[01:10:06] Elsie: "(Though contractors can still leak, heh.)" thanks for that, made me giggle on my bus ride home [01:10:20] :D [01:17:30] Elsie, I like how you said it's not going great because it's not going your way [01:18:29] I don't think thats why he said it. [01:18:38] It's not going well when a bunch of people threaten to resign. [01:18:55] The idea of community elected staff sounds good, though I'm not sure how that would work money-wise [01:19:08] Krenair: Which way is my way? :-) [01:19:18] There are eight people (for now) saying they're going to quit over it. [01:19:24] Including a few stewards. [01:19:33] That worries me. [01:23:08] Elsie, I'm guessing you would prefer that most people supported the majority of this draft. [01:25:27] Krenair: I think it's a whole lot of change at once. :-) [01:25:37] And mixing issues just complicates things. [01:25:46] Similar to a big changeset vs. small changesets. [01:26:11] The "confidentiality pledge" is a big step. Retaining IDs + retaining IDs for three years is another big step. [01:26:23] Adding "volunteer developers" (whatever that means) is a big step. [01:26:38] If you try to take too many big steps too quickly, you'll trip up. [01:26:43] volunteer developers = developers who are volunteers, not paid wmf/wmde staff [01:26:44] That's what I think we're seeing here. [01:26:52] Who's a developer? [01:26:55] Anyone with a Labs account? [01:27:06] Anyone with a Tool Labs account? [01:27:24] No. Didn't Tim define some arbitrary line count or something for this? [01:27:44] I'm not sure Board policies should rely on some arbitrary line count made up by Tim one day. ;-) [01:27:55] <^d> We should go back to when you had to beg commit access. Then it was easy to know who developers were. [01:27:58] I think Tim would agree, though you'd have to ask him. [01:28:27] ^d: Back to CVS. [01:28:40] For the purposes of this policy, the distinction is largely useless [01:29:02] Well, as I just asked on the talk page, I'm not sure how you determine whether the policy is applicable. [01:29:06] Who makes that determination? [01:29:57] It's only supposed to deal with volunteer developers who happen to have access to nonpublic data. It certainly doesn't seem to cover all volunteer devs [01:30:01] Krenair: Re: paid staff for stewards and checkusers, I think you'd cut costs by sacrificing some privacy, right? [01:30:22] You don't need any many staff if you can directly query the database and do pattern analysis on all of the IPs. [01:30:28] "Go fishing" as they say on the English Wikipedia. [01:30:37] And staff or paid contractors would presumably be able to do that. [01:30:42] Cutting down sockpuppeting dramatically. [01:31:14] Krenair: Toolserver users have access to nonpublic data (archive table, etc.). [01:31:52] Hm, true. I hadn't considered that. Usually all the TS/labs stuff is considered public - anyone can get an account there - It might be useful to spell out clearly what is nonpublic [01:32:06] Toolserver has partial views for watchlist, user_properties, etc. [01:32:10] And maybe one day Labs will. [01:32:20] Though I suspect some people might be holding off on adding views until this policy is settled... [01:32:28] Adding views to Labs, that is. [01:32:45] * Elsie shrugs. [01:33:13] I don't think the Board will pass this, so I'm not particularly concerned at the moment. [01:34:45] Elsie, there was some concern about this policy being illegal for German functionaries to comply with [01:34:56] then someone else said it was the exact opposite [01:35:10] that should be looked in to [01:40:41] I agree. [01:40:42] :-) [06:46:20] can mediawiki be used to track access to pages [06:46:29] such as which IP or user read which article? [06:46:40] this'd probably be an extension [17:11:27] yikes; all the links on Commons are suddently redirecting to wikimediafoundation.org [17:11:53] someone broke something :( [17:11:53] commons is down o_O [17:12:19] someone just opened a bug about it happening at meta too [17:13:18] We know [17:14:58] Is it really just me? http://downforeveryoneorjustme.com/http://commons.wikimedia.org [17:15:20] marco__: they are aware of the issue [17:22:10] Hey, what is wrong with commons?? [17:22:11] we need moar nonbroken .wikimedia.org wikis [17:22:58] also, on some wikis insecure javascript is injected from wikimediafoundation.org (random banner) [17:41:24] not sure this is the right place to ask [17:42:19] i want to make a dictionary for ebook reader based on wikipedia [17:42:33] i will need just the first paragraph [17:43:09] is there an automated way to do this from the dump? [17:44:20] Yeah, write some code to process the dumps [17:45:15] i know that, the thing is i'm not good at the writing code part :) [17:45:40] maybe someone did it before or has a tutorial [17:46:06] There's numerous scripts for processing XML dumps [17:46:32] Working out a paragraph, removing infoboxes, images links... [18:05:23] I'm getting some weird behavior on outreach.wikimedia.org [18:05:31] still? [18:05:38] 404 index.php not found on this server [18:06:03] Example pages? [18:06:10] and occasionally links redirecting to the corresponding page on wmf.org [18:06:15] mutante just had to kick a few apaches that were misbehaving [18:06:38] I've tried a few times to save an edit here, and this gives the 404: https://outreach.wikimedia.org/w/index.php?title=Welcome_to_Wikipedia_%28Bookshelf%29/2013_edition/draft&action=submit [18:07:11] worked that time. [18:07:12] WFM and just saved a null edit [18:07:40] Possible your browser has some stuff cached too [18:08:35] okay. I'll clear caches and keep an eye out for any problems that stick around, and assume the null hypothesis that the problems are fixed. [18:09:02] Certainly purge any pages you see with screwy links [18:11:07] okay. the links don't appear screwy until clicked. [18:12:42] Reedy, clearing cache isn't helping. [18:13:00] I go to this page: https://outreach.wikimedia.org/wiki/Welcome_to_Wikipedia_%28Bookshelf%29/2013_edition/draft [18:13:04] and I edit a section [18:13:28] and then, for example, if I click the logo to try to reach the main page, I get redirected to wikimediafoundation.org [18:14:42] works in Chrome [18:14:58] but in Firefox, a hard refresh doesn't fix it. [18:15:37] and neither does a null edit. [18:16:45] hello hello [18:16:59] Who's hello? [18:17:01] ragesoss, I tried restarting my browser, and that seems to have fixed it for me just now. [18:17:10] I think the browser was caching the DNS lookup [18:17:25] *seems* is of course in italics, bold, and underline [18:18:05] doh [18:18:08] All wikimedia.org sites would resolve to the same thing anyway ;) [18:18:13] it's still not working sometimes guys :-/ [18:18:22] it seems like some of your nodes are still broken [18:18:38] usually goes to commons.wikimedia as it should, but occasionally redirects to wikimediafoundation.org [18:18:40] Magog_the_Ogre: restarting browser didn't work for me. [18:19:01] FYI I have Chrome set to enforce HTTPS on wikipedia.org and wikimedia.org subdomains [18:19:08] but not wikimediafoundation.org [18:19:42] Strange that wikimediafoundation is being used as a redirect target [18:19:52] I've got too many dynamic windows open to restart my browser [18:20:09] You need a better computer than can handle it then :P [18:20:49] Dragonfly6-7, on both Chrome and Firefox, if you terminate the program, you should be able to recover your tabs and windows [18:21:13] no, you miss the point. [18:21:16] These are *dynamic* windows. [18:21:21] if you're using IE, as a programmer, I do not have enough evil things to say to you :-/ [18:21:23] If I close them and reopen them, the data will be lost. [18:21:44] rephrase rephrase [18:21:45] you mean you have data entered into the textboxes? [18:21:45] dynamic pages [18:21:48] no [18:21:55] dynamic is pretty vague [18:21:58] office wiki is intermittently having the same problems, and I just used it for the first time today. [18:22:16] I mean I have lists of "pages that are 720 hours old", but the lists themselves are a few days old, and if I reload one of those lists, everything on it will be lost [18:22:29] Print as PDF [18:22:32] oh now that's a problem [18:22:37] heh [18:22:43] File -> Save as [18:22:47] lol [18:22:50] Computers are hard. [18:22:52] Dragonfly6-7: if it's a dns caching issue, just wait long enough [18:23:02] What DNS caching issue? [18:23:05] There's no DNS issues [18:23:13] yeah Dragonfly6-7, a restartdidn't fix it for us, so don't repeat [18:23:18] I think the browser was caching the DNS lookup [18:23:25] ok. [18:23:28] It probably was [18:23:34] no, because it's occasionally still redirecting MatmaRex [18:23:36] But all wikimedia sites point to the same load balancer [18:23:47] ah, so it'll eventually balance itself out. [18:23:47] nevermind me, i don't know what the talk is about [18:23:48] I think it's just that a node or two is still goofing up, while the rest are OK [18:23:49] well, i do [18:23:57] but nevermind me still. [18:28:43] i guess everyone knows about commons.wikimedia.org returning a php file instead of html? and there are bigger things to worry about? [18:28:58] Hum, why do I get redirected to http://wikimediafoundation.org/wiki/Special:Watchlist when I try to access the watchlist on commons? [18:29:06] lol [18:29:25] This is going to be good fun [18:29:36] heh [18:29:45] oh boy [18:30:35] TommyG - it's affecting people sporadically [18:30:38] lmao [18:30:48] TommyG: pleas clear your browser cache [18:30:58] ah, see Dragon [18:31:28] I'm sporadic then, it affects me everytime [18:31:45] Which urls? [18:31:49] Clicking what on where? [18:32:06] strange. after clearing the cache. for me works perfect o_O [18:32:10] Reedy: it's happend to me: clicking links in my watchlist [18:32:32] Links are fine, actual target is not? [18:32:33] clicking the main page icon both in edit mode and read mode, on various pages on outreach wiki. [18:32:48] links are correct on hover [18:33:02] but target ends up being the corresponding page on wikimediafoundation.org [18:33:18] and it's indeed very intermittent for me. [18:33:30] i don't get why wikimediafoundation [18:34:06] Reedy: I just got the php problem edsu was talking about. [18:34:17] tried just going to office.wikimedia.org [18:34:26] and was given this as a download... [18:34:45] http://pastebin.com/uFNdRUDc [18:35:15] That sounds very much like something is either cached or an apache still isn't working right [18:35:23] Reedy: http://i.imgur.com/kwwH8y5.jpg [18:35:54] Reedy: do you need more help on this issue? [18:36:20] TommyG: better now? [18:36:34] greg-g: Actually replicating it would be nice [18:36:54] greg-g: yea, but.. we reverted.. [18:37:35] purges office.wm from squid [18:37:37] wfm [18:37:47] Reedy: I can login and go to random pages successfully on officewiki [18:38:03] Ditto [18:38:16] There's very little on officeiwki that would be cached [18:38:21] (in squid etc) [18:38:37] ragesoss: still seeing it? [18:38:42] yes. [18:38:54] lol [18:38:58] I press F5 and get the behaviour [18:39:09] Uhh [18:39:11] mw1052 [18:39:13] congrats! you've replicated! [18:39:35] got it here with mw1042 [18:39:36] I note mw1052 isn't in the output from apache-fast-test.... [18:39:41] http://office.wikimedia.org/ * 301 Moved Permanently https://office.wikimedia.org/ [18:39:45] mw1052 [18:40:03] * 301 Moved Permanently https://office.wikimedia.org/wiki/Main_Page [18:40:15] I promise I saw 1042 :) [18:40:23] and again [18:40:27] with 42 [18:40:42] mw1042 ! yes, wait [18:40:55] :) [18:41:02] * greg-g waits [18:41:04] try again, restarted [18:41:12] https://commons.wikimedia.org/wiki/Special:UploadWizard -> http://wikimediafoundation.org/wiki/Special:UploadWizard :) [18:41:35] mutante: I think you need a bigger hammer [18:41:42] Crap, it was working an hour ago [18:42:53] can't get it with f5 here anymore [18:43:16] on officewiki [18:44:44] restarted some more..sigh [18:44:45] nor on commons [18:44:54] so, I can no longer replicate [18:44:56] greg-g: try this link: https://office.wikimedia.org/wiki/IRC#Alternate_instructions_if_you.27re_willing_to_depend_on_marktraceur [18:45:43] Heh. [18:45:47] L) [18:45:53] ragesoss: wfm [18:46:01] rather :) [18:46:10] weird weird weird. [18:46:21] i just reloaded like 20 tuiumes [18:46:27] -us [18:48:34] mutante: much :-) [18:54:32] mwalker: heya, when ya'll wanna do your fix today? [18:55:05] I was thinking the LD [18:55:21] k [19:03:15] Might have the cause of the problem [19:03:28] And why fixing all the apparently bad apaches isn't actually fixing all the apparently bad apaches [19:06:07] ragesoss: How about now? [19:06:23] TommyG: dr0ptp4kt Magog_the_Ogre ^^ [19:06:57] TommyG, I just recycled my connection. would you please resend? [19:07:40] reedy, the errors I've been experiencing have been intermittent, but I haven't seen any for about 15 minutes [19:08:08] That's just fixed a problem Krinkle reported about loads of stupid redirects on mediawiki.org due to fundraising [19:08:30] Magog_the_Ogre: It was only improved in the last 4 or 5 minutes [19:08:56] idk. I'm just reporting what I observed [19:09:03] Yeah, watchlist/userpage/etc works as expected now [19:09:06] :) [19:09:24] Hi, there is a problem in http://br.wikimedia.org/, the Special:RecentChanges (second link on side menu) is redirecting to wikimediafoundation.org, it was not happening yesterday [19:09:34] lmfao [19:09:40] danilo: Still? When did you last try it? [19:09:51] Problem was hopefully fixed in the last 5 minutes or so... [19:09:52] now [19:10:05] ok, thanks [19:10:06] mutante: http://br.wikimedia.org/wiki/Special:RecentChanges [19:11:00] WFM though [19:11:15] works with test script (now) [19:12:14] danilo: it should just be browser cache now [19:14:22] I reload the cache (shift+F5), but the problem persists here [19:22:15] danilo: how about really hitting "clear cache" button / restarting browser [19:22:27] all servers look ok to me [19:28:46] Server error 404 after uploading chunk: Not Found Response: 404 Not Found

Not Found

The requested URL /w/api.php was not found on this server.

Additionally, a 404 Not Found error was encountered while trying to use an ErrorDocument to handle the request.

[19:31:42] mutante: i am still seeing http://commons.wikimedia.org/ returning a file [19:32:11] aude: Try clearing your browser's local cache. WFM [19:32:19] i am trying [19:32:28] i don't cache anything [19:32:35] How rude [19:32:55] And probably not entirely true. [19:33:16] It's pretty tough to get most browsers to not cache in memory [19:33:16] there was one last server left that may have been it, mw1044, i just fixed it a little while ago [19:33:23] well, i disable much cache as possible [19:34:11] tested upload and commons one more time, they all look ok [19:34:15] * aude doesn't want to clear my entire cache, like all my cookies [19:34:19] on eqiad apaches [19:34:32] also problem on firefox [19:34:38] which i hardly ever use [19:35:00] Cleared all history. Restarted firefox. Still getting the 404 for api.php sometimes [19:35:02] firefox (no proxy), chrome (proxy) [19:35:09] can't be dns [19:35:17] mutante: I did it now, but it is still redirecting to wikimediafoundation.org [19:35:55] aude: not related to dns, it was wrong apache config [19:35:55] * aude wonders what was changed [19:35:58] ok [19:36:02] and reverted [19:36:14] https://gerrit.wikimedia.org/r/#/c/90669/ ? [19:37:23] aude: https://gerrit.wikimedia.org/r/#/c/91195/ [19:38:55] mutante: ok [19:41:12] alright, i am sure the issue will correct itself or seems so [19:41:56] The obviously broken sources should now at least be fixed [19:42:12] ok [19:45:03] Part of my plan to remove 7 docroots [19:45:05] aude: Have you looked at the downloaded file you are sent? Wondering if it contains any clues as to the origin. Maybe even better if you can trace in firebug or similar. [19:45:30] aude: Is it an unparsed extract2.php by any chance? [19:45:30] I would suggest getting the full http headers sent/received [19:45:48] awww <3 flow works [19:45:59] let me look [19:46:20] SarahStierch is reporting the same issue still on bug 56006 [19:46:29] it is extract [19:46:32] .php [19:46:35] extract2 [19:46:59] https://gist.github.com/filbertkm/75c3a56dfa116b10bcb1 [19:47:30] Can you see the http headers? /me thinks they show the varnich/squid that sent [19:47:31] if i request http://commons.wikimedia.org? that bypasses cache, obviously [19:47:33] I'd see what the headers next like apergos said [19:47:35] let's see [19:48:12] Then maybe go squid hunting [19:48:22] Have we got any dolphins? [19:48:25] * bd808 calls Reedy Ahab [19:48:27] https://gist.github.com/filbertkm/04d793032b8a773b52de [19:48:30] headers ^ [19:48:33] varnish [19:48:35] I wonder if a Unicorn would work [19:48:50] X-Cache cp1054 hit (2), cp1054 frontend hit (34) [19:49:03] we have another one from cp1055 [19:49:04] hrm [19:49:49] mark banned php [19:49:56] but doesn't help when it's not served as php [19:49:57] Content-Type text/html; charset=UTF-8 [19:50:05] :/ [19:50:10] Can we ban it on content length? [19:50:11] Content-Length 33518 [19:50:12] huh, now it works [19:50:27] if it's 33518 it's gotta likely be extract2 [19:52:28] aude: Thanks for that data. [19:54:09] I purged "http://commons.wikimedia.org" and "http://commons.wikimedia.org/" [19:55:58] I have just installed Chrome to test, but even in chome is redirecting to wikimediafoundation.org [19:56:45] From where? [19:57:08] http://br.wikimedia.org/wiki/Especial:Mudan%C3%A7as_recentes [19:58:23] fixed! \o/ [19:58:34] reedy@tin:/a/common/php-1.22wmf22$ echo "https://br.wikimedia.org/wiki/Especial:Mudan%C3%A7as_recentes" | mwscript purgeList.php --wiki=commonswiki [19:58:34] Purging 1 urls [19:58:34] Done! [19:58:34] reedy@tin:/a/common/php-1.22wmf22$ echo "http://br.wikimedia.org/wiki/Especial:Mudan%C3%A7as_recentes" | mwscript purgeList.php --wiki=commonswiki [19:58:34] Purging 1 urls [19:58:36] Done! [19:59:04] Interesting it wasn't redirecting me... [19:59:12] Though, for brwikimedia, I'm guessing you are in Brazil? [19:59:21] Hitting different frontend caches to what I would (esams) [19:59:22] yes [20:00:05] I guess if anyone else needs it doing and you have shell ^ something like that should remove [20:00:08] Reedy: I think bans on content-length should be possible. /me guesses `ban obj.http.content-length == 33518` [20:00:11] bd808: At least the purging works now ;) [20:00:53] As long as the relay stays alive [20:00:59] * bd808 crosses fingers [20:01:49] We should setup an irc bot that can send the purges :) [20:02:21] !purge http://blah [20:03:48] Reedy: everything seems fine for me [20:04:27] Yay [20:05:13] bd808: Might be worth getting someone to do that ban as an extra step [20:05:17] Reedy: if you want to deploy https://gerrit.wikimedia.org/r/#/c/91283/ and https://gerrit.wikimedia.org/r/#/c/91284/ [20:05:20] would be awesome :) [20:05:42] just small but annoying bugs that shouldn't wait until next week [20:09:34] FYI, reedy et. alia, I am STILL having issues with the JavaScript not loading properly on every page load [20:09:45] every page load on what wikis? [20:09:59] Commons [20:10:15] I rely heavily on the JSs [20:10:32] on a page load, occasionally I have to hit the reload button atm because the JS doesn't load for the tool I want. [20:11:22] Certain JS things? Or everything? [20:11:29] Can you pastebin from your js console? [20:11:44] eh [20:11:50] no, it changes on each page load [20:11:57] sometimes when HotCat won't load [20:12:04] sometimes the no permissions box isn't present [20:12:27] sometimes if I hit "watchlist", it redirects to the "your page is watchlisted" page instead of performing an AJAX request [20:13:25] cp1054 [20:13:36] * Reedy looks what exactly cp1054 is [20:14:10] text varnish [20:14:48] quick question for Lua people: does anybody know how exactly /doc pages are transcluded on Module pages? There doesn't seem to be transclusion code on those pages, unless I'm missing something [20:15:36] I think the extension does magic [20:16:10] Reedy was that answering mine? [20:16:14] Indeed [20:16:24] public static function getDocPage( $title ) { [20:16:24] $docPage = wfMessage( 'scribunto-doc-page-name', $title->getText() )->inContentLanguage(); [20:17:42] equazcion: I guess it should be of a question of why are you asking? Is something broken? [20:17:43] reedy: http://tinypic.com/r/eh0kdw/5 [20:18:02] notice in the dropdown that none of the normal extra script goodies are loaded [20:18:05] anomie: ^^ [20:18:24] like "Google Images", "Tineye", "Global usage", etc. [20:18:40] Have you had a look at the js console errors? [20:18:46] equazcion: Extension does magic, yes. [20:18:51] :D [20:19:01] mutante: It should at least tell us if it's being redirect somewhere wrongly et [20:19:11] i'm asking because we need to update pp-template -- i'm trying to find out where modules' protection templates come from, like pp-indef etc at the moment. templates get them from the {{documentation}} template, but with the magic I'm not sure what is specifying them on modules [20:20:08] equazcion: MediaWiki:scribunto-doc-page-show ? [20:21:16] ah ok. so i guess if i wanted to test changes to pp-template on module pages, that would present a problem [20:22:20] there's no documentation template i can duplicate to call the sandbox version of pp-template [20:28:19] anyway thanks for the info reedy and anomie :) [20:34:22] Reedy: I'm still getting redirects to WMF.org from the outreach logo link. [20:34:42] https://outreach.wikimedia.org/wiki/Special:MyLanguage/Main_Page ? [20:34:48] yes [20:35:24] Just purged it.. [20:37:54] working now [20:38:29] (after clearing browser cache) [21:47:54] ori-l: Could I get a sanity review of https://gerrit.wikimedia.org/r/#/c/91199 ? [21:48:10] csteipp: sure; looking. [21:48:11] csteipp: your links in mail last night were broke [21:49:39] Wow... google stripped all my newlines. Awesome [21:53:44] greg-g: can I sync a mw.inspect change during LD (or earlier)? [21:54:22] ori-l: earlier, probably, csteipp is doing a deploy starting in 5 minutes, coord with him [21:54:43] LD already has two users [21:54:54] kk. csteipp, can I go after you? [21:55:01] alternately I can deploy your config change if you like [21:56:20] ori-l: If you want to deploy, I just have that and a beta config change [21:56:40] sure, happy to. what's the beta change? [21:56:48] https://gerrit.wikimedia.org/r/#/c/90670 [22:25:10] Question: Does a gadget need to have applied [ResourceLoader] for [top] to take effect? [22:25:32] YairRand__: Yes, [top] is a ResourceLoader feature. [22:25:49] Krinkle: thanks [22:26:10] Note that not using ResourceLoader is discouraged, and within a few months it will likely become the default and only way to load gadgets [22:26:25] ah. good to know. [22:27:04] RL on itself shouldn't be any trouble though, there's other things around it we recommend you migrate at the same time, but it's OK to leave it as is, and just enable it. There may be some scope issues that need to be addressed after adding [ResourceLoader] but most of it should just work as-is. [22:52:07] ori-l: does httparchive do geographical differences? [22:52:24] jeremyb: no, I don't think so. [22:56:47] ori-l: also, there were some recent comments about how much JS main page loads on some wikis vs. commons. seems like exactly the kind of intra-cluster differences that that site will monitor [22:57:10] rollback links aren't showing up for me post-Thanks deploy [22:57:20] PinkAmpersand: You're welcome [22:57:44] well thanks for the Thanks (love that feature), but I need my rollback too :P [22:57:49] jeremyb: that is tracked in Ganglia now and is mostly the result of gadgets [22:57:49] Where exactly? [22:57:50] (cur | prev) 04:10, 27 September 2013‎ Tariqabjotu (Talk | contribs | block)‎ m . . (6,391 bytes) (-1)‎ . . (removing unnecessary pipe) (rollback 2 edits | undo | thank) [22:57:57] in the case of commons, the use of the media player on the main page [22:58:00] https://www.wikidata.org/w/index.php?title=Q14923848&action=history [22:58:12] (not actually a bad edit; we just use rollback as a quick way to track down where merges went) [22:58:37] hmm [22:58:39] i dont see it either [22:58:43] ori-l: still seems relevant to httparchive [22:58:53] jeremyb: yes, sure [22:59:05] unless i just didn't notice, I'm pretty sure I didn't see the "thank" button till just now [22:59:20] and the first two or three times I tried to load that page, it took way longer than it should have [22:59:39] PinkAmpersand: thanks was deployed today [22:59:41]