[00:00:21] great [00:00:36] test build passed now that the line length is not angering it [00:01:34] i'll circle back tn and run PCC on some hosts and verify the puppet catalog changes in the way we expect (should be pretty simple, just the net-new file resource [per instance] and the slightly different `elasticsearch_6%2540.systemd.erb` contents) [09:29:44] mpham: I won't be able to make it to the "WMF/WMDE query service check-in". Can you do it without me? [11:47:19] i can't seem to clone the repo git@gitlab.wikimedia.org:mhurd/mediawiki-cirrus-docker.git [11:47:35] and i added my ssh key to gitlab [11:50:29] nvm figured it out [14:01:53] greetings [14:29:05] Trey314159: I'll be a few minutes late [14:31:59] Trey314159: and I'm there! [14:42:24] gehel: yeah, i can. Lydia is restarting the check ins between the two teams since those meetings seemed to have disappeared [14:43:54] sorry, been here for awhile [15:10:43] gehel or anyone else, lmk if you have time to show me the 'check experimental' magic, don't seem to be running on https://gerrit.wikimedia.org/r/c/operations/puppet/+/766876/ or maybe I just don't remember where to check? [15:11:44] ah, I see it on https://integration.wikimedia.org/zuul/ but it just links back to the patch, no output [15:15:03] inflatador: in zuul, if you click on the job it opens and has a link to the jenkins build: https://integration.wikimedia.org/ci/job/operations-puppet-catalog-compiler-test/1207/ [15:15:55] I think it should add a comment on the patch when done [15:17:29] gehel gotcha, looks like you have to click on the blue bar, then it'll show you the job [15:19:35] note that you should probably add a "Host:" entry on the patch comment. Here, you seem to be compiling against all servers, which is probably not needed and taking way more time than selecting either a few hosts of each relevant clusters, or a cumin selector for the impacted classes / profiles. [15:19:50] * gehel is in meeting, going back to not reading IRC (at least trying) [15:20:29] you have intermediate results already in https://puppet-compiler.wmflabs.org/pcc-worker1003/1207/ [15:20:37] https://wikitech.wikimedia.org/wiki/Help:Puppet-compiler#Gerrit_integration [15:22:58] note for later: maybe check-experimental with no "Hosts:" should error instead of running on all nodes [15:29:58] +1 for me :) [15:32:08] oh yeah , thanks for the host-targeting advice. Dropping off kids but will check the job when I get back in ~20 [15:51:39] \o [15:51:49] still running. I understand what you are saying re: limiting hosts but I don't remember it ever running against all hosts before, is that just because this particular code path is used by more services? [15:52:47] wow, at first i didn't believe you but opening the zuul console output for this..yea it's doing them all :P [15:53:16] i also thought before it had failed instead of running against everything [15:54:14] I've never limited the amount of hosts before, and it's never tried to run against everything before either. It's not like I've been around long, but I still don't understand why this is happening [15:55:25] it's indeed odd, doesn't seem like an intended behaviour [15:56:08] yes, historically not specifying hostslist in the pcc means run on all site.pp picking one host per node block [15:56:19] you can interrupt the run if you want [15:56:58] volans sure, just aborted [16:01:41] OK, trying again with Hosts: specified as in volans ' link above [16:01:56] you can also use a subset of cumin query capabilities [16:02:03] docs should be there too [16:05:15] OK, looks like that was a success! Thanks for the advice [18:03:33] Quick errand, back in ~15-20 [18:41:40] sorry, been back, but now I must eat lunch! See you in ~30m [18:51:46] :) [19:03:30] does anybody know where zbyszko uploaded the wcqs usage data for the last 14 days? [19:09:40] mpham: if i was going to guess, https://drive.google.com/drive/u/1/folders/1ojrcehL7Bz0Cc4wKgdtD8CccruNK8lyh sort by last midified, there are two files dates 2022_02_28 [19:09:59] * ebernhardson fails at english yet again...some day [19:10:53] thanks! [19:11:41] or "thonks" to keep the theme [19:12:59] aaand back [19:17:57] Please let me know when we take down WCQS beta 1 today so I can get the announcement out about it [19:25:38] i can switch it in a couple minutes, sec [19:27:45] * ebernhardson has to lookup how to get a redirect instead of a proxy [19:33:55] mpham: it redirects now, although i'll have to check back in with gehel later as i didn't see how to accomplish this from the horizon interface. It's doing a redirect from the instance now [19:34:44] thanks ebernhardson . I'll draft up the comms and send them out, since it sounds like it effectively redirects for the user now either way [19:34:50] yup [20:48:50] hmm, i guess i didn't say i was taking lunch. but back now [21:04:13] (╯°□°）╯︵ ┻━┻ ＴＨＩＳ　ＩＳ　ＲＩＤＩＣＵＬＯＵＳ [21:04:18] ;P [21:21:37] * ebernhardson is surprised to find Special:Nearby usage mostly disapeared in july 2019, but we still do 4-8M geo searches per day [21:21:48] i guess it's primarily from the mobile apps [22:58:18] meh, my isp seems to have broken it's routes to wiki. mtr doesn't leave the ISP's AS # [23:15:59] ebernhardson: there was just a ulsfo outage [23:28:20] until tomorrow...keep it searchy! [23:43:31] going to depool wcqs2001, so i can make it's token expiration super short and try to reproduce T301650 without waiting multiple hours between tests [23:43:32] T301650: WCQS "Application Connection Error" E009 - https://phabricator.wikimedia.org/T301650 [23:47:04] err, sigh. I should have looked at the code first, we read an integer and denominate it in hours :( [23:47:30] i guess instead it's figuring out if i can make the mw-oauth integration environment trigger the error