[13:40:29] RobH: hi! you awake? [13:40:45] did you get my mail about the server numbers, etc? [14:04:08] Daniel_WMDE: yep, trying to pull it up now [14:04:15] since our mail migration, searching my inbox takes minutes. [14:04:17] =[ [14:21:26] RoanKattouw: good morning [14:21:32] ready to deploy stuff? [14:23:15] Alright [14:24:04] TrevorParscal: You don't need any other core stuff, do you? [14:24:25] no [14:24:37] TrevorParscal: Also, wasn't there some config change that was needed to not make the tracking weights 0:100 against? [14:24:46] i just did that [14:24:49] on prototype [14:24:50] also [14:24:56] the email message not being changed [14:25:16] that's because AF configures EC and it must be included EC,AF not AF,EC as it was on prototype [14:25:17] OK could you copy that config change here then? [14:25:21] Oh OK [14:25:30] Let me verify that the order is correct on WMF [14:25:52] It is [14:26:08] one last thing [14:26:09] hey guys [14:26:10] EmailCapture is on line 1082 and ArticleFeedback on line 2287 (CommonSettings.php) [14:26:13] *RoanKattouw documents this [14:26:19] is there a way to customize the title of a special page? [14:26:36] the confirmation page says "Email Capture" at the top, which is confusing and aggressive [14:26:39] Sure [14:26:45] Just change the message [14:26:48] Also, title case == bad [14:27:04] it actually says "E-mail address capture" [14:27:38] I was saying changing it on a per-view basis using logic... [14:27:47] I guess using "Email verification" would work well [14:27:50] for all cases [14:28:11] Howie? [14:28:28] yes [14:28:31] ok [14:28:42] as long as the user gets an obvious "thank you" message when verificadtion is done [14:29:10] which should be in the body of the page [14:29:51] yeah, it says "Your e-mail address was successfully verified." [14:29:58] awesome [14:30:53] TrevorParscal: OK are you done now? [14:31:05] damn it, i never know what channel i'm looking at in colloquy - i thought i was chatting with you guys in wikimedia-dev... :-/ [14:31:31] RoanKattouw: did you see that last i18n change? [14:31:39] Yes [14:32:04] Is that the last change you needed? [14:32:13] If so awjr and I can get started [14:33:38] RoanKattouw: can you merge that to wmf1 so I can verify it on prototype? [14:33:49] EmailCapture is running from deployment branch on prototype [14:34:05] nice - i think i just got bucketed into the no expertise group [14:34:34] TrevorParscal: Then switch it to trunk [14:34:39] :( [14:34:47] you have to merge it anyways [14:34:56] *TrevorParscal types long switch statement [14:37:47] ok [14:37:50] looking good [14:39:18] howief: can we get your blessing? [14:39:34] i'd like to confirm the a/b behavior [14:39:57] also howief, you're email provider may have been spam-blocking the emails - hotmail thought it was spam at first until I trained it... [14:39:58] who's getting which version? [14:40:14] howief: I get a random version when I remove my cookie [14:40:16] yeah - my gmail was filtering the messages [14:40:22] I've tested in about 10 times [14:40:26] awesome [14:40:31] and sure enough, half of the time I see the options [14:40:34] Alright, we all good? [14:40:59] prototype looks good, but is this on test? [14:41:06] not yet [14:41:14] are we going to put this one test? [14:41:17] on test [14:41:29] Yes, if you say we can [14:41:30] if you're satisifed with how things are on prototype, yes [14:41:43] ah - that's what you're asking -- sorry [14:41:46] We've been practically begging you to greenlight just that for 10 mins now :) [14:41:46] :) [14:41:50] yeah, everything looks awesome on prototype [14:42:07] w00t [14:42:08] sorry i thought you were asking about pushing to production [14:42:21] Only after we confirm test is fine [14:42:32] sorry for the ambiguity, test is one of the first steps in getting it out to prod [14:42:58] RoanKattouw: is everything in the deployment branch or shall i merge? [14:43:04] You should merge [14:43:19] But you're updating an existing extension (two actually) two trunk state, so this is a little bit different [14:43:33] k [14:45:40] OK so [14:45:48] If you look at this https://secure.wikimedia.org/wikipedia/mediawiki/w/index.php?path=%2Fbranches%2Fwmf%2F1.17wmf1%2Fextensions%2FArticleFeedback&title=Special%3ACode%2FMediaWiki%2Fpath [14:46:18] You'll see that ArticleFeedback was most recently updated to trunk state in r86085 (path creations are not shown, must be a bug in CodeReview) [14:46:34] There is one cherry-picked MFT after that so let's hope it doesn't cause too much trouble [14:46:48] gulp [14:46:59] What you need to do is go to your /path/to/your/checkout/of/branches/wmf/1.17wmf1/extensions/ArticleFeedback [14:47:19] And run svn merge -r 86085:HEAD http://svn.wikimedia.org/svnroot/mediawiki/trunk/extensions/ArticleFeedback . [14:47:34] (You may replace the URL with the filesystem path of your trunk AFT checkout) [14:47:47] i've got a local copy of AFT from trunk, i like to do a dry run first [14:47:55] This is a dry run [14:48:01] Merges don't get committed immediately [14:48:09] if you do url -> url they do [14:48:11] oh i see [14:48:13] that's url -> wc [14:48:14] nm [14:48:17] This is URL->fs [14:48:19] yeah [14:48:28] (the dot at the end was part of the command) [14:48:40] ok - wait, 86085? it looks like the last r86382 [14:49:12] Yeah that's just a single cherry-pick [14:49:21] The last wholesale update-everything was in 86085 [14:49:26] ah i see [14:49:27] ok [14:52:45] ok RoanKattouw, AFT is merged [14:52:53] Alrighht [14:53:02] now the same with Emailcapture? [14:53:04] Sanity-check the diff for any conflict markers, then commit it [14:53:19] With a commit message like "1.17wmf1: Merge ArticleFeedback to trunk state" or something similar [14:53:19] already committed [14:53:28] Ah yes [14:53:30] Thanks [14:53:32] Now EmailCapture, yes [14:53:48] CodeReview is unhelpful finding the original rev, because creations aren't listed [14:54:04] But it's 86624 [14:54:12] yeah, im just looking in viewvc [14:55:39] ok EmailCapture merged and checked in [14:56:03] Nice [14:56:09] OK now run svn up on fenari [14:56:18] It should also pick up my MFT of the jquery.json stuff from last night [14:57:03] oh, then should i svn up from docroot? [14:57:19] i was just up'ing the extensions themselves [14:57:55] Yes, just svn up from the docroot [14:58:07] Usually that was a good idea because you don't want collateral updates [14:58:11] Except this time you do want them [14:58:42] :p [14:58:52] doh locked [14:59:01] cleaningup [14:59:44] TrevorParscal: What's that config change we need again? [14:59:51] A resources/jquery/jquery.json.js [14:59:51] UU resources/Resources.php [14:59:51] UU resources/mediawiki/mediawiki.js [14:59:57] RoanKattouw: look right? ^ [15:00:03] Looks right [15:00:23] cool - we need to make sure email capture is configured before aft [15:00:30] It is [15:00:35] By like a thousand lines [15:00:43] well shucks [15:00:45] I checked and added a comment saying it has to stay that way [15:00:50] great [15:01:15] RoanKattouw: http://pastebin.com/XudSZ3d8 [15:01:19] that's what's on prototype anyways [15:01:29] the categories on the cluster are different [15:01:33] es [15:01:40] but the buckets and version thing need to be changed to match that [15:01:40] We just need to copy the buckets part [15:01:45] And the version I guess, yeah [15:01:54] and we aalso need to to add something for the tracking [15:01:59] awjr: Copy the buckets and version thingies from that pastebin to CommonSettings [15:02:01] howief: are we wanting to track everyone? [15:02:04] k [15:02:07] yes please [15:02:44] can we track everyone without compromising performance? [15:03:11] For now we should be able to [15:03:17] As we ramp up we'll want to drop that [15:03:18] ok [15:03:23] are we getting rid of lottery odds and using 'feedback options' instead? [15:03:29] (When was the ramp-up thing supposed to be again?) [15:03:49] ramp-up is scheduled for the end of next week [15:04:07] also, do these need to change: [15:04:08] 'wmgArticleFeedbackCategories' => array( [15:04:09] 'default' => array(), [15:04:09] 'testwiki' => array( 'Really big category' , 'Article Feedback Pilot' ), [15:04:09] 'enwiki' => array( 'Article Feedback Pilot', 'Article Feedback', 'Article Feedback Additional Articles' ), [15:04:09] ), [15:04:35] so, more like this then http://pastebin.com/fuUX1MLf [15:04:58] awjr: lottery odds aren't for bucketing users, it's for articles [15:05:16] k [15:05:36] brb [15:07:06] RoanKattouw: i should translate those configurations into per-wiki configs in InitialiseSettings? [15:07:27] No [15:07:34] Just set them in CommonSettings for now [15:07:38] ok [15:09:03] done [15:09:14] now we test on test? [15:10:00] Yes [15:10:10] back [15:10:18] test.wp.o should already be running the new code and config [15:10:32] It may need syncing, I'm not sure when that occurs exactly [15:11:13] it doesnt look like it's running the new stuff [15:12:11] I saw an e-mail prompt [15:12:13] caching? [15:12:16] Except I'm not receiving the e-mail [15:12:23] OK awjr let's sync test by hand then [15:12:29] did we need to make any schema changes? [15:12:30] is email working on test? [15:12:32] ssh into srv193 and run ~catrope/sync-common [15:12:44] from fenari? [15:13:20] running [15:13:41] (I mean run that command on srv193) [15:13:59] It'll give you half a screenful of errors but they can usually be ignored [15:14:07] yeah, time set errors [15:14:07] My e-mail address is in the emailcapture table [15:14:16] done [15:14:20] Thanks [15:14:32] so now test is sync'd with fenari? [15:14:33] things look fine on chrome [15:14:41] Yes [15:14:47] That's usually automatic [15:14:52] But sometimes it needs kicking [15:14:58] I'm not entirely sure when it does and when it doesn't [15:15:00] cool [15:15:03] lol [15:15:06] k [15:15:47] BTW guys, the whole greying out of the e-mail address and coloring the box green/red doesn't work on IE [15:15:51] *in IE [15:16:19] Oh hey it briefly becomes green when I submit the form, then it disappears [15:16:43] RoanKattouw: that behavior is not idea, but it's not going to block deployment [15:17:24] we do need to figure out how to grey the email address since right now it looks like there's something already there [15:18:01] OK e-mail verification works for me [15:18:09] ok, things appear to be working fine for me on test - are we set to move to production? [15:18:21] have we spot checked the major os/browser combo's? [15:18:28] chrome/osx works fine [15:18:59] i can dbl check ff on os x [15:19:07] and safari [15:19:08] (checking chrome on xp) [15:19:35] can someone take safari on osx and ff on win7/xp? [15:20:21] im not seeing the checkboxes in FF on os x [15:20:40] TrevorParscal: if i shift-refresh, will that always give me the expertise checkboxes? [15:20:59] I don't think so, because of cookies [15:21:10] no, you have to be in the right bucket [15:21:14] hmmmm [15:21:31] that's what i thought the behavior should be, but i seem to always get the checkboxes when i shift-refresh [15:21:46] let me try a few more combos [15:22:01] well, you are in the options show bucket [15:22:03] the bucket! is refreshing a bunch of times enough to eventually get me in the bucket? [15:22:23] no [15:22:26] i don't think so since the bucketing happens by cookie [15:22:33] Exactly [15:22:35] but i'm on the same browser [15:22:38] clearing cookies and getting lucky will [15:22:40] and when i shift refresh [15:22:40] refreshing + clearing cookies then? [15:22:48] So you have to remove the cookie or using a different browser to be rebucketed [15:22:55] yes,but not on that order [15:22:59] sometimes i get the checkboxes and sometimes i don';t [15:23:06] there we go [15:23:20] you should always get them or not [15:23:29] unless you clear cookies [15:23:33] ok ff on os x looks good, checking sfari [15:23:34] that's not whawt i'm observing [15:23:54] can someone reproduce what i'm seeing? [15:24:12] (i.e., shift-refresh a bunch of times; sometime checkboxes sometimes not) [15:24:50] what browser? [15:24:51] Well, shit, yeah [15:24:54] I see that in IE8 [15:25:14] i'm on chrome [15:25:38] hmmm [15:25:49] FF4 too [15:25:53] the same is true in safari [15:26:26] I'm on my mobile, so I can't patch it right this sec. [15:27:04] so if the only issue with this behavior is on the user experience, i think it's okay to go fowrard [15:27:05] but [15:27:28] I am running an errand, will be back at computer in like 1 hour [15:27:28] if this issue throws off the tracking, then we may need to fix [15:27:42] (wow you type fast on a mobile) [15:27:59] I think I may be onto something [15:28:35] its going to be in user.bucket most likely [15:28:39] mw.user.bucket [15:28:47] Looks like fixes weren't merged [15:28:56] ha ha [15:29:00] makes sense [15:29:23] I fixed the cookie name mismatch, but it never got merged [15:29:53] Found it, fixing [15:30:01] sweet [15:30:02] It's about treating the bucket value as a number [15:30:06] 'hide' is not a number [15:30:34] ah [15:31:25] Merge committed, updating test [15:31:37] RoanKattouw: did i miss something? [15:31:42] svn: Can't open file '.svn/tmp/entries': Permission denied [15:31:44] WTF [15:32:14] awjr: Please type umask in your fenari shell and tell me what it says [15:32:31] 0022 [15:32:41] *sigh* [15:32:47] Run umask 0002 [15:32:52] Then svn up [15:33:07] Also put umask 0002 in your bash profile or somthing [15:33:24] I have it in my ~/.bashrc [15:33:34] :( i assumed that would be something puppet did [15:33:38] It should b [15:33:40] e [15:33:43] *RoanKattouw files RT ticket [15:34:32] ok svn up'd but there are still files with bad permissions [15:36:11] I see [15:36:16] OK but you picket up the change? [15:36:18] *d [15:37:32] yes [15:38:10] OK [15:38:45] Bug looks fixed to me [15:39:06] great - btw did we need to do any schema updates? [15:39:26] No [15:39:29] k [15:39:57] so the fix has been pushed to test? [15:40:40] Yes [15:40:48] awesome [15:41:13] looks good to me [15:41:13] let's do a check on all the major browser/os combos [15:41:20] roan - can you do ie and ff on windows [15:41:36] I don't have FF in my Windows VM [15:41:44] i'll take ff on windows then [15:41:45] install it? it's not hard [15:41:48] :P [15:42:02] awjr: can you do ff, chrome on osx? [15:42:05] Fine [15:42:11] yeah, ff chrome and safari on os x look legit [15:42:31] awesome [15:42:47] let me check ff and chrome on windows [15:43:13] IE8 on Windows XP looks fine [15:43:37] Aha, Nimish also has a bad umask [15:43:57] Oh and someone's been naughty and run svn up as rot [15:44:00] *root [15:44:33] Hm wait this is not necessarily all umask-related [15:44:55] doh [15:45:01] i fixed the perms on the files i up'd this morning [15:45:28] awjr: When we're done could you run find . -user arichards -not -perm -020 | xargs chmod g+w ? [15:45:47] (trying to get the checkboxes on ff by clearing cookies. . .just not getting lucky) [15:46:40] FF4 on Windows works [15:46:50] worked for me in ff4 on os x [15:46:59] checking chrome on windows [15:47:04] RoanKattouw: done - find ftw [15:47:31] find and xargs, a match made in heaven [15:47:38] sweet sweet find [15:48:47] chrome on windows looks good [15:48:50] so i think we're good to go [15:48:54] OK [15:48:55] huzzah [15:49:02] Alright, awjr, run scap [15:49:50] RoanKattouw: does scap pick up the config changes also? [15:49:53] Yes [15:49:56] neat [15:49:57] And it clears message blobs [15:50:06] ok, brace yourselves [15:50:11] scap basically does everything but it was broken until earlier this week [15:50:14] wheee [15:50:17] When Tim fixed the sudo config [15:50:23] nice [15:51:34] it must be checking an awful lof ot syntax [15:51:44] does the syntax check normally take a while?' [15:51:50] Yeah like 20 seconds [15:51:57] It checks the syntax of the entire tree [15:51:57] ah there we go [15:52:00] oh wow [15:52:02] fancy [15:52:12] For what it does (over NFS too!) it's actually surprisingly fast [15:52:41] srv277: rsync error: some files could not be transferred (code 23) at main.c(1385) [generator=2.6.9] [15:52:42] im seeing those [15:52:57] do i care? [15:52:58] Are the errors just timestamps? [15:53:11] Timestamp errors = don't care [15:53:26] and: [15:53:27] The code 23 error is a generic "oh look something failed before" aggregation error [15:53:27] srv212: rsync: mkstemp "/var/tmp/texvc/.texvc_tex.ml.d0OGeW" failed: Permission denied (13) [15:53:35] Ah yes [15:53:46] It tries to recompile texvc and fails [15:53:49] That's another nasty thing [15:53:58] ok it is done [15:54:01] It's also the reason why I made you run ~~catrope/sync-common instead of the normal sync-common [15:54:12] Because the normal sync-common dies when it fails to mkstemp [15:54:16] oh i see [15:54:25] but scap perseveres [15:54:26] Makes me wonder why mkstemp in /var/tmp should even fail for non-roots [15:54:33] srsly [15:54:43] Aha, because /var/tmp/texvc exists [15:54:48] And is owned by root [15:54:51] doh [15:54:58] well anyway, scap is done [15:55:05] I can tell [15:55:13] Clear spike on Ganglia [15:55:33] howief, TrevorParscal: We should now be live on enwiki [15:56:03] awesome!!! [15:56:14] hmm i am not seeing emailcapture [15:56:16] let's check out justin beiber [15:56:36] Oh [15:56:38] ah EmailCapture is not enabled [15:56:41] Did you enable it on enwiki? [15:56:54] hmmm - i see email capture [15:56:54] Right better do that then [15:56:56] Are you on it? [15:57:03] yes [15:57:36] something is wrong with the survey [15:57:51] "an error has occured. Please try again later" [15:58:08] Let me reproduce that [15:58:11] It worked on test [15:58:58] indeed i am seeing that as well [15:59:04] just enabled emailcapture on enwiki [15:59:13] Bah getting 504 Gateway Timeouts [15:59:23] For the API calls [15:59:38] :( [16:00:06] This looks like an ops issue [16:00:15] PROBLEM - LVS HTTP on api.svc.pmtpa.wmnet is CRITICAL: CRITICAL - Socket timeout after 10 seconds [16:00:19] Plus lots of Apaches unreachable [16:00:37] did scap shake things up too much? [16:01:01] Doubt it [16:01:07] I think these were API apaches [16:11:29] RoanKattouw: so we roll back changes to AFT? [16:11:40] Yes, I'll do that [16:12:04] ok no sweat [16:12:24] are you just reverting to the previous revisions for AFT and the configs? [16:12:35] Yes [16:12:40] I'm reverting our AFT merge on 1.17wmf1 [16:12:47]