[00:42:21] odder: Still needed ? [00:42:43] yes [00:44:25] (we're in pm) [10:40:40] is there a maximum filesize for Wikimedia? [10:41:00] Other than the 100MB upload limit? [10:41:02] I might have a > 1,4 GiB file and am wondering if it might be uploaded to Commons even by a dev [10:41:10] Oh it's possible, for sure [10:41:16] Question is whether apergos will let me :) [10:41:55] :-)) [11:05:14] will let you do what? [11:05:18] *apergos does the backread [11:05:25] yes, do it [11:05:31] well, what is it? :-D [11:05:35] ( RoanKattouw ) [11:05:47] odder I might have a > 1,4 GiB file and am wondering if it might be uploaded to Commons even by a dev [11:06:04] odder: How much data do you have for these conference videos, all together [11:06:16] 26-30 GiB [11:06:44] not sure yet [11:08:01] what's the conference? [11:08:53] 10th birthday of Polish Wikipedia [11:09:01] cool [11:09:21] I say put em up, just don't upload too many gb in the same day, you know the drill Roan [11:09:28] OK [11:09:39] http://www.thenews.pl/1/12/Artykul/55748,Polish-Wikipedia-marks-10th-anniversary [11:09:46] we're still trying to string the host along til we can make swift take over [11:09:50] someday... [11:10:56] or, for the German-speaking people here, http://www2.polskieradio.pl/zagranica/de/news/artykul161583.html and http://de.wikipedia.org/w/index.php?title=Wikipedia:Kurier&oldid=94085558 [11:12:04] Cool, I didn't know dewiki had their own signpost [12:04:42] RoanKattouw, how's the state of things? [12:05:14] Some prot rel URL fallout [12:05:17] Some dead Apaches [12:05:23] No reponsive ops people :( [12:05:31] The job queue kind of took off this morning [12:05:45] Two Apaches that were job runners and something else (memc or ES) went to swapdeath [12:05:55] Job runners were sitting on 3G of mem, killing them fixed it [12:06:10] But three other Apaches are still completely dead [12:08:45] three apaches should have 0 impact on the site [12:09:26] I'd hope not [12:09:47] memc is still stopped on 163, though [12:09:57] I think we can put it back up [12:10:15] what did it have on it, just apache and memcached? [12:10:26] But I wasn't gonna do that myself because 1) mark took it down and 2) I've already technically broken the rules once by killing the job runners on srv163 without anyone approving it [12:10:31] Job runners, too [12:10:31] I should do some shell requests then maybe some other stuff [12:10:35] But Mark stopped those [12:10:41] I see [12:10:51] Ah [12:10:55] I mean srv161 [12:11:05] It shows as memc down in Nagios [12:11:23] WTF: spence check_job_queue CRITICAL 09-27-2011 12:09:36 52d 14h 15m 6s 3/3 (Return code of 127 is out of bounds - plugin may be missing) [12:11:29] *RoanKattouw files RT ticket for that [12:11:38] Is that the one we use? [12:11:43] It can't be... [12:11:59] It's been failing for 52 days [12:12:15] So there's something wrong with the chec [12:12:15] Yeah, the enwiki one wherever it is was showing problems with the queue [12:12:20] so that's gotta be the old one [12:15:15] Filed #1580 [12:21:39] Reedy: I could use your help with something [12:22:38] Sure? [12:23:11] Look at /home/catrope/enotif-clean [12:23:27] That's a list of wikis where [[MediaWiki:Enotif_body]] is overridden locally [12:23:42] I am editing it on each of those wikis to do s/fullurl/canonicalurl/g [12:23:54] Could you start from the bottom? I just did anwiki (the topmost one) [12:24:48] Can I grant myself staff rights? [12:24:56] You don't have them? [12:25:00] staff/dev/ops/whatever [12:25:01] Nope [12:25:06] Sure [12:26:10] Apparently sysadmin doesn't have editinterface O [12:26:11] But staff does [12:26:15] lol [12:26:21] seems a daft oversight [12:26:29] *RoanKattouw gives Reedy +staff [12:26:51] there [12:26:59] Cheers [12:28:29] Hmm, brwiki has {{SERVER}}{{localurl:Special:Watchlist/edit}} [12:28:31] That's weird [12:28:40] as does zhwiktionary [12:28:45] Been in there since 2006 oO [12:29:09] The zh wiktionary has from 2005 [12:29:15] "MediaWiki default" [12:29:23] You can change {{SERVER}}{{localurl:foo to {{canonicalurl:foo [12:29:30] And {{fullurl:foo to {{canonicalurl:foo as well [12:31:49] What the ... [12:31:59] Why am I not allowed to edit interface messages on bswikibooks [12:32:16] Oh now I am, weird [12:33:31] srwiki is in your list but has no modifications [12:33:44] same for sq.. [12:33:49] OH [12:35:09] It has an override, probably [12:35:13] Should have [12:35:21] I ended up at MediaWiki with an accented e [12:37:45] heh [12:38:09] We have such funny wiki names [12:38:15] I've done bswiki, now doing bugwiki [12:38:54] "Yer freendlie {{SITENAME}} notification seestem" [12:39:06] What language is *that*? [12:39:09] sco [12:39:11] haha [12:39:16] Thought so [12:39:22] I saw ang (Old English) the other day [12:39:47] I could understand most of it, but eerily enough I derived that from my knowledge of Dutch and Frisian rather than English [12:42:40] Hmm, is cy Welsh? [12:42:53] i think so [12:42:54] Yup [12:42:55] I like "pms" [12:43:01] 'cy' => 'Cymraeg', # Welsh [12:43:03] haha [12:43:32] http://pms.wikipedia.org/w/index.php?title=Ciaciarade:Reedy&redirect=no [12:43:42] Google thinks it's italian, but won't translate it [12:44:11] dick-wiki [12:44:14] Well, diqwiki but still [12:44:39] Just deleted the page on pms wiki as it was blanked [12:46:24] upto mtwiki [12:46:53] eswikisoruce [12:47:41] did all the wikis on the 118 list yesterday get switched? [12:48:03] (as in, they are running on 1.18 now) [12:48:46] Yes [12:48:49] ok [12:48:53] yay [12:48:55] See also /h/w/common/wikiversion.dat [12:48:55] I'll do some testing with them later today [12:49:14] We should almost do another deployment this week, instead of "everything else" next week [12:49:22] s/almost// [12:49:39] I'll propose it to robla later [12:49:57] cause it's gonna be like and then 750+ more [12:50:13] hell, even do 5-10 a day till friday [12:50:14] wikiversionS.dat [12:52:16] Hmm or do some of the big ones separately [12:52:28] Like, do the remaining 9 of the top 10 wikis in a separate batch or something [12:52:38] big ones separately = +1 [12:52:47] Cause not only is it gonna be "oh, and 800 more" but in those 800 we're gonna have enwiki, dewiki, commons, etc. [12:53:00] Hmm, the job queue seems to have calmed down [12:53:06] Good [12:53:23] To do those deploys, it's gonna take quite a while [12:53:25] Well, it's gone down over the past hour [12:53:28] the other wikis should be simpler [12:53:54] Hmm yeah it's pretty much back to normal now, but it's been erratic all day [12:54:15] We also still have four dead Apaches [12:56:18] I seem to be getting bad sessions errors quite frequently [12:56:31] I just get slowness [12:56:49] I sometimes have to hit an edit page twice because the first time CentralAuth hasn't fully logged me in yet [12:56:54] So it gives me a permissions error [12:57:03] This only happens if I go to the edit URL directly, not if I go via the page view [12:57:17] frwiktionary [12:57:26] Ah, the middle [12:57:41] Oh [12:57:44] I just did fywiki [12:57:54] So I guess we're done [12:57:57] yay [12:58:01] I'll run a script to check, but it's a slow script [12:58:05] Why doesn't testwikis job queue get dealt with automagically? [12:58:15] I don't know? [12:58:17] (i did log an RT ticket about it) [12:58:27] Do the regular runners not run testwiki jobs? [12:58:33] Not that i've seen [12:58:41] I realised a few weeks ago, there was a HUGE backlog [12:58:43] I can kind of see why not, but then there should be a process on srv193 running testwiki jobs [12:58:44] i'm presuming it must not be in some list [12:58:55] indeed, considering it sits relatively idley [12:59:59] Oh [14:31:18] Morning robla [14:31:29] mornin! [14:31:56] what a fine day for a deployment ;-) [14:32:01] apergos, RoanKattouw_away and I were wondering if it might make more sense to do some deployments of 1.18 every day this week [14:32:06] Rather than doing ~800 next week [14:32:15] And also, staggering the big wikis somehow [14:32:25] or at the least to do the big 9 or whatever in a separate last batch [14:32:42] and prolly enwiki by itself cause that's how it likes to be [14:33:13] robla: tldr? [14:33:25] hexmode: yup [14:33:49] <^demon|away> Do we have links to consensus on all the wikis for migrating to 1.18? [14:34:00] apergos: Reedy: RoanKattouw_away: sounds like a reasonable idea, but let me talk more after my phone call [14:34:07] worksforme [14:34:11] Yaa [14:34:15] ^demon|away: no, this is a technocracy :-P [14:34:44] <^demon|away> :) [14:35:54] <^demon|away> Ok, off to the dentist. Just wanted to get a last troll in before my nervous breakdown. [14:36:18] happy toothfixing/maintenance, ^demon|away [14:36:29] <^demon|away> Thanks :) [14:36:35] <^demon|away> Hopefully I'll be back in one piece. [14:37:17] good luck [15:01:22] so.... Reedy and apergos: re: spreading out 1.18. the selfish side of me says that I really don't have the time to spread the pain out more than we already have [15:01:53] You'd rather everything exploded at the same time? :D [15:01:59] I was not thinking of delaing next week [15:02:07] but rather inserting a day or two before that [15:02:14] Exactly, just interject it [15:02:16] *delaying [15:02:24] the reason for not going earlier is because binasher is going to be gone [15:02:39] Ooh [15:02:40] crap really? [15:02:52] when is he out? [15:02:55] I see it on the calendar now [15:02:57] Wed-Tue [15:03:07] So from tomorrow until the day before the big day [15:03:09] bite me. how unfortunate is that... [15:03:22] Are the schema changes done on all wikis? [15:03:23] he'll be gone during our deployment, but will be nominally reachable on Tuesday [15:03:33] RoanKattouw: my understanding is that he's wrapping up [15:03:40] Oh, right, the deploy is on Tuesday [15:03:45] that's been the reason for most of the explosions I'm aware of [15:03:48] It shows up on Wednesday on my calendar because of the time difference [15:03:55] Yes, s7 and s5 exploded [15:03:56] s6 didn't [15:03:59] I know s3 happened [15:04:06] s2, too, I think [15:04:07] well if we have something load-related we won't know til big-traffic wikis get converted [15:04:13] Mostly wondering about s1, and to some degree s4 [15:04:20] apergos: nlwiki is converted [15:04:35] apergos: that was the point of nlwiki. last time around, it was a good litmus [15:04:37] that may or not be enough [15:04:40] yes, I recall [15:04:51] The job queue was kind of crazy today [15:05:05] RoanKattouw: on nlwiki? [15:05:10] On the Apaches [15:05:13] Don't know which wiki [15:05:20] Lot of job queue-related CPU utilization [15:05:22] Took some boxes down [15:05:41] I would like us to ensure that there are no Apaches that run jobs in combination with either memc or ESE [15:05:42] *ES [15:05:44] job queue generally has been a problem, even prior to this deploy [15:05:59] I haven't talked to ops about that but Mark seemed supportive of that general notion [15:06:06] However, Mark has been incommunicado for most of the day [15:06:15] I'll suggest it in -ops now [15:07:24] I would like us to have no ES on apaches [15:07:35] which is a related discussion if not exactly on topic [15:09:23] That's already gonna happen [15:09:30] The stuff on srv* is gonna move to es1001-1004 [15:09:46] But it's not gonna happen before Tuesday [15:09:47] yes [15:09:57] I've been talking with ben about it off and on [15:10:04] it can't happen soon enough, is my view [15:29:49] RoanKattouw_away, https://bugzilla.wikimedia.org/show_bug.cgi?id=31188 :-)) [15:56:23] RoanKattouw, thanks again :) [17:41:09] http://meta.wikimedia.org/wiki/Talk:Limits_to_configuration_changes [17:41:50] updates? [17:45:28] jorm: http://en.wikipedia.org/w/index.php?title=Wikipedia:Wikipedia_Signpost/2011-09-19/In_the_news&diff=prev&oldid=451404929 [17:47:31] heh [17:48:01] It's unfortunate how one guy gave the German word for "leader" such an enormously negative connotation [17:50:33] can I propose this for deletion since it is original research? [17:51:12] it's completely non-notable. I don't see any references to any credible media source [17:51:17] :) [18:22:34] Krinkle: there's a problem on nlwiki that looks very likely to be a common.js problem there. as a logged in user, the html in loggedOutTalkPage() still shows up [18:24:05] (actually, it looks like it's only a problem on the post login page) [18:24:50] robla: Yeah I can reproduce it. That looks bad [18:24:52] checking out now [18:25:05] robla: Bug in mediawiki [18:25:10] mw.user.name is null on that page [18:25:14] weird [18:26:27] wgUserName rather [18:28:23] robla: https://bugzilla.wikimedia.org/show_bug.cgi?id=31192 [18:38:31] thanks Krinkle [18:42:05] binasher_: I don't recall: have you been including the LQT schema updates in the updates you've been making? (sorry for asking yet again) [18:42:57] yes, it should be applied everywhere that's had any of the 1.18 updates now [18:43:11] binasher_: So, which schema updates are NOT yet done? [18:43:15] robla, should push REL1_18's LQT back into 1.18wmf1 [18:43:28] RoanKattouw: the ones i don't know about [18:43:35] haha [18:43:37] Is s1 (enwiki) done [18:43:39] ? [18:43:47] Like, do we have any more master switches coming up? [18:43:57] but really, the current enwiki and commons masters need to be done, which should be today [18:44:01] You should attend meets RoanKattouw ;) [18:44:22] all of their slaves are done though [18:44:24] s/meets/meetings/ [18:44:27] I do [18:44:35] Had a conflicting meeting [18:44:55] OK, so there are master switches for s1 and s4 in our future [18:44:57] Reedy: at least the trunk version....that's what werdna tested [18:44:58] *RoanKattouw shudders [18:45:15] TW have been using SVN for the last year fine [18:50:35] Back in 15 or so [18:55:04] Nikerabbit: is Lior Kaplan someone you especially want to encourage to consider going to the Mumbai hackathon? [19:56:22] AaronSchulz: I need het deploy help [19:56:29] You know how dumpInterwiki is a wiki-less script? [19:57:08] yes [19:57:13] That makes it impossible to run the 1.18 version of dumpInterwiki [19:57:27] Or, more generally, any version of dumpInterwiki but the one belonging to the MW version that aawiki is on [19:57:53] aawiki is convention, but does it have to be aawiki and not some 1.18 wiki? [19:58:11] it shouldn't matter [19:58:13] aawiki is hardcoded in MWScript.php [19:58:19] I can work around it [19:58:35] By changing MWScript.php locally to use metawiki or somethign [19:59:16] But really, if I run a wikiless script and pass in a wiki argument, I want that argument to be used [19:59:20] Not overridden with aawiki [20:00:59] yep [20:01:14] *AaronSchulz wishes you could run those without any wikiname hacks at all [20:04:24] I'm not sure whether you can [20:04:31] But I should be able to instruct it to use a different version [20:06:13] *AaronSchulz changes it [20:07:03] catrope@fenari:/home/wikipedia/common$ mwscript dumpInterwiki.php --o php-1.17/cache/interwiki.cdb [20:07:05] --wiki must be the first parameter. [20:17:58] AaronSchulz: while we're at it... what I have is a python script which calls one of the maintenance scripts, using MWscript as a wrapper (all good) but this maintenance script can take a plugin, which anyone could write a new one, it could have any path whatsoever, but *we* pass in a plugin that's an extension... living in directory php-1.x. I need that version number [20:18:41] so I wrote a 5 line php script to do it but I am sure it would be nicer to have an "official" multiversion script that, given the name e.g. enwiki will return the version string [20:20:54] the script does the very obvious open /usr/local/apache/common-local/wikiversions.cdb, dba_fetch the thing taken right from argv[1], and echo it. that's all I need [20:21:25] where is your script? [20:21:41] not checked in even, not even deployed. [20:21:52] pastebinz [20:21:55] sure [20:22:54] http://pastebin.com/FCab2edk [20:22:59] something like that I need [20:26:42] mostly ok [20:26:54] *AaronSchulz puts it in the queue [20:27:41] the main things are: called with one arg just like it looks in all.dblist, and outputs one thing, the versioned part of the directory [20:27:44] thanks dude [20:36:10] RoanKattouw: should work now [20:44:22] brion, brion-mac, neilk_ - are y'all having a regular meeting with Trevor to talk about mediawiki.next stuff? I thought I saw that on a calendar... [20:45:05] sumanah: hasn't really been happening [20:45:22] sumanah, we've been trying to make sure we at least have a lunch-n-greet or a poke-whats-up standup [20:45:24] neilk_: as in, it's been attempted & never got off the ground? [20:45:25] not always on time :) [20:45:45] for my part, I haven't had much to contribute, but this should change this week [20:46:14] got it [20:46:20] and trevor gave a nice update on his & inez's status at lunch :) [20:46:24] :) [23:14:23] tfinc: Have you seen this? http://lists.wikimedia.org/pipermail/mediawiki-api/2011-September/002362.html [23:14:47] RoanKattouw: in a meeting [23:16:18] let me take a quick look [23:16:45] yeah, i saw roberts mail earlier today [23:16:48] Basically it claims all iOS mobile clients are now broken [23:17:07] our app works pefectly [23:17:27] and i haven't received any other contacts [23:17:31] I'm inclined to say it's their fault for 1) using that hack and 2) not paying attention to mediawiki-api-announce (the guy skipped it because I forgot to put BREAKING CHANGE in the subject, my bad there)