[02:30:04] !log LocalisationUpdate completed (1.20wmf6) at Thu Jun 28 02:30:03 UTC 2012 [02:30:17] Logged the message, Master [02:58:43] !log LocalisationUpdate completed (1.20wmf5) at Thu Jun 28 02:58:42 UTC 2012 [02:58:49] Logged the message, Master [07:31:59] good morning [13:12:47] o hai. [13:13:00] git-related questions go here? [13:13:19] in particular, i need the rights to push tags [13:20:16] <^demon> Jens_WMDE: Which repo? [13:20:30] ^demon: already got it. [13:20:37] ^demon: thanks! [13:20:44] <^demon> Oh ok :) [13:27:00] ^demon: will you be available to review some changes I made in mediawiki-config . [13:27:02] ? [13:28:26] <^demon> https://gerrit.wikimedia.org/r/#/q/status:open+project:operations/mediawiki-config+owner:Hashar,n,z [13:28:26] <^demon> ? [13:28:39] xactly [13:28:57] I am polishing up 'beta' [13:29:05] which run out of master not of a wmf branch [13:29:08] so need some tweaks [13:29:21] also I have set up a bits cache this morning (bits.beta.wmflabs.org) [13:29:36] which uses varnish, should works like bits.wikimedia.org [13:32:32] <^demon> aww, changeMerge.test has a memory leak. Can't enable it :( [13:33:32] are you deploying gerrit 2.4 ? [13:34:04] <^demon> Next week. [13:34:06] <^demon> Need to pick a day [13:34:20] can't wait for the rebase button :-] [13:35:13] <^demon> Me too :) [13:35:50] <^demon> I'm not 100% sure what 13316 is doing. [13:36:36] <^demon> I understand the symblinks, but why are they needed? [13:36:40] <^demon> *symlinks [13:46:32] ^demon: ah, i might need your help after all! [13:46:45] <^demon> What's up? [13:47:02] ^demon: the repo is https://gerrit.wikimedia.org/r/#/admin/projects/mediawiki/extensions/Wikibase [13:47:12] ^demon: I have no idea :-/ I just did the same that is done for wmf parts [13:47:21] <^demon> hm. [13:47:25] aka the docroot/bits/static-1.20wmf6/ [13:47:39] that is the path used by resource loader I think [13:47:46] ^demon: i tagged it. after i found the place in gerrit to give myself the right to push the annoted tag, i pushed it. now i want to remove it :) [13:48:07] <^demon> Hehe, removing tags is dark magic :) [13:48:39] <^demon> You'll need to grant yourself Push rights on refs/tags/*, with the force option. [13:49:03] <^demon> Then, do the following: `git tag -d mytag && git push origin :refs/tags/mytag` [13:49:10] yeah [13:49:11] <^demon> The latter is the "delete from the remote repo" part [13:49:20] the latter i actually knew [13:49:29] but how to get the right? [13:49:36] <^demon> I'll do it so you can see. [13:50:06] <^demon> Ok, it should be set now. [13:50:17] thank you! [13:50:21] <^demon> We'll revoke it after you're done, since it's kinda dangerous to leave open :) [13:51:01] done. you can remove it now. [13:51:07] thanks a lot! [13:51:19] <^demon> You're welcome. [13:54:00] <^demon> hashar: Hehe, https://gerrit-review.googlesource.com/#/c/36510/ [13:56:58] ^demon: I should have learned Java :-D [13:57:11] had an occasion to enter Sun as a first job [13:57:19] took the internet way instead :-d [13:57:26] <^demon> :) It's not hard [13:57:59] <^demon> But yeah, found out yesterday there's a loophole in gerrit. If you close a branch to changes (so nobody can push for review anymore), you could restore an old change to that now-closed branch. [13:58:03] <^demon> Kinda stupid :p [13:58:13] well I don't even know where to look at nor how to compile gerrit [13:58:23] needless to say I have no idea how to start a java app ;-] [14:04:08] <^demon> hashar: `git clone && cd gerrit && mvn package` [14:04:08] <^demon> :) [14:08:32] ahh maven [14:08:46] the great install tool that download ton of stuff on my disk [14:14:47] ^demon: any idea where maven write all the stuff it downloads? [14:15:07] <^demon> The only stuff it downloads should be in that package you're building. [14:15:10] <^demon> Dunno about anything else. [14:15:37] well If I ever find out a dir holding 4GB of .war, I will not that is the maven cache dir ! [14:15:53] [INFO] BUILD SUCCESS in 7:03 [14:16:44] ask a few [14:27:38] ATTENTION!! rebooting maganese (aka gerrit) for dist-upgrade [14:31:49] now you say it [14:33:06] At least it was prior, not post [14:33:58] no wonder it was slow [14:35:13] ^demon: why reviews of draft changes are public? [14:35:39] <^demon> Bug in hook, need to fix it. [14:53:29] it's sitll a bit slow [14:56:41] <^demon> Nikerabbit: It's always gonna be a tiny bit laggy after a restart, caches will be cold. [15:02:48] ^demon: any objection to adding docroot/bits/static-master sym links ? ( https://gerrit.wikimedia.org/r/#/c/13316/ ) [15:03:09] <^demon> No objections, I just don't understand it. Anyone who does? [15:10:54] ^demon: well we ship urls such as //bits.wikimedia.org/static-1.20wmf5/skins/vector/images/search-ltr.png [15:11:04] the 'static-1.20wmf5' comes from mwMultiVersion [15:11:27] on labs, the version is master, so the URL above would be : //bits.beta.wmflabs.org/static-master/skins/vector/images/search-ltr.png [15:12:05] bits virtual host is bound to /usr/local/apache/common/docroot/bits/ ;) [15:21:01] Reedy: would know :) [16:14:16] Ryan_Lane, ^demon: the plan is not to use php5-memcache anyway [16:14:19] but php5-memcache*d* [16:14:24] (confusing, I know) [16:14:53] yeah. I was just removing stuff for now, till we had a plan [16:14:57] and did it in production :) [16:15:52] Quick question, have we found the cause of yesterdays downtime? [16:17:23] Nathan2055|food: things being broken [16:17:25] heh [16:17:36] * Nathan2055|food laughs [16:17:45] we're still investigating [16:17:55] it's the same problem we've been having occasionally the past week or so [16:18:06] not sure the exact cause yet [16:51:52] thx Krenair [16:52:02] np [16:52:08] check the changes to make sure I didn't miss anything :P [16:52:46] already done [18:16:59] bah [18:17:02] connected again [18:24:25] so hmm I need to sync some changes on cluster [18:24:30] for mediawiki-config [18:24:35] ping reedy ^demon [18:26:58] > print "realm: $realm cluster: $cluster"; [18:26:59] realm: production cluster: pmtpa [18:27:00] \O/ [18:27:25] \o/ [18:27:42] syncing my labs hack on production :/ [18:28:54] hmm [18:29:03] I really need to write a test suite to test out CommonSettings.php [18:29:07] I thought about using phpunit [18:29:16] but will most probably end up using eval.php :-] [18:31:49] !log hashar synchronized wmf-config/CommonSettings.php [18:31:54] Logged the message, Master [18:34:17] Did someone blow up bits? [18:34:19] http://en.wikipedia.org/wiki/Main_Page [18:34:24] !log hashar synchronized wmf-config/CommonSettings.php [18:34:27] Somebody broke en.wiki. :( [18:34:30] Logged the message, Master [18:34:41] it's back up [18:34:49] grrr [18:34:53] Oh, there we go. [18:34:55] stylesheets were missing [18:35:04] and I did test the stylesheets paths :-( [18:35:20] Who just switched me to Classic? [18:35:35] To Monobook? [18:35:42] No, Classic. [18:35:49] i.e. Phone format. [18:35:53] Nathan2055: Ctrl-Shift-R [18:35:58] reload page :-( [18:36:03] I did. [18:36:04] Or Cmd-Shift-R if you're using a Mac. [18:36:10] Try it again. [18:36:17] Ctrl+F5 also work [18:36:20] The only way to fix it was to switch nack to Vector in settings. [18:36:24] *back [18:36:34] Did someone mess up the CSS? [18:36:43] [20:34] hashar stylesheets were missing [18:36:45] [20:34] hashar and I did test the stylesheets paths [18:36:47] :P [18:36:51] Ah. [18:36:56] hashar: SURE. ;P [18:36:58] hehehe, they reverted the chagne [18:37:07] You scared me for a second. [18:37:11] :P [18:37:16] I'm off now. [18:37:17] Bye! [18:37:36] backing off change [18:37:38] Selling trouts! Gets yours while the supply lasts! ;) [18:38:15] actually, i think we should go back to a geocities, circa 1997 style [18:38:18] blink tags everywhere [18:38:23] the dripping blood ani gif [18:38:35] ooo and the water reflection javascript applet [18:39:02] It would cut required server resources quite a bit i guess. [18:52:13] about to run a scap if that's cool with everyone [18:54:20] scap started [19:05:35] Excirial: which mean that even when doing lot of tests there still some magic that happens: ( [19:06:07] Isn't that what keeps a job fun and interesting? [19:06:55] it is sure exciting [19:07:04] * Damianz beats HASHAR with a CAPS lock [19:07:07] but preventing knowledge distribution is nowhere near fun [19:07:15] OH YEAH SORRY I FORGOT ABOUT MY CAPSLOCK KEY [19:07:39] :D [19:07:49] "Honey, i'm home, and i had such a blast at work! I dropped the worldwide average IQ by 30 points for an hour by blowing up some servers." [19:08:00] feel free to try it out HTTPS://GERRIT.WIKIMEDIA.ORG/r/#/c/13313/ then append uselang=en-x-capslock [19:08:19] !log kaldari Started syncing Wikimedia installation... : [19:08:25] Logged the message, Master [19:20:37] so hmm I am moving out now [19:20:41] since cluster is fine :-] [19:20:51] and knowledge is broadcasted for proper world improvement [19:20:57] see you tomorrow ! [19:39:49] hiiii guys [19:39:57] and particularly brion_ [19:40:05] I have a lucene question if anyone has answers [19:40:07] :) [19:41:13] hahahahaha [19:42:18] funny question eh? [19:45:27] ottomata, ask away [19:45:31] we'll see what we can answer :) [19:45:51] yay! [19:45:53] ok so [19:46:06] how much have we changed with our version of lucene? [19:46:10] just the udp2log stuff? or more? [19:46:24] jeremyb told me that we created lsearchd ourselves, so that we could query lucene as a service [19:46:45] if so, did we have to hack original lucene code to do that? or is it a standalone set of classes? [19:47:48] ottomata, should be pretty much a standalone server using the stock lucene library [19:47:55] cool, that's good [19:48:08] afaik we customized a couple of tokenizers or something for some languages but all as add-ons, no changes to core lib [19:48:10] do you know of any patches to the lucene source that we have done, other than the addition of udp2logger code? [19:48:22] ok that's good news [19:48:27] not that i know of :D [19:48:31] cooooool [19:48:35] i'm working on some lucene logging stuff [19:48:39] if that is true [19:48:44] then we shoudl be able to revert to upstream packages [19:48:46] MAYBE [19:48:50] if I can figure out some log4j weirdness [19:49:00] at the very least we should only have to add an import statement [19:49:05] very most* [19:49:39] rather than doing function search() { … if (UDPLogger.enabled) { UDPlogger.log(search_query); } … } [19:49:45] or whatever it is we are doing in lucene code right now [19:49:55] should be able to configure that in log4j.properties [19:49:55] no idea how that's done currently [19:50:05] yeah, that's how they are doing it right now [19:50:25] someone patched the lucene source (95% sure about that…unless I was looking in the wrong class, but ja) [19:51:58] ok, brion_ a q for you about that [19:52:01] not sure if you know [19:52:10] about log4j, buuuuuut [19:52:12] maybe you know [19:52:41] if there is a way to specify an import of a custom log appender [19:52:49] without modifying source code? [19:52:56] somethign in log4j.properties? [19:53:11] (or maybe you have no idea what I am talking about :p ) [19:53:59] (ottomata is getting class not found error when using a 3rd party logger class) [19:54:41] !log kaldari Finished syncing Wikimedia installation... : [19:54:47] Logged the message, Master [20:01:51] ottomata, i'm not familiar with whatever present logging code there is in there [20:02:20] yeah, its not current code, i'm trying to do some new stuff [20:02:34] that's more of a generic esoteric java + log4j question [20:02:38] but salright, i'll figure it out [20:02:42] thanks for your help! [20:29:52] ah holy crap, lsearch is huge! [20:30:01] i didn't realize that we had built this whole thing ourselves [20:38:42] lolololololol [20:45:47] yeah seriously lolz [20:45:55] i was all "who hacked the lucene source"? and the answer is [20:45:58] no one! [20:46:12] we just built our own search engine that uses lucene [20:47:22] hehehe [20:48:23] we have more of our own code for this than makes up lucene itself! [20:50:35] barkeep looks nice. [20:52:19] HASHAR, i love you. [20:52:30] ottomata: you sound... suprised... [20:52:42] i am so surprised! [20:52:53] <^demon> Well, back in the day I don't think solr was as big/stable as it is now. [20:52:54] i saw the UDPLogger patches [20:52:58] aye probably not [20:53:01] <^demon> So we didn't really have a search daemon to work with [20:53:10] <^demon> If we did it all over again, I imagine we'd just use solr out of the box. [20:53:24] <^demon> Or with very minimal glue. [20:55:15] that sounds reasonable to me. [21:01:42] The DB is in read-only. What's going on? [21:01:57] no recent changes for more than 5 mins... [21:02:31] Nathan2055: which db? [21:02:35] Evening guys, can anyone just update me quickly on the RO mode at enwp please? Are you fixing stuff or something? :) [21:02:40] ... [21:02:48] binasher: ? [21:02:55] was about to do that ;) [21:03:20] 1017 is 266s [21:03:21] I'm sorry, bad client. [21:03:25] What did you say? [21:03:40] we're not sure hwat'sup yet Nathan2055 [21:03:41] binasher: 5 enwiki slaves are lagged, 1 is fine [21:03:52] 240-300s [21:04:03] that's what's up, the question is why [21:04:21] anything stuck in the job queue? [21:04:37] Job queue trouble? [21:04:42] Don't know [21:04:47] unikely [21:05:12] db38, db36, db32, db59, db60, db12 [21:05:29] ok, Reedy - I'll let you get on for now, but I'll make sure #wikipedia-en's topic is updated to save people asking :) [21:05:36] Due to high database server lag, changes newer than 458 seconds may not appear in this list. - Returns on watchlist. [21:05:55] yup [21:06:12] binasher: looks db57 is showing a 5M+ unpurged transactions [21:06:53] 5 million? [21:06:55] wait, db57 is another cluster [21:07:27] Reedy: R/O is off on enwp [21:07:28] so it is [21:07:55] db38, db36, db32, db59, db60, db12 [21:08:01] Back online! [21:08:40] Taking almost two minutes to load my watchlist. [21:08:50] stop loading things when you know they don't work [21:08:55] 59 and 60 seem to be reading a lot more rows [21:09:21] they aren [21:09:21] t [21:09:32] lagged so are getting most actual queries [21:09:40] ah [21:09:58] this was all my fault :) [21:10:04] schema change? [21:10:06] running scap for Translate and ClickTracking updates [21:10:07] Blame features! [21:10:09] engarde! [21:10:12] kaldari: don't [21:10:16] It looks good, no lag notices. [21:10:22] ok [21:10:50] how long should I wait? [21:11:39] Reedy: yeah, there was something in the aftv5 migration that really shouldn't have been run as written, and i didn't mean for it to get into the binlog [21:12:03] have you just done something on db36? [21:12:35] ah I see, some comments from Krinkle on ClickTracking [21:13:00] kaldari: pushing patch now :) [21:13:18] Krinkle: thanks [21:13:38] !log asher synchronized wmf-config/db.php 'temp pulling db36' [21:13:44] Logged the message, Master [21:14:44] Krinkle: the logged event array needs to be exposed, tho [21:14:50] sure [21:15:13] Krinkle: otherwise lgtm, thanks for reviewing. [21:16:03] ori-l: (why though?) [21:16:12] the public array, not that I care, just curious :) [21:16:26] validating now, almost done [21:17:35] Krinkle: mostly tactical considerations. i have a user script with a UI for displaying events as they are logged. it was too big/risky to merge to clicktracking for now. so i just wanted to include the one bit of functionality that the extension had to implement itself. [21:17:54] db36 is being bad [21:17:57] ori-l: but do you need the event and the public array? [21:18:33] Krinkle: how else would the userscript side know about tracking calls? installing a global ajax handler doesn't work because some calls are made before the user script loads. [21:19:03] ori-l: Right, so the array is to start up with what happened before you listen to the event [21:19:09] alrighty [21:19:53] Krinkle: it's a kludge. i'm working on a re-write of a big chunk of that extension, but until it's ready i want what is already there to be a bit more transparent. [21:21:07] ori-l: I started a rewrite last december, if you're interested I can send it to you (it isn't finished so it may have syntax errors or incomplete code, but could be useful to see how I went with it) - I am unlikely to be able to finish that any time soon [21:21:25] Krinkle: yes, definitely! [21:21:57] kaldari: pushe [21:22:37] Krinkle: btw did you see the skeleton js extension thing on wikitech-l? [21:22:59] nope? [21:22:59] not yet [21:24:29] Krinkle: http://www.mediawiki.org/wiki/Extension:SkelJS / https://github.com/atdt/skeljs [21:25:00] Krinkle: (i requested a gerrit repo & will migrate once it's up) [21:30:33] <^demon> ori-l: Is this meant to be an example for people to point to? If so, that kinda makes me leery about having open pushing available to it. [21:31:01] ok, changes merged, gonna run scap now unless anyone objects :) [21:31:53] ^demon: it should definitely be gated eventually. i thought of keeping it open push for initial development. but what do you think? [21:34:41] <^demon> Well, repo's made, you're owner, history imported from github. https://gerrit.wikimedia.org/r/gitweb?p=mediawiki%2Fextensions%2FSkelJS.git;a=shortlog;h=refs%2Fheads%2Fmaster [21:34:50] <^demon> Feel free to open yourself up to pushing if you'd like. [21:35:58] <^demon> :) [21:36:15] <^demon> Even got your first contribution https://gerrit.wikimedia.org/r/#/c/13413/ ;-) [21:39:49] ^demon|away: thanks! [21:39:56] !log kaldari Started syncing Wikimedia installation... : [21:40:02] Logged the message, Master [21:40:03] <^demon|away> ori-l: you're welcome. have a good afternoon :) [21:40:22] ^demon|away: you too! thanks again :) nice of you to do all this. [21:50:29] !log kaldari Finished syncing Wikimedia installation... : [21:50:35] Logged the message, Master [23:03:56] !log asher synchronized wmf-config/db.php 'returning db36' [23:04:02] Logged the message, Master [23:30:25] !log reedy synchronized php-1.20wmf6/includes/resourceloader/ResourceLoader.php [23:30:31] Logged the message, Master