[00:25:37] 3MediaWiki / 3Uploading: SVG upload get blocked on correct encoding (windows-1252, with wrong/unspecific warning) - 10https://bugzilla.wikimedia.org/70937#c4 (10Bawolff (Brian Wolff)) (In reply to PRO from comment #3) > (In reply to Tisza Gergő from comment #1) > Yes, the [[Windows-1252]] encoding is preferr... [07:31:53] 3MediaWiki / 3Uploading: Generate thumbnails based on buckets - 10https://bugzilla.wikimedia.org/67525#c21 (10Gilles Dubuc) That's very odd, the chaining-free one I had was equally fuzzy as the chained one, which is why I didn't notice the difference. I'll try to figure out what happened there. This might be... [07:37:47] marktraceur, tgr: do you guys think it would be useful doing an engineering focused post-mortem/overview of MV at the developer summit? [07:43:36] (03CR) 10Gilles: [C: 032] Adjust metadata panel colors [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/161242 (owner: 10Pginer) [07:44:13] (03Merged) 10jenkins-bot: Adjust metadata panel colors [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/161242 (owner: 10Pginer) [07:45:28] (03CR) 10Gilles: [C: 032] Fix API help for query=videoinfo [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/157619 (https://bugzilla.wikimedia.org/55744) (owner: 10Brian Wolff) [07:45:51] (03Merged) 10jenkins-bot: Fix API help for query=videoinfo [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/157619 (https://bugzilla.wikimedia.org/55744) (owner: 10Brian Wolff) [07:59:49] (03CR) 10Gilles: [C: 032] Dedicated page for UW funnel analysis [analytics/multimedia/config] - 10https://gerrit.wikimedia.org/r/161290 (owner: 10Gergő Tisza) [08:16:14] gi11es: it should be definitely done sometime [08:24:15] (03CR) 10Gilles: [V: 032] Dedicated page for UW funnel analysis [analytics/multimedia/config] - 10https://gerrit.wikimedia.org/r/161290 (owner: 10Gergő Tisza) [08:32:12] (03CR) 10Gilles: Log errors (031 comment) [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/161363 (owner: 10Gergő Tisza) [08:57:45] (03CR) 10Gilles: [C: 04-1] "- Attribution tabs still look like the have broken CSS (firefox): https://www.dropbox.com/s/6klaole0343j1by/Screenshot%202014-09-19%2010.5" [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/160814 (owner: 10MarkTraceur) [08:59:02] (03CR) 10Gilles: [C: 04-1] "The last option of the embed dropdown is always out of view: https://www.dropbox.com/s/zjye5z3d9tmhtlj/Screenshot%202014-09-19%2010.58.08." [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/160041 (owner: 10MarkTraceur) [09:11:47] (03CR) 10Gergő Tisza: Log errors (031 comment) [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/161363 (owner: 10Gergő Tisza) [09:22:00] (03PS2) 10Gergő Tisza: Log errors [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/161363 [13:29:23] 3MediaWiki / 3Uploading: SVG upload get blocked on correct encoding (windows-1252, with wrong/unspecific warning) - 10https://bugzilla.wikimedia.org/70937#c5 (10PRO) (In reply to Bawolff (Brian Wolff) from comment #4) > Its not like one is better than the other really. If the file is > windows-1252, you shou... [16:05:41] Guest13488: will there be a community meeting today? [16:05:56] going from the calendar description, it seems very pointless [16:14:17] (03PS3) 10Gergő Tisza: Add tracking categories for files with attribution problems [extensions/CommonsMetadata] - 10https://gerrit.wikimedia.org/r/160580 [16:15:46] (03PS4) 10Gergő Tisza: Add tracking categories for files with attribution problems [extensions/CommonsMetadata] - 10https://gerrit.wikimedia.org/r/160580 [17:15:05] (03CR) 10Kaldari: "@Gilles: I agree it should be upstreamed. Thanks for noticing that. I don't, however, think we should delay merging this fix in the meanti" [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/161174 (https://bugzilla.wikimedia.org/53840) (owner: 10Kaldari) [17:51:17] Guest13488 / Keegan: ^^ [18:00:51] (03PS7) 10Kaldari: Protect against exceeding memory errors when parsing Ogg files [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/161174 (https://bugzilla.wikimedia.org/53840) [18:00:54] (03CR) 10jenkins-bot: [V: 04-1] Protect against exceeding memory errors when parsing Ogg files [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/161174 (https://bugzilla.wikimedia.org/53840) (owner: 10Kaldari) [18:04:07] (03PS8) 10Kaldari: Protect against exceeding memory errors when parsing Ogg files [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/161174 (https://bugzilla.wikimedia.org/53840) [18:10:30] (03CR) 10Kaldari: "Created an upstream pull request: https://github.com/pear/File_Ogg/pull/2" [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/161174 (https://bugzilla.wikimedia.org/53840) (owner: 10Kaldari) [18:14:44] (03CR) 10MaxSem: [C: 032] "looks plausible" [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/161174 (https://bugzilla.wikimedia.org/53840) (owner: 10Kaldari) [18:14:57] (03Merged) 10jenkins-bot: Protect against exceeding memory errors when parsing Ogg files [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/161174 (https://bugzilla.wikimedia.org/53840) (owner: 10Kaldari) [18:33:18] tgr: What was the fix for that weird markup in the size selectors? [18:33:42] Never mind, my bad [18:33:59] Placed rules in the wrong order. [18:35:11] (03PS1) 10MarkTraceur: Style fixes for the UI changes [extensions/MultimediaViewer] - 10https://gerrit.wikimedia.org/r/161496 [18:35:36] I got tired of rebasing things [19:02:40] gi11es: hey! Dan here. i have the mmv perf tests running an a brand new labs instance. there's still some work to be done on the traffic shaping front, but first i want to verify the numbers against what you've been logging [19:03:26] gi11es: what's the easiest way to get these numbers up in limn in a way that won't pollute what you're already logging? [20:17:36] marktraceur: know anything about the limn setup for mmv metrics? ^ i'm trying to get my data up there for comparison but i know very little about eventlogging and limn [20:18:18] i tried logging in but it seems the oauth setup is a little wonky (i get redirected to localhost after authorization against meta) [20:20:06] marxarelli: It's all configured on the backend... [20:20:12] marxarelli: Where does your data live? [20:22:52] marktraceur: it lives in my cwd at the moment. :) i didn't want to pollute the data that's already there [20:23:28] should i maybe use a custom user agent so it can be filtered? [20:25:35] sorry, to clarify, the test as it's currently written logs to the MultimediaViewerVersusPageFilePerformance eventlogging schema. i've temporarily disabled that in my local test until i can be sure i'm not going to muck up your team's data [20:26:39] Ah [20:26:53] Well, the data lives on stat1003 and gets used by limn1 [20:32:21] marktraceur: oh, ok. so everything that's pushed through eventlogging ends up on stat1003? [20:47:30] marxarelli: depends on how eventlogging is set up on the wiki where you do the logging [20:48:01] specifically, $wgEventLoggingBaseUri [20:48:08] by default, it logs locally [20:49:37] tgr: got it. do you know which wiki is being targeted for the selenium perf tests by chance? i've only been able to successfully run it against beta (test2 and prod don't have the pages used by the test) [20:50:04] so, set up your local eventlogging to actually work (which it doesn't by default, and does not seem trivial to do, ori can probably help), run the SQL to create the tsv locally, point limn to that tsv [20:50:55] well, the first line of the test is [20:51:02] @en.wikipedia.beta.wmflabs.org @mediawiki.org @firefox @internet_explorer_8 @internet_explorer_9 @internet_explorer_10 @internet_explorer_11 @test2.wikipedia.org [20:51:22] so I assume it runs on beta-en, mediawiki and test2? [20:51:34] huh [20:51:57] if that's true, it doesn't make much sense that we show it in a single graph [20:52:08] hmm.. ok. it fails when i point it to test2 but it works fine against beta [20:52:18] anyway, if you have stat1003 access, you can double-check in the DB [20:53:01] tgr: cool. that sounds like a good next step. who can i bug about getting access? [20:53:31] there is an email address for creating new RT tickets [20:54:54] ops-requests@wikimedia.org [20:55:19] fantastic. i really appreciate the help! [20:55:53] https://wikitech.wikimedia.org/wiki/Analytics/Data_access#Access_requests has the details [20:56:55] marxarelli: takes a few days so let me check the DB data for now [20:58:51] 3MediaWiki / 3Uploading: Uppercase picture extensions like .PNG are not allowed - 10https://bugzilla.wikimedia.org/66667#c8 (10Cindy Cicalese) We also experienced this bug after upgrading a wiki from 1.13.2 to 1.23.1. Following the suggestion above ("Removing the ! in includes/libs/IEUrlExtension.php at line... [20:58:56] mysql:research@s4-analytics-slave.eqiad.wmnet [log]> select distinct wiki from MultimediaViewerVersusPageFilePerformance_7907636; [20:58:59] +---------------+ [20:59:02] | wiki | [20:59:04] +---------------+ [20:59:07] | commonswiki | [20:59:09] okay, that's... surprising [20:59:11] | mediawikiwiki | [20:59:14] +---------------+ [20:59:16] 2 rows in set (0.02 sec) [21:00:31] mysql:research@s4-analytics-slave.eqiad.wmnet [log]> select wiki, count(*) from MultimediaViewerVersusPageFilePerformance_7907636 group by wiki; [21:00:34] +---------------+----------+ [21:00:36] | wiki | count(*) | [21:00:39] +---------------+----------+ [21:00:41] | commonswiki | 429 | [21:00:44] | mediawikiwiki | 1645 | [21:00:44] mysql:research@s4-analytics-slave.eqiad.wmnet [log]> select wiki, count(*) from MultimediaViewerVersusPageFilePerformance_7907636 where timestamp > '20140901000000' group by wiki; [21:00:48] +---------------+----------+ [21:00:51] 2 rows in set (0.00 sec) [21:00:53] +---------------+----------+ [21:00:56] | wiki | count(*) | [21:00:58] +---------------+----------+ [21:01:01] | commonswiki | 114 | [21:01:03] | mediawikiwiki | 219 | [21:01:06] +---------------+----------+ [21:01:08] 2 rows in set (0.03 sec) [21:01:45] so it runs twice on mw.org and once on Commons [21:02:23] and six times a day, apparently [21:04:03] hmm... that seems off. the way the test is written seems like it would push 4 mw metrics for every 1 commons metric [21:04:44] the url for the target wiki is taken from an env variable, but the commons one is hardcoded [21:07:57] correction, there are 5 scenarios that target a configured wiki (mw.org from the looks of it) and 1 that targets commons [21:09:04] ok, well this at least tells me where the test it pointing so i can compare my own results. i appreciate you digging in the db [21:09:17] marxarelli: btw if you are going to investigate that test up close, something that would be great if you could look at is caching [21:09:55] there are versions of the test with cold and warm cache and the difference is pretty much zero, which is hard to believe [21:10:13] tgr: really? i'm seeing big differences [21:10:17] in my own data that is [21:10:38] when I mentioned this to Gilles, he pointed me to the relevant selenium docs and those do say quite clearly that the test is set up correctly [21:10:45] so I don't know what's up [21:11:08] tgr: weird. yeah, i'll have a look and compare [21:11:10] http://multimedia-metrics.wmflabs.org/graphs/mmv_versus_file_page [21:11:41] there used to be a difference back in may, then it disappeared [21:11:51] tgr: do you know if these jobs are running on integration.wikimedia.org? [21:12:04] I don't [21:12:20] SauceLabs was thrown around a lot in the discussions [21:12:34] hashar can surely tell [21:13:09] hmm... sauce would be a bit problematic given the latency between the runner (cucumber) and the webdriver [21:13:21] awesome. i love bugging hashar! :) [22:01:23] 3MediaWiki extensions / 3TimedMediaHandler: Allowed memory size of 183500800 bytes exhausted (tried to allocate 2018670161 bytes). - 10https://bugzilla.wikimedia.org/53840#c12 (10Ryan Kaldari) 5PATC>3RESO/FIX Hopefully, this is fixed now.