (10Dzahn) subtask resolved. sessionstore1001 now has actual super_password fro... [23:28:08] win 46