[15:57:00] duesen: Do you know "tosfos"? [15:57:10] Seems to be re-assigning your tasks. [16:00:08] Krinkle: https://www.mediawiki.org/wiki/User:Tosfos [16:01:01] yet he doesn't leave any explanation indeed [16:01:29] ah, the phab account has no name, description, or link to mw.org account [16:01:49] Didn't think of checking mw.org directly. i suppose it could be a different person but probably not indeed [16:02:11] yup, but the name sounded familiar from https://www.mediawiki.org/wiki/Gerrit/New_repositories/Requests [16:09:39] Krinkle: was can ask him to leave a comment in the future. The reassignment occurred after a meeting with him [16:10:37] that's fine, just thought it was maybe a vandal. We get some of those and they looked rather similar (fairly anonymous/non-descript account, little to no activity in core or wmf repos, and then this kind of change). [16:12:59] <_joe_> Krinkle: I wanted to start sending 1 out of 1000 users to php7, do you see any blockers? [16:15:41] _joe_: we don't have the sampling profiler set up yet. (we=I) [16:16:23] We can start boiling ourselves, but we need to stop at some point, if we want to be able to identify parts that are slower and on-going regressions unrelated to php7 (the new normal). [16:16:36] e.g. week over week etc. [16:18:50] <_joe_> yeah I thought we could start with a low sampling rate (1 in 1000) and we fix profiling in the meanwhile. [16:19:01] <_joe_> btw, how can I help? [16:19:14] <_joe_> I would love to do it this week as I'm off next week [16:19:38] <_joe_> but if you can't work on this, it's ok, it can wait until I'm back [16:24:28] <_joe_> then I have to find a better way to find php7 errors in logstash than having to specify AND NOT phpversion:5.6.99-hhvm [16:26:45] _joe_: next step is https://gerrit.wikimedia.org/r/#/c/operations/puppet/+/493247/ [16:27:23] need to generalise it a bit and create a second instance of arclamp that consumees from the excimer redis topic [16:27:51] <_joe_> Krinkle: I can take a stab at it if you're too busy [16:29:11] _joe_: Sure, that'd be great. [16:29:35] The plan right now is to rename /arclamp to /xenon, then change xenon-log.py to accept some kind of parameter that will tag the log and svg files with, by default none (which is for current xenon), and for the second instance we'd set it to "excimer". [16:29:37] So under https://performance.wikimedia.org/xenon/ [16:30:07] we'd get 2019-03-04.index.log (xenon) and 2019-03-04.index.excimer.log [16:30:10] or some such. [16:30:22] <_joe_> ack [16:30:31] <_joe_> where is xenon-log.py? puppet? [16:30:32] and then once that is working without breaking xenon, puppetise the second instance of it that consumes from that redis topic. [16:30:34] Yes [16:30:40] <_joe_> ack [16:30:55] I want to move it to performance/arclamp.git at some point (exists, but unused), but that's for another time I think. [19:17:47] * duesen does the inbox zero dance [22:05:28] hm, how do I tell jenkins to run a specific job, like mediawiki-quibble-vendor-sqlite-php70-docker, that'S usually only run on the gate pipeline? [22:11:20] you can't make it run just one job [22:11:27] specifically that job seems to be part of the experimental pipeline so you can comment 'check experimental' to trigger it [22:12:15] in general I don't think it's possible; you could add a 'check gate-and-submit' command to the CI config