[00:14:20] hey, under what product/component should i file a bug to update the description that comes up when you Google "Wikipedia"? [00:14:54] *under what component [00:15:02] (since I imagine the product is Wikimedia) [00:15:24] Ummm. [00:15:37] What comes up? [00:15:50] "The biggest multilingual free-content encyclopedia on the Internet. Over 10 million articles in over 200 languages, and still growing." [00:16:04] out of date by at least 15 million articles and at least 75 languages [00:17:17] I would think that is something that Google would have to fix. [00:17:45] really? i'd think it's something on our side... most Google results come from the client, n'est-ce pas? [00:24:14] Pink|dinner: http://www.dmoz.org/search?q=Wikipedia&start=0&cat=/Top/Computers/Software/Groupware/Wiki/Topics&all=no [00:24:19] but it appears broken somehow. [00:27:41] Pink|dinner: http://www.dmoz.org/public/update?cat=/Top/Computers/Software/Groupware/Wiki/Topics [00:28:16] I'm not exactly sure how I'm supposed to update the description. [00:28:25] * twkozlowski zzzZZZzzz... [01:52:01] twkozlowski: :/ [04:15:47] People using Chrome on en Wikipedia are complaining about full width TOC issue. [04:16:01] Anyone know whats up? [04:16:22] VPT? [04:16:49] I'm on droid and about to pass out... [04:16:53] * theopolisme is doing [04:17:29] Check for changes to common.js/css theopolisme [04:18:17] T13|sleeps: There are several sections explaining what the issue is. [04:18:19] And how to resolve it. [04:18:22] I'm not sure what more you'd like. [04:19:17] Elsie, just a hug. :) [04:19:28] !hug | T13|sleeps [04:19:34] I made a joke there. [04:19:37] Aww [04:19:47] * theopolisme groans [04:19:56] lol [04:20:08] * T13|sleeps passes out... [04:50:55] T13|sleeps: had that issue the other day [06:45:53] I just got a [06:45:54]

502 Bad Gateway

[06:45:54]

nginx/1.1.19 [06:45:55] on meta. [06:46:18] reloading worked. [07:04:15] legoktm: Do you know which URL? [07:04:36] https://meta.wikimedia.org/wiki/User:MF-Warburg/abuse [07:04:53] legoktm: Can you please make a note on https://bugzilla.wikimedia.org/show_bug.cgi?id=50891 ? [07:05:03] At the moment, I'm the only one to have reported the issue. [07:05:08] Ok, sure [07:05:28] And if it's outside en.wikipedia.org, we can update the bug summary. [07:06:29] ty [11:15:01] http://bn.wikipedia.beta.wmflabs.org/ is stopped?? [11:27:22] jayanta: I am not sure it ever worked [11:33:47] I need to check some twinkle js for testing...! [11:34:21] Now where I test?? [13:26:52] Does anyone know if the SUL finalisation will automatically create a user account on all sites when an account is first created? Or will that continue to function in the same way as it does now (wait until the user visits the site and then create the account) ? [13:28:37] addshore: The plan is to continue like we do now [13:28:46] Asking because of WD cross site editing? [13:28:47] okay :> [13:28:55] yup [13:29:16] specifcally https://bugzilla.wikimedia.org/show_bug.cgi?id=43192 which I just stumbled upon [13:29:24] ah, ok [13:29:25] naturally having every account exist everywhere would help :P [13:29:45] Sure, but that would blow our databases :P [13:29:52] :D [14:20:41] addshore: I don't think user IDs are the main problem for that bug [14:34:26] (commented on the bug) [14:52:21] Steinsplitter: hoo: ping [14:54:59] I'm here ;) [14:55:38] :) [14:56:14] Don't manually replace files on the server, we have version control for this. [14:56:25] So it is easy to see how did what, when, and how to undo it. [14:56:28] And review of course. [14:58:12] Krinkle: Ok... you are aware that there already are change in public_html which aren't in git? [14:58:36] https://svn.wikimedia.org/viewvc/pywikipedia/trunk/pywikipedia/commonsdelinker/ [14:58:45] ther ar no such fiel in the SVN [14:58:48] Yes, but that is no excuse. And the bigger the change (this new UI Steinsplitter is working on is a major change) the bigger the problem. [14:59:17] And a new entry point (kill switch) is especially worrysome and more than just front-end. that needs to be secure and tight, and what I saw is not that (yet). [14:59:23] Krinkle: Didn't say that as an excuse [14:59:36] whatever.... the php code is the same. [14:59:38] Just saying it in advance to prevent problems. [14:59:47] I didn't look at kill switch [14:59:47] No, it is most certainly not the same. [14:59:52] k [14:59:53] and I think Steinsplitter didn't touch it [15:00:07] I'm going to be updating the repo tomorrow so we can do this in a pull request [15:00:29] I guess that's fine [15:00:50] Krinkle: Thank you. The php error: ther is actually no such process on my vhost and no lofiel. For this greaso the error. On toolserver the cod schold work finde :) [15:01:29] Like I said before in the PM, it not being there is not the problem. It failing the way it does is the problem. [15:02:17] It exposes path information that may want to be kept private, and more importantly, it causes unpredictable output by PHP at a point that can be problematic. [15:02:30] Output your own warning, don't rely on PHP to cover it. [15:02:40] so check is_readable before opening a file etc. [15:03:11] The file will disappear at some point on Toolserver. things will fail at some point. This is reality :) [15:03:13] the php code os exactly the same. [15:04:01] We'll see [15:04:12] and if my help is not welcom, feel free to say "stop" ;-) [15:05:15] your help is welcome, I'm just making sure we don't move too quickly. Getting and keeping things stable is not easy, and ensuring integrity and doing review is part of that. [15:05:18] I know you ans siebrand hav a lot of work to do for the wmf. And for this reaso i hav decided to fix now the "webinterface". [15:05:21] Nothing went wrong, we're just not there yet. [15:06:34] Okay. Thanks for your time, Krinkle :-) [15:15:45] Is single log in not working anymore? [15:16:45] I'm having to log in to meta, en, commons, mediawiki, etc all separately... [15:17:21] really? we launched SUL stuff on Wednesday to make stuff easier [15:17:29] are you doing odd things with cookies or redirects? [15:17:43] T13|college: http://lists.wikimedia.org/pipermail/wikitech-ambassadors/2013-July/000291.html has more info [15:17:49] I'm not.. [15:18:31] Been acting like this since the "automatically logging you in to all Wikimedia sites" page went away last week. [15:18:39] Firefox 22.0 [15:18:53] pretty sure there is already a bug open for that [15:19:18] T13|college: Thanks for asking the right channel. Another user reported the problem yesterday and today at a project channel, too. [15:20:15] Yeah, I try to get the right channel... It was a toss up between here and #wikimedia and I thought it was more on the technical side. [15:20:23] gry, is there a ticket for it already? [15:20:27] Possibly clear cookies and check for the issue again. [15:20:30] I don't know. [15:21:03] !bzqs unified login [15:21:09] http://bots.wmflabs.org/~wm-bot/dump/%23wikimedia-tech.htm [15:21:36] !bugsearch unified login [15:21:56] ... petan is infobot database linking broken? [15:22:08] o.o [15:22:09] http://bots.wmflabs.org/~wm-bot/db/%23mediawiki.htm [15:22:13] btw T13|college - do you use https://meta.wikimedia.org/wiki/Tech/News ? might be nice for you - get it delivered to your talk page [15:22:19] well, it could be... but it shouldn't be [15:22:42] This infobot is linked to mediawiki which says there should be a !bugsearch... [15:22:52] omg yes it was [15:22:53] !bugsearch unified login [15:22:53] https://bugzilla.wikimedia.org/buglist.cgi?quicksearch=unified+login [15:22:57] damn it [15:22:59] another bug [15:23:00] there we go. :) [15:23:26] I seem to have a knack for finding bugs... [15:24:22] that's good, we like that :) [15:25:37] Technical_13: you might also like to check out https://www.mediawiki.org/wiki/Quality_Assurance and join the mailing list in case you want to not just file bug reports but actually help write the infrastructure that prevents future bugs [15:30:00] !b 17937 [15:30:00] https://bugzilla.wikimedia.org/17937 [15:30:08] Unified login is not unified (at least for me) [15:30:14] ^^ Think that is it.. [15:30:54] Apparently not... 2009... [15:31:21] andre__: ^ just wanted to give you a heads-up in case you can help Technical_13 narrow down the problem & find the relevant bug that's open, or open 1 [15:32:29] I don't see any current ones in https://bugzilla.wikimedia.org/buglist.cgi?bug_status=UNCONFIRMED&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&field0-0-0=product&field0-0-1=component&field0-0-2=alias&field0-0-3=short_desc&field0-0-4=status_whiteboard&field0-0-5=content&field1-0-0=product&field1-0-1=component&field1-0-2=alias&field1-0-3=short_desc&field1-0-4=status_whiteboard&field1-0-5=content&list_id=219514&query_format=adva [15:32:43] * Technical_13 wishes those URLs were shorter... [15:33:04] * andre__ reads backlog [15:33:13] "Please try clearing cookies" doesn't mean "go to bugzilla and look for duplicates" [15:33:51] gry: It does it on both my android and my laptop, and I've cleared my cookies a couple times for other things since I first noticed it. [15:35:49] What browser version? [15:36:37] Firefox 22.0 on laptop. [15:38:50] Works correctly on Opera... Testing Chrome, Safari, IE next... [15:39:19] Suppose Chrome and Safari could work, dunno about IE. [15:39:56] I have latest versions of all Firefox/Chrome/Safari/Opera and IE7 [15:41:43] Chrome is good... [15:42:06] Only Firefox-specific problem? [15:46:35] Gry safari is locking up... will restart amd try again... [15:47:16] Got Safari to load... It is slow, but after about 20 seconds it picks up I am logged in... [15:47:42] Testing IE7 [15:47:58] ftr, I finially found it and my android is using FF22 as well... [15:49:40] IE7 works as well.. .So, I would assume it is a Firefox if not v22 specific thing. [15:51:43] Technical_13, I wonder if you run into https://bugzilla.wikimedia.org/show_bug.cgi?id=51644 [15:51:49] (sorry for the late answer, was in a meeting) [15:53:42] Not sure andre__ but I can play with that idea in a bit... If so, I can add a note to the ticket that it seems to only affect Firefox (22.0)... [15:55:35] more testing would be great [16:17:06] apergos, parent5456: hello [16:17:41] Hey [16:18:14] hello [16:18:52] ohh new stuffs [16:18:55] i have just commited a change that adds LZMA-compressed revision texts [16:19:38] so I should feed it content dumps now ? [16:19:43] Mhm just saw it. [16:20:04] yeah [16:21:32] though in adding LZMA to the code, i kind of made a mess out of the build process, i should look at that at some later point [16:22:16] it seemed to build and run [16:23:23] I guess the read doesn't do anything new yet (displaying rev text) [16:23:35] yeah, it should, it's just that it's not pretty: the platform independent cmake file for dumps calls GCC-specific makefile for LZMA [16:23:42] heh [16:24:12] it should display the text (or rather first ~80 character of ot) [16:24:17] so size-wise, this version produces a format a bit bigger than the standard bz2 file [16:24:31] not that we expect it to be efficient at this stage, compressing each rev separately [16:24:38] a lot bigger [16:25:35] for tenwiki: uncompressed XML: 99 MB, bz2 compressed XML: 4 MB, my compressed incremental dump: 37 MB [16:26:24] I have 4.6M for the compressed incremental [16:26:26] that's weird [16:28:03] that has to be wrong [16:28:37] was that for full history dump? [16:28:56] ah [16:29:07] the build in fact bailed someplace along the line [16:29:08] this is why [16:29:15] ah [16:29:33] oh [16:29:39] different executable name [16:29:40] sec [16:30:03] yeah, i should have told you, i changed that [16:30:38] well I shoulda read the output instead of assuming (after not seeing obvious errors, just a couple whines) [16:33:13] yep that's going to be a lot larger [16:35:56] so what's next after this? (it's still running) [16:36:33] i think i'll write the format specification, as you suggested [16:36:53] ok. I would suggest snding that to the list(s) too [16:38:29] ah next week we are supposed to do mid term evaluations btw [16:38:48] very soon compared to start time, because of your compressed timetable [16:38:58] and then make the indexes into a tree; right now, each index is basically just a single array, which is why i limited it to 64k entries [16:39:03] 41M for me. hm [16:39:21] ah, I was curious about that :-) [16:41:15] I know it's way too soon to do real world timing tests, but at some point we will want to see how much things slow down as things are written to the file [16:41:56] i think it doesn't make sense to do that before i have reasonable compression implemented [16:42:21] oh yeah, you can't do it now [16:43:01] just that because you're not doing only serial writes, you'll want to look out for that later [16:44:20] yeah [16:45:21] so did anything come up that you want to chat about? [16:46:42] 5: User:DerHexer, 4 revs [16:46:42] terminate called after throwing an instance of 'std::ios_base::failure' [16:46:42] what(): basic_ios::clear [16:46:42] ^CΑκυρώθηκε (core dumped) [16:46:45] ouch [16:46:50] and about the evaluations: yeah, i know about that, looking at what i said has to be done by then, i'm mostly just missing the library for reading dumps; since we decided to have XML output instead of that, i'll do that after i fix the indexes [16:47:20] yup, there's the deletion stuff too [16:47:51] hm repeatable core dump it seems [16:48:55] hmm, did you delete the dump from yesterday? because if the file already exists, it will try to add to it and i haven't tested that much yet [16:49:40] ah [16:49:43] nope [16:49:45] sec then [16:49:52] well not sec, it takes a while to run it [16:49:59] so I'll poke at it later [16:50:10] ok [16:50:34] if there's an actual bug there I'll report it tomorrow :-) [16:51:11] blubb... SUL problems O_O [16:51:56] ok, if that's everything for today, see you tomorrow [16:52:01] I got nothing else [16:52:02] parent5456: ? [16:52:13] Nope I'm good. [16:52:21] Hmm why's my username like that. [16:52:23] ok, great to see steady progress [16:52:28] see ya tomorrow [16:52:43] which username were you wanting, parent5456/ [16:52:44] ? [16:52:59] parent5446, my usual one. [16:53:10] Hmm already in use. [16:53:14] I never remember which number you have :-D [16:54:06] :P [16:55:15] is it in use by you or someone else? [16:55:20] you could try ghosting it [16:55:25] By me I think [16:55:45] Anyway, I'll see you both tomorrow. [16:56:03] all righty, have a good day [16:58:56] bye [18:16:20] @techs: In the last days a lot of problems with SUL: Always logged out... :( [18:30:27] ah, seems to be https://bugzilla.wikimedia.org/show_bug.cgi?id=51644 [18:35:45] greg-g: hoo: take a look at https://gerrit.wikimedia.org/r/#/c/74682/ and https://bugzilla.wikimedia.org/show_bug.cgi?id=51644 ? [18:37:26] sumanah: Want to deploy that soonish? [18:37:42] or rather have it deployed [18:38:12] lets see, one second [18:40:53] hoo: well, I figure it's greg-g's decision whether to get it deployed, but you are one of the people who could review it [18:41:28] is this needed to happen now now, or should it be a part of the mediawiki train? anomie ^^ [18:42:10] Code looks sane at first glance [18:42:41] CHOO-CHOO!!!!!!! [18:42:56] * sumanah puts on a striped hat! [18:43:19] All aboard the dinosaur train! [18:43:42] T13: Dinosaur train? [18:44:01] greg-g: Would be nice to get it deployed, if people are complaining (which it seems they are). Needs a +2 first though. [18:44:36] T13: Thanks, now I have that in my head. [18:44:45] sumanah: https://en.wikipedia.org/wiki/Dinosaur_Train [18:45:55] ROFLMMFAO!!!!! [18:46:36] the TOC on that page looks broken :P [18:46:54] a pteranodon cannot lay an egg that hatches into a T-Rex [18:47:02] also, children's steampunk I guess [18:47:06] Buddy was adopted. [18:47:07] sumanah: It was adopted [18:48:37] anomie: I have a 21 month old, what's your excuse for knowing that? [18:48:46] He's also got children? ;) [18:48:47] T13: 6 year old [18:49:21] s/he [18:49:31] oh wait [18:49:32] nm [18:49:42] yeah, anomie has a young'un [18:50:09] T13: anomie: should I attempt to watch an actual ep? I've seen the intro song [18:51:28] * sumanah does like trains and dinosaurs but possibly not at the level of a child [18:51:34] sumanah: If you want. It's a kid's show, of course, but they at least try to not be too inaccurate beyond the fact that they have talking dinosaurs riding a train and stuff like that. I think it's on Netflix, if you have that, otherwise check your PBS schedule I guess. [19:31:21] We're going to ride-rride-ride... The Dinosaur Train!!! [19:49:00] anomie: What about the keep me logged in now? [19:49:38] hoo: What about it? [19:49:40] " * With securelogin, token and user cookies are set encrypted only, when user opts-out of https." Is this blocking/ a big problem? [19:49:49] Do you want to deploy it soon? :P [19:50:08] It works for me, but I can't play around with https [19:51:07] All those problems are if $wgCentralAuthUseOldAutoLogin = true and $wgCentralAuthSilentLogin = false, the latter of which is no longer the case on WMF wikis and the former of which will be deleted in If81765b2. [19:51:24] I saw that [19:53:02] ok, in these cases I guess I'll +2 that now... the code isn't super beautiful, but nothing in CA is that :P [19:54:11] Merged now... go ahead to deploy it [19:54:52] greg-g: window? Or should we just go with the lightning deploy slot? ^ [20:05:19] Erg, I got another 502 bad gateway. [20:05:37] legoktm: Please report it on the bug. [20:05:38] With a URL. [20:06:32] Ryan_Lane or notpeter: If there are intermittent nginx issues (502 bad gateway error), who's the correct person to talk to about that? [20:06:40] likely me [20:06:54] are all users getting this in europe? [20:06:56] err [20:06:57] the US? [20:06:59] US [20:07:02] https://bugzilla.wikimedia.org/show_bug.cgi?id=50891 [20:07:08] It's intermittent. A page refresh resolves it. [20:07:13] Only via HTTPS. [20:07:15] hm [20:07:18] that makes no sense [20:07:21] I copied your Gmail address yesterday. [20:07:30] yeah. I saw it this morning [20:07:33] I'm wondering if there's a log of these that can be looked at? [20:07:39] not really [20:07:43] :-/ [20:07:48] https is a stupid transparent proxy [20:07:57] it doesn't really do anything [20:08:12] Except occasionally error. ;-) [20:08:14] it's possible that a squid or varnish server isn't responding properly [20:08:32] we may have error logs still enabled [20:08:38] And that would trigger nginx to get a bad gateway error? [20:08:41] I should probably syslog them somewhere central [20:08:55] the gateway would be the backend [20:11:02] legoktm: Thanks for adding a comment. [20:11:16] np. [20:15:43] so, looking at the error logs, there's some timeouts to the backends [20:15:50] to a specific ip address [20:15:59] which makes me wonder if that ip isn't bound on all hosts [20:18:36] Internal IP address? [20:28:53] Elsie: is this IPv6 or 4? [20:29:31] IPv4 for me, as far as I know. [20:29:34] legoktm? [20:29:39] IPv4 [20:29:47] wow. wtf: https://ganglia.wikimedia.org/latest/?r=hour&cs=&ce=&s=by+name&c=SSL%2520cluster%2520eqiad&tab=m&vn= [20:29:51] I don't think I'm using IPv6 at all (or at least I'm not doing so intentionally). [20:31:50] ssl1004 seems to be receiving substantially more load? [20:33:00] Hi [20:33:09] hi Fox2k12 [20:33:12] can't login into wiki [20:33:20] http://i.imgur.com/J6jaF8n.jpg <-- this is all I get [20:33:46] Fox2k12: You're trying to log in to wikimediafoundation.org. [20:33:54] yes [20:33:54] That's not a public wiki. :-) [20:33:57] Did you previously have an account at wikimediafoundation.org, Fox2k12? [20:34:00] login.wikimedia.org# [20:34:05] Are you Fox2k11 or Fox2k12 ??? [20:34:11] 11 [20:34:18] only 2k12 on irc [20:34:39] anomie|away: LD this afternoon should be ok [20:34:42] login.wikimedia.org is a completely separate site from wikimediafoundation.org. [20:35:00] but it redirects me to that page [20:35:09] What is it? :-) [20:35:41] Fox2k12: If you can describe where you started, that would be helpful. [20:35:55] when i login into de.wiki or en.wiki it works but only until I close the tab even when i checked the rember me check box [20:36:08] Okay. [20:36:10] it worked before.. my browser also does not delete cookies... [20:36:17] okay [20:36:19] I think there is a bug with remember me. [20:36:35] I get an email saying.. "page x has been edited" i click the diff link in the mail [20:36:50] and the articel shows up but I'm logged out [20:36:59] Right... you're bouncing between HTTPS and HTTP. [20:37:08] E-mails use HTTP, but you log in via HTTPS. [20:37:23] Bad usability is bad. [20:37:43] so i go to login.wikimedia.org to login and i get forwarded to the site.. [20:37:57] There are a few issues getting conflated. [20:38:22] well when i close the tab and go back later I'm logged out either.. [20:38:46] as i said it worked before had never a problem wenn clicking on links [20:39:45] Fox2k12: You understand the difference between HTTPS and HTTP? [20:39:55] of course.. [20:40:00] how can i fix that? [20:40:06] Fix what? [20:40:31] the https issue ? [20:40:50] As I said, there are a few issues that are getting conflated here, I think. [20:41:00] okay [20:41:02] E-mails and other links should be using HTTPS. [20:41:05] They don't currently. [20:41:24] There should be a cookie that forwards any request from HTTP to HTTPS. That may not be working properly. [20:41:38] There should be a cookie that allows you to stay logged in for up to 30 days. That may be not be working properly. [20:41:40] apparently [20:42:01] I said this was a bad idea about 20 times before it was deployed. I got ignored. ;-) [20:42:42] well.. [20:43:08] Ryan_Lane: The SSL traffic spike would be related to &wpStickyHTTPS or equivalent, right? [20:43:08] unlikely [20:43:12] it's not enabled [20:43:35] I have a patchset in to enable it [20:43:37] Ryan_Lane: You sure? :-) [20:43:41] https://login.wikimedia.org/wiki/Main_Page [20:43:44] I see it in the URL there. [20:43:49] it's only enabled for that one wiki [20:43:55] Aren't people passing through that one wiki now? [20:44:10] again, I have a patchset in to enable it everywhere :) [20:44:16] Hmmm. [20:44:26] yes, but that wouldn't explain this: https://ganglia.wikimedia.org/latest/graph_all_periods.php?c=SSL%20cluster%20eqiad&m=cpu_report&r=hour&s=by%20name&hc=4&mc=2&st=1374525841&g=network_report&z=large [20:44:36] this specifically: https://ganglia.wikimedia.org/latest/graph.php?r=year&z=xlarge&c=SSL+cluster+eqiad&m=cpu_report&s=by+name&mc=2&g=network_report [20:46:21] I wonder if https results in google are showing up more [20:46:29] we need the rel=canonical change [20:48:25] well at least i know now that it's http(s) releated when i change the email link into https link it works... [20:48:40] well that a workaround for now... ;) [20:53:21] heh [20:53:30] well, this is an issue with too much traffic going through ssl [20:53:41] it wasn't really intended to handle this much with only 4 nodes [20:53:59] I blame HTTPS Everywhere [20:54:01] j/k [20:55:09] I prefer a Secure connection ;) [20:55:42] I have a good feeling it's google indexing https [21:05:38] Does enwiki really have 0 jobs or is it broken? https://en.wikipedia.org/w/api.php?action=query&meta=siteinfo&siprop=statistics [21:06:06] legoktm: broken... [21:06:15] Been that way for a bit.. [21:06:21] Is there a bug about it? [21:06:39] andre__: ^^^ ??? [21:06:42] Found it. https://bugzilla.wikimedia.org/show_bug.cgi?id=50635 [21:06:47] :p