") in the message on Article Feedback Activity Log on testing. It should be: ( | ) hid <'feedback post' # ID> on

() [18:46:49] rsterbin: Was this meant to do other things? The commit summary mentions changes to hooks.php etc [18:46:52] fabriceflorin: We know [18:47:40] fabriceflorin: Is it OK if I deploy the e-mail switch /now/ ? [18:47:51] (Settings pasted above for confirmation) [18:48:08] RoanKattouw: that's the new custom formatting method [18:48:21] the whole commit dropped out of the deploy squash [18:48:23] Also, it is using the word 'feedback' twice ('feedback feedback post'). This is not a show-stopper for this deployment, but I do want to point it out. Here is the requirements page for this request (check Column D for the central log): https://docs.google.com/a/wikimedia.org/spreadsheet/ccc?key=0Aq_75_5y5sKWdDVDNjJNejRUYW9yaU82WDAzQ21ZWGc#gid=0 [18:48:28] on our end, not, yours, that is [18:48:43] rsterbin: What I mean is, the commit summary mentions changes to AFTv5.hooks.php and AFTv5.php , but the commit itself only changes .i18n.php [18:48:49] fabriceflorin: those messages have not yet been fixed [18:48:52] aaaaaa [18:49:25] well [18:49:29] RoanKattouw: Yes, please switch the email address for oversight requests from aft@wikimedia.org to oversight-en-wp@wikipedia.org . [18:49:33] So either the summary is wrong or you're missing stuff [18:49:37] fabriceflorin: OK, making that switch now [18:50:38] RoanKattouw: i'm in master, and the method's there. maybe it *was* part of the squash commit, and git just applied the changes in the wrong order and lost the i18n ones? [18:50:54] Possile [18:51:14] the stuff is being pushed through the custom formatter - the messages are just wrong ;) [18:51:21] rsterbin: meta:R:AFT/Clicktracking | sed 's/S3/SE/g' ---> completed [18:51:23] If everything's there and only the i18n changes are missing, then please fix the commit summary to describe what the commit really does :) [18:51:38] RoanKattouw: i'm on master locally and seeing the right thing, so i think that's okay [18:51:43] ok [18:51:43] Ok [18:53:33] RoanKattouw: fixed message [18:53:44] *commit message, not i18n message :-P [18:56:43] RoanKattouw -- https://gerrit.wikimedia.org/r/5346 updates the bucketing for links, as per DarTar's question earlier; however, it's a config change, so if you just wanted to do it in LocalSettings, you could. [18:56:55] RoanKattouw, wanted to confirm that the email switch was done to production, so I can ping Philippe and have him check it there. [18:57:18] rsterbin: thanks [18:57:23] np [18:58:25] does a version update force rebucketing? [18:58:33] DarTar: yep [18:58:36] cool [18:59:02] fabriceflorin: Sorry yes it's done [18:59:10] Now deploying Reha's i18n fix [19:01:42] Thanks, RoanKattouw! I will test the email switch on production and send a note to Philippe to have him check it. [19:02:07] Rebuilding l10n cache now [19:06:03] rsterbin: Is https://gerrit.wikimedia.org/r/5346 meant to be included too? [19:06:22] yeah, that's the config change i was talking about above [19:06:35] OK [19:13:29] OK, roankattouw, rsterbin and auroraeosrose, how are we coming along with the deployment on your end? What should I be testing next? We only have 15 mins. to go on this deployment, technically. [19:13:43] I'll push Reha's latest fix [19:17:06] fabriceflorin: the messages in the logs should now be correct [19:17:22] Yes, i18n cache rebuild is done [19:17:28] Sorry, I was talking to my boss for a minute [19:17:33] Now pushing Reha's fix [19:18:53] Cool. I assume that is still on testing, right? rsterbin, can you please let me know what should be tested? [19:19:09] the fix was for bucketing [19:19:18] (of the link options, that is) [19:19:44] when Roan's done, you should only get link option E if you don't specify one [19:20:10] unless you're on a protected page and getting the option 4 fallback, of course [19:20:29] or if you're getting option 0 / no form [19:20:33] Got it. So this bucketing fix would disable the feedback link Option A and Option X, and only show Option 4, right? [19:21:33] OK Reha's fix should be live now [19:21:34] Option E [19:23:27] auroraeosrose and RoanKattouw: Thanks for fixing the incomplete messages on the central activity logs. I can confirm that these messages are correct on testing. [19:25:33] fabriceflorin: I am ready for a production deploy once you approve [19:26:39] RoanKattouw, rsterbin, fabriceflorin: confirmed that the above fix buckets users as we expect [19:26:43] I just tested all 6 links on to the feedback form on testing and can confirm that only feedback link Option E is displayed (not A or X), and that it does not appear when it's not needed (when there is no feedback form, for example). [19:26:53] cool [19:27:00] my tests are all good as well [19:27:20] fabriceflorin: are we cool to go to prod, then? [19:28:37] RoanKattouw: From my perspective, we are good to go to production. I just checked with DarTar, and he's OK with this as well. [19:29:24] OK, then let's go [19:29:42] yep, good to go [19:29:45] Excellent. Ready to test production when you give us the word. [19:37:20] fabriceflorin: Production deployment done [19:37:27] woo-hoo [19:37:31] off to test [19:37:35] Schema changes and SQL update scripts also done [19:37:59] RoanKattouw: Thanks! Testing right now. [19:38:01] except for the DROP COLUMN parts, running those now [19:38:10] fabriceflorin, dartar: don't forget about the caching delay :) [19:38:18] sure [19:38:25] and your browsers' cache, too :) [19:38:29] yep [19:38:32] DROP COLUMN done [19:38:43] I guess I could go grab a sandwich then [19:40:07] DarTar: I bring in lunch from home on deployment days :) [19:41:02] yeah I was waiting for fabriceflorin's blessing but I took silence = consent ;) [19:41:07] I'll be back in 15 mins anyway [19:42:09] Ok, that's a good idea to wait. Let's go get lunch, and come back in 30 mins. [19:42:31] To make it easier when we get back, here are the links to an *unprotected* page on production: [19:42:31] http://en.wikipedia.org/wiki/Golden-crowned_Sparrow?aftv5_form=1 [19:42:33] http://en.wikipedia.org/wiki/Golden-crowned_Sparrow?aftv5_form=4 [19:42:34] http://en.wikipedia.org/wiki/Golden-crowned_Sparrow?aftv5_form=0 [19:42:36] And here are the links to a *semi-protected* page on production: (for anyone that is not an auto-confirmed editor or logged in): [19:42:37] http://en.wikipedia.org/wiki/Barack_Obama?aftv5_form=1 [19:42:38] http://en.wikipedia.org/wiki/Barack_Obama?aftv5_form=4 [19:42:39] http://en.wikipedia.org/wiki/Barack_Obama?aftv5_form=0 [19:54:45] New patchset: Hashar; "Jenkins 1.458 can not fingerprint .git/FETCH_HEAD" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/5351 [19:55:12] https://gerrit.wikimedia.org/r/5351 [19:55:17] New review: Hashar; "(no comment)" [integration/jenkins] (master); V: 1 C: 2; - https://gerrit.wikimedia.org/r/5351 [19:55:19] Change merged: Hashar; [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/5351 [20:25:27] RoanKattouw, rsterbin, auroraeosrose and DarTar: I am back from lunch and testing the 6 links I posted earlier on production. I'm sorry to say that even after clearing my caches, Option 4 is not displaying correctly yet at this URL: http://en.wikipedia.org/wiki/Golden-crowned_Sparrow?aftv5_form=4 (it should be showing a call to edit, not the feedback form. Am I using the right URL? Should I do anything else besides clearing my cache? [20:26:30] something didn't deploy right [20:26:38] Same problem with the Obama page: http://en.wikipedia.org/wiki/Golden-crowned_Sparrow?aftv5_form=4 [20:26:39] I"m getting feedback errors on teh feedback page [20:29:51] fabriceflorin: confirmed here on a browser with no cache [20:30:32] DarTar, are you experiencing the same issue? Should we do anything else besides clearing our individual caches? how about clearing the server caches? Would that be needed? [20:31:24] that's the only explanation that I have, what we are seeing client-side is not a combination of old and new designs, it's just the plain old design (the clicktracking codes are for Option1A) [20:32:56] auroraeosrose: I can confirm the problem you report for the feedback page as well. It shows me this error message: 'Error loading feedback' - http://en.wikipedia.org/wiki/Special:ArticleFeedbackv5/Golden-crowned_Sparrow [20:35:33] And the article feedback log is also showing the old format: [20:35:33] https://en.wikipedia.org/w/index.php?title=Special%3ALog&type=articlefeedbackv5&user=&page=&offender=&year=&month=-1&tagfilter= [20:35:44] RoanKattouw, any suggestions? [20:35:53] we're getting the first post-deployment events in the clicktracking log, meaning that someone is seeing it [20:36:14] how long does it take a code push to filter through? it looks like the code isn't done [20:36:21] also I dont' think the filter_counts script was run [20:36:25] there is still some weirdness, like event prefixes such as: 2-optionSE_4X_edit which we shouldn't be seeing [20:36:50] (there should be only a 2-optionSE_4E_edit) [20:40:55] RoanKattouw? [20:41:57] there's no record yet in aft_article_feedback with the new experiment IDs [20:42:15] which I imagine will be prefixed with SE_, right? rsterbin ? [20:43:11] as in, SE_4E, SE_1E, SE_0X [20:43:31] yes, that's correct [20:43:48] It sounds like the deployment didn't go through properly. Let's wait for RoanKattouw to return. [20:43:49] right, count is 0 for all of these [20:50:47] Hey TrevorParscal, can you see RoanKattouw anywhere on the 3rd floor? We need his help on an AFT deployment that's gone wrong. [20:51:01] fabriceflorin: Oh sorry. This is Roan at Trevor's computer [20:51:04] Reading backscroll [20:51:12] doesn't look like the code is live... [20:51:51] It should be.. [20:51:57] Let me (Roan) get back to my desk and figure this out [20:52:26] Thanks, TrevorParscal and RoanKattouw, much appreciated. [20:53:43] TrevorParscal (Roan), there's a bunch of new events in the log which clearly come from the new code, but the AFT table has no new record yet [20:54:30] DarTar: kinda confirms that php code is there but JS and CSS have not updated [20:55:07] OK I'm here [20:55:14] Let me see if we have any DB errors [20:55:45] Unknown column 'af_hide_user_id' in 'field list' [20:55:50] yeah, that's old cod [20:55:57] it looks like the code did not get pushed [20:56:04] the log messages haven't changed either [20:56:08] wtf you're right that is old code [20:56:09] Weeeird [20:57:24] Trying to redeploy now [20:58:52] Thanks, RoanKattouw, let us know when you think the code has been pushed. [21:03:14] WTF, this happened because scap is broken in some way [21:04:42] RoanKattouw: fwiw, wikisource doesn't seem to have that patch you mentioned earlier either: https://en.wikisource.org/w/index.php?title=Lucas,_Charles_(1769-1854)_(DNB00) [21:04:49] Yeah [21:04:51] It's going out now [21:04:54] * RoanKattouw stabs scap [21:05:08] What's scap? [21:05:18] The main deploy script [21:05:35] I'm wondering if there's some other version out there that gets run instead, but that doesn't seem to be the case [21:05:46] Got it, thanks. [21:05:51] The deploy script just stops three steps before the end, right before the "actually push code to the servers now" step [21:06:33] This is gonna take a while [21:06:46] The first set of 30 servers is finishing up now, and that's like 1/10th of the cluster [21:07:41] Oh, there's only 200 servers on the mediawiki-installation list, not that bad then [21:07:46] It'll take 7 cycles [21:08:44] Thanks, RoanKattouw, I really appreciate your going the extra mile on this. Let us know when you think would be a good time to test. [21:09:14] * RoanKattouw is just pissed that the deployment infrastructure is apparently broken [21:11:23] Yeah, I feel your pain. Hang in there. [21:15:45] Option 4 is now working for me on production: http://en.wikipedia.org/wiki/Golden-crowned_Sparrow?aftv5_form=4 [21:16:14] So is Option 0: http://en.wikipedia.org/wiki/Golden-crowned_Sparrow?aftv5_form=0 [21:18:50] The sync is still running [21:18:55] So you may not get consistent results [21:18:57] And the Option 4 is working as intended (Learn more) on a controversial page like Obama: http://en.wikipedia.org/wiki/Barack_Obama?aftv5_form=4 [21:19:51] So is this Option 0 on Obama semi-protected: http://en.wikipedia.org/wiki/Barack_Obama?aftv5_form=0 [21:25:01] The new feedback page is also working for me now: http://en.wikipedia.org/wiki/Special:ArticleFeedbackv5/Golden-crowned_Sparrow -- and the Activity log seems correct. [21:25:38] And the activity log seems to be working as well for me: https://en.wikipedia.org/w/index.php?title=Special%3ALog&type=articlefeedbackv5 [21:26:41] And the suppression log is also working for me, displaying this correct info: 11:04, 19 April 2012 Aft5hide (talk | contribs) requested oversight for feedback post #86969 on Barack Obama (Testing that oversight requests are now correctly sent to oversighters. This feedback post deserves to be hidden, though it's not oversight material, IMHO.) [21:27:10] So from my standpoint, we are good to go! Can you guys reproduce as well? [21:27:12] rsterbin: I see the 4E fallback on protected pages generates an optionSE_4X_learn_more clicktracking prefix, that's not as specced but it makes sense as long as it's consistent [21:27:36] OK the deploy is essentially done now [21:27:48] It's still updating snapshot1001-1004 but those boxes don't serve web requests [21:27:51] ok [21:28:28] DarTar: huh, i thought that was the right one [21:28:47] according to the specs it should have been optionSE_4E_learn_more but I am happy to update the page on meta to reflect the new prefix (we'll just have to keep this in mind if we want to grep on optionSE_4 events) [21:29:22] not a big deal, it's just that optionSE_4* users all come from the same bucket [21:29:27] yeah, there should never be a 4E -- it rewrites to X if it shouldn't have one [21:29:32] a feedback link [21:29:55] er, a 4E_learn_more [21:30:07] right [21:30:15] makes sense, updating the specs on Meta [21:30:27] thanks [21:31:09] optionSE_4E-init will still count all Option 4 combined, right? [21:31:59] * DarTar looking at http://meta.wikimedia.org/wiki/Research:Article_feedback/Clicktracking#Stage_3_.28Impact_on_engagement.29 [21:33:09] clicktracking data coming through [21:34:53] oh, no, there will be some optionSE_4X-init, because it switches to X before it logs the init [21:34:57] uuugh [21:35:10] ha [21:35:29] ok, there should be no problem as long as I know what we call what :) [21:35:39] do you want to take a stab at that page? [21:35:56] you mean updating the clicktracking page? [21:36:09] yeah – I don't know what to expect for other events like optionSE_4E-noedit-init [21:36:30] I imagine this will be optionSE_4X-noedit-init, right? [21:39:47] robla: https://www.mediawiki.org/w/index.php?title=MediaWiki_1.20/Roadmap&diff=527601&oldid=527216 -> 1.20wmf1 deployment to non-wikipedias is not yet done, there are several wikimedia domains still on 1.19, such as wikimania wikis, incubatorwiki and outreachwiki [21:40:04] yep [21:40:10] i [21:40:14] 'm updating now [21:40:17] thx [21:41:07] all set [21:46:55] thanks, there's just one missing (optionSE_4E_learn_more-impression-bottom) which I'll fix as it will also use a 4X [21:49:12] SPQRobin: feel free to update that status based on what you know. [21:57:49] thanks [21:58:47] RoanKattouw, rsterbin, auroraeosrose and yoni_omniti: So are we all comfortable with the AFT deployment? DarTar confirms we are getting the data he wanted. Works on my end as well. Anyone else found any issues? [21:59:33] no issues here [22:01:07] robla: ok, I updated the page, but can't those wikis just be upgraded? [22:02:54] I think we're good, thanks everybody [22:03:21] Looks fine to me [22:03:26] I apologize for the deploy script breakage [22:06:31] fabriceflorin: if we tweet, we actually shouldn't say anything specific about the test or we risk to affect the results [22:07:31] I'll keep it generic [22:07:47] Git experts around? How do I get details of a specific revision from the web interface? [22:10:41] Wonderful. Nice work, everyone! And special thanks to RoanKattouw for going beyond the call of duty to get through all this, despite the recalcitrant scap ;o) [22:11:20] I agree with DarTar that we shouldn't tweet about the exact nature of the test. [22:11:43] OK, everyone, I'm calling it a day. Thanks for all your good work! [22:11:50] Over and out. [22:15:21] vvv: Through gitweb, hang on [22:15:39] vvv: Through URLs like this https://gerrit.wikimedia.org/r/gitweb?p=mediawiki/extensions/Collection.git;a=commitdiff;h=109dbc76b80dd9cd8447396f493c07bb4557e8f9 [22:15:58] (And yes, you need to know both the repository name and the commit hash, that's unavoidable) [22:17:19] RoanKattouw: good; intuitive, also [22:17:23] Thanks [23:00:59] let's see who's here for the 20% checkin [23:01:21] probably no one since I never really set it up or pushed for it before [23:01:55] Friday UTC/late Thursday SF - 00:00 UTC (4pm PST) [23:04:17] sumanah: its only 23:00 UTC :-) [23:08:31] yeah [23:08:34] (never mind) [23:41:44] hi chrismcmahon - got a moment? [23:42:03] hi sumanah I have about 15 [23:42:07] ok! [23:42:23] TrevorParscal: git reset --hard fd49e8df32be3d08ac02843fd5df4620d6c3421f [23:42:26] basically, chrismcmahon, I made contact with Joshua Van Davier, QA associate in the fundraising department [23:42:38] chrismcmahon: and he will want to chat with you while you are in SF next. [23:42:44] chrismcmahon: basically to pick your brain about QA practice. [23:42:59] sumanah: great! I'll be sure to set that up if he doesn't [23:43:29] chrismcmahon: fantastic. [23:43:37] That is all. [23:43:56] chrismcmahon: He's jvandavier (in case you want to write him email, schedule a meeting, etc) [23:46:29] sumanah: Do you know anything about the bots project on Wikimedia Labs? [23:46:46] kaldari: I don't think I do, no, but if you have a question I might be able to direct you [23:47:56] well my main questions were, can I join the group, and do they have some sort of English Wikipedia db mirror for the bots to use? [23:48:22] But it looks like none of the labs people are online [23:48:25] I think the answer to the latter question is "no" [23:48:34] kaldari: Faidon is a new Labs person and is here on the 3rd floor [23:48:58] kaldari: you can almost certainly join the group. You might wish to join the labs-l mailing list so you can ask for such things asynchronously [23:49:29] Someone should add Faidon to this list then: https://labsconsole.wikimedia.org/wiki/Help:Access#Access_FAQ [23:49:46] kaldari: can you? or do I need to? [23:50:06] I don't know any of his info, but I could probably look it up I suppose [23:51:15] Is this the same Faidon that is supposed to be in Athens (according to the contact list)? [23:51:25] He does live there, but he is in SF right now [23:51:29] ah cool [23:51:41] OK, I'll add his info then [23:56:13] TimStarling: hi! any word on the Score extension? [23:58:17] I'm going to look at it today [23:59:59] Got it, thanks TimStarling. Also I consolidated some of our discussion about the current status & plans for Lua/Scribunto into https://www.mediawiki.org/wiki/Lua_scripting#Status - hope it's accurate