[00:06:30] <Brooke>	 Reedy: No lleaking!
[00:06:33] <Brooke>	 leaking!
[00:43:01] <TimStarling>	 !log put DolphinBrowser files in docroot/bits (from preilly's Ie3fefec6) and now running scap
[00:43:06] <morebots>	 Logged the message, Master
[00:53:57] <logmsgbot>	 !log tstarling Started syncing Wikimedia installation... : 
[00:54:02] <morebots>	 Logged the message, Master
[01:09:01] <Betacommand>	 Reedy: reading scroll back, it was dispenser
[01:10:41] <Betacommand>	 Reedy: poke me when you are around and Ill explain what Dispenser did
[01:34:28] <logmsgbot>	 !log tstarling Finished syncing Wikimedia installation... : 
[01:34:34] <morebots>	 Logged the message, Master
[02:24:46] <logmsgbot>	 !log LocalisationUpdate completed (1.20wmf5) at Tue Jun 19 02:24:46 UTC 2012
[02:24:54] <morebots>	 Logged the message, Master
[02:45:18] <Brooke>	 Ryan_Lane: Thread on wikitech-l about HTTPS (load) seems like it's for you.
[02:47:42] <logmsgbot>	 !log LocalisationUpdate completed (1.20wmf4) at Tue Jun 19 02:47:42 UTC 2012
[02:47:48] <morebots>	 Logged the message, Master
[09:04:44] <logmsgbot>	 !log nikerabbit synchronized wmf-config/InitialiseSettings.php  'Merge I8572d5f4 to fix workflowstates in Translate'
[09:04:49] <morebots>	 Logged the message, Master
[09:23:51] <Hydriz>	 Can any ops help me deploy 2 small changes?
[09:24:43] <Hydriz>	 https://gerrit.wikimedia.org/r/#/c/11837/ and https://gerrit.wikimedia.org/r/#/c/11840/
[09:28:16] <hashar>	 Hydriz: ops dont do mediawiki configuration changes: -]
[09:28:21] <hashar>	 platform engineering does though
[09:28:29] <Hydriz>	 platform engineer :P
[09:28:42] <hashar>	 (aak me, sam reed and other folks)
[09:28:46] <Hydriz>	 s/ops/platform engineer/
[09:28:56] <hashar>	 you need a review first :-]
[09:29:02] <hashar>	 and you are not alone hehe
[09:29:03] <hashar>	 https://gerrit.wikimedia.org/r/#/q/status:open+project:operations/mediawiki-config,n,z
[09:29:05] <Hydriz>	 :(
[09:29:07] <hashar>	 doh
[09:29:11] <hashar>	 so mannnyyyy pathes
[09:29:13] <hashar>	 patches
[09:29:24] <hashar>	 I love how making that repo public attract lot of volunteeers
[09:29:26] <hashar>	 \O/
[09:29:37] <hashar>	 I will process some this afternoon
[09:29:54] <Hydriz>	 :)
[09:30:03] <hashar>	 I have added myself as a reviewer on the two changes above
[09:30:03] <Hydriz>	 hope I get pinged
[09:30:09] <hashar>	 I might process them this afternoon
[09:30:13] <hashar>	 no guarantees though
[09:31:10] <Hydriz>	 I see much lesser shell requests in Bugzilla after this repo went public, I think?
[09:32:19] <p858snake|l>	 Hydriz: well they should all still have their matching Bz tickets
[09:32:35] <Hydriz>	 yeah
[09:32:40] <Hydriz>	 lesser as in open bugs
[09:39:44] <hashar>	 for a long time it has been only Sam Reed handling them
[09:39:58] <hashar>	 and he was really busy with API + MediaWiki release + other stuff
[09:40:15] <hashar>	 so we have offloaded some of his work to the community hehe
[09:56:21] <Nikerabbit>	 hmm
[09:56:31] <Nikerabbit>	 no logging anymore here? Anyway, I'm running scap
[09:56:47] <logmsgbot>	 !log nikerabbit Started syncing Wikimedia installation... : Updating TranslationNotifications
[09:56:52] <morebots>	 Logged the message, Master
[09:57:15] <Nikerabbit>	 lol
[10:02:06] <hashar>	 Nikerabbit: congrats :-]
[10:02:13] <hashar>	 and no fatal errors around
[10:02:25] <Nikerabbit>	 hashar: that would be unexpected
[10:02:45] <hashar>	 always expect the unexpected when running scap :-D
[10:11:43] <logmsgbot>	 !log nikerabbit Finished syncing Wikimedia installation... : Updating TranslationNotifications
[10:11:48] <morebots>	 Logged the message, Master
[10:12:24] <Nikerabbit>	 so scap takes 25 minutes nowdays
[10:15:16] <hashar>	 :(
[10:15:42] <hashar>	 I think one of the main issue is copying the l10n caches around
[10:16:12] <hashar>	 aka hundred of servers * X wmf branches * hundred of MBytes of cache
[10:17:29] <domas>	 use bittorrent!  
[10:17:56] <p858snake|l>	 use sneakernet!
[10:18:09] * p858snake|l  nominates domas to do it

[10:18:30] <domas>	 though I'd suggest having less languages
[10:18:32] <domas>	 and less branches
[10:18:33] <domas>	 :)
[10:18:50] <aokomoriuta>	 Hello, all. Anyone knows how can we change the links in [[Special:MobileFeedback]]?
[10:19:16] <Ryan_Lane>	 hashar: we should compress the localization cache before delivery, and uncompress after
[10:19:35] <Ryan_Lane>	 it was nearly 10x smaller when Tim tested it
[10:19:49] <hashar>	 isn't rsync able to do that?
[10:19:54] <hashar>	 something like rsync ---compress
[10:20:11] <Ryan_Lane>	 meh, easier to do it when the cache is created
[10:20:21] <Ryan_Lane>	 and have the apaches uncompress when its delivered
[10:20:49] <Ryan_Lane>	 it also means fenari spends less time compressing the same file over and over again
[10:21:27] * hashar  starts learning C and implements a compression cache in rsync

[10:21:45] <Ryan_Lane>	 or, just compress it ahead of time, and afterwards ;)
[10:21:53] <Ryan_Lane>	 *uncompress afterwards
[10:22:01] <Ryan_Lane>	 it's a single line in the scripts
[10:22:12] <Ryan_Lane>	 well, one on each side
[10:22:24] <Ryan_Lane>	 the localization cache scripts could add the one liner to compress
[13:52:35] <Dispenser>	 Tim-away: https://commons.wikimedia.org/w/api.php?action=query&prop=extlinks&titles=File:Zoar_Chapel_of_Rest,_Llanteg_-_Rabbit_Runway^_-_geograph.org.uk_-_1161626.jpg
[14:03:29] <logmsgbot>	 !log hashar synchronized wmf-config/InitialiseSettings.php  'Syncing misc changes
[14:03:34] <morebots>	 Logged the message, Master
[14:03:39] <hashar>	 : 9df0438 {{bug|36972}} activate the patroller group on nn.wiki
[14:03:39] <hashar>	 : a5fdddb {{bug|36895}} Restrict media upload to autopatrolled users on sv.wikisource
[14:03:41] <hashar>	 : 69c7e87 {{bug|35712}} Add an alias to Help namespace in ml.wikisource
[14:03:42] <hashar>	 : 19675e2 {{bug|37679}} zhwiki let rollbacker abusefilter-log-private
[14:03:43] <hashar>	 : f666f29 {{bug|37661}} Change Vietnamese Wikibooks logo
[14:03:44] <hashar>	 : b3fe639 {{bug|37644}} Enable subpages on be.wikimedia.org for NS 0 and 4
[14:03:45] <hashar>	 that is the change synced
[16:21:26] <Merlissimo>	 could somebody explain me why some language names have been changed to lower case, but not all?
[16:22:52] <Reedy>	 Merlissimo: yup. Because that's how they are in that language
[16:23:17] <Reedy>	 https://gerrit.wikimedia.org/r/#/c/7306/
[16:23:43] <Reedy>	 https://bugzilla.wikimedia.org/36819
[16:26:31] <Merlissimo>	 so "englisch" , but "Simple English" - strange
[16:27:31] <Reedy>	 Changes may be incomplete
[17:58:19] <Nemo_bis>	 omg en.wiki still has the vandalized anti-watchlist css version
[17:58:29] <Nemo_bis>	 such a wiki can't be helped
[17:58:58] <Betacommand>	 Nemo_bis: ??
[18:11:01] <Krenair>	 Nemo_bis: link?
[18:13:58] <Nemo_bis>	 Betacommand, Krenair, I don't know where to start, maybe https://en.wikipedia.org/wiki/Wikipedia:Customizing_watchlists
[18:14:17] <Nemo_bis>	 or the hundreds KBs on https://en.wikipedia.org/wiki/MediaWiki_talk:Common.css#Nope.2C_it_is_still_bolding_stuff
[19:13:01] <logmsgbot>	 !log mlitn synchronized wmf-config/InitialiseSettings.php  'Enable AFTv4 on testwiki'
[19:13:06] <morebots>	 Logged the message, Master
[19:28:28] <logmsgbot>	 !log mlitn Started syncing Wikimedia installation... : Update ArticleFeedbackv5 to master
[19:28:33] <morebots>	 Logged the message, Master
[19:45:05] <logmsgbot>	 !log mlitn Finished syncing Wikimedia installation... : Update ArticleFeedbackv5 to master
[19:45:11] <morebots>	 Logged the message, Master
[20:36:59] <logmsgbot>	 !log mlitn synchronized php-1.20wmf5/extensions/ArticleFeedbackv5  'desc'
[20:37:04] <morebots>	 Logged the message, Master
[21:52:23] <Excirial>	 Anyone around in tech?
[21:53:01] <LeslieCarr>	 there are many around, what is the issue ?
[21:53:20] <Excirial>	 I'm having some fun with a userpage that refuses to load on EnWiki: http://en.wikipedia.org/wiki/User_talk:BeryllLA
[21:54:34] <Excirial>	 As far as i can see it was created and not deleted, but it cannot seem to load correctly. Editing it (http://en.wikipedia.org/w/index.php?title=User_talk:BeryllLA&action=edit) displays a very neat error warning.
[21:55:23] <LeslieCarr>	 hrm, purge is fail
[21:55:40] <Excirial>	 *Nods*
[21:56:09] <James_F>	 Intriguing issue.
[21:57:14] <LeslieCarr>	 i have no idea...
[21:57:37] <Reedy>	 The revision was never saved
[21:58:02] <AaronSchulz>	 page_latest is 0
[21:58:17] <AaronSchulz>	 that gets set in transaction, but should never be committed until the rev is added
[21:58:20] <Excirial>	 It should have been create with the AFC helper script
[21:58:59] <Excirial>	 But if i look at the editor that used the script, i can only see a contribution that edits the AFC page - edit to the users talk page is MIA from contributions altogether.
[22:01:34] <masti>	 looks like the name contains pipes which cannot be a part of URL
[22:02:53] <Excirial>	 To me those look like two lowercase "L"'s
[22:04:44] <masti>	 hmm, right
[22:06:21] <Excirial>	 Thanks Aaron :)
[22:24:31] <AaronSchulz>	 binasher: you know how chronologyprotector check the master pos? What if cached that per-user?
[22:24:36] <AaronSchulz>	 *if we cached
[22:24:56] <AaronSchulz>	 (per-user to avoid contention)
[22:26:20] <binasher>	 AaronSchulz: do we also do that for semi-anon logged out users based on a cookie value?
[22:27:13] <AaronSchulz>	 yes
[22:27:27] <binasher>	 what would we key on in that case
[22:29:18] <AaronSchulz>	 the session ID or something?
[22:35:18] <Platonides>	 I think the master pos is stored in the session
[22:35:36] <Platonides>	 it then refuses to serve any content previous to the last pos you saw
[22:35:49] <Platonides>	 so that you can't view events getting backwards
[22:36:03] <Platonides>	 which is the point of chronologyprotector
[22:37:56] <AaronSchulz>	 Platonides: yeah it already works the way I'd like
[22:38:02] <AaronSchulz>	 for some reason I thought it was hitting the master more
[22:49:45] <MaxSem>	 okay, it's our window and I'm deploying a config change - anyone I can collide with?
[22:51:46] <AaronSchulz>	 binasher: does revision have the page_timestamp index yet?
[22:51:55] <AaronSchulz>	 I mean page_user_timestamp 
[22:52:20] <MaxSem>	 *deploying*
[22:53:02] <binasher>	 AaronSchulz: nope.. i don't think it ever actually made it onto the schema migration list 
[22:54:35] <AaronSchulz>	 yeah, well it's new and in tables.sql
[22:54:53] * AaronSchulz  is happy to have fucked up binasher's projections ;)

[22:55:35] <logmsgbot>	 !log maxsem synchronized wmf-config/InitialiseSettings.php  'bug 37611'
[22:55:41] <morebots>	 Logged the message, Master
[22:56:03] <binasher>	 AaronSchulz: noooooooo
[22:56:31] <binasher>	 hah, wow.. brion reverted patch-revision-user-page-index.sql in 2007
[22:58:00] <binasher>	 AaronSchulz: how often do we get "All edits by a user to this page" queries?
[22:58:20] <Reedy>	 Isn't that the one I re-added pre-emptively?
[22:58:27] <Reedy>	 ie the feature hasn't been done yet
[22:58:28] <binasher>	 yup
[22:58:32] <AaronSchulz>	 binasher: never right now
[22:58:44] <binasher>	 Reedy: what's the feature?
[22:58:55] <Reedy>	 something to use that index
[22:58:56] <AaronSchulz>	 you mean the feature that's not going to happen
[22:59:03] <binasher>	 hah
[22:59:12] * AaronSchulz  greps for 'revision' queries

[22:59:28] <AaronSchulz>	 Reedy: I really don't think we can do that
[22:59:59] <binasher>	 AaronSchulz: how useful is the usertext_timestamp index?
[23:00:02] <Reedy>	 something else to hide behind $wgMiserMode
[23:01:09] <AaronSchulz>	 we can live without that feature I bet
[23:03:24] <Reedy>	 https://bugzilla.wikimedia.org/show_bug.cgi?id=10788 Filter page histories by user, or contributions by title
[23:09:21] <logmsgbot>	 !log maxsem synchronized wmf-config/InitialiseSettings.php  'bug 37611, plan B'
[23:09:25] <morebots>	 Logged the message, Master
[23:11:19] <binasher>	 aggh why is revision.rev_sha1 a var when its a constant length 
[23:11:28] * binasher  smacks self for applying that 

[23:41:10] <TimStarling>	 AaronSchulz: you noticed a higher image backend load?
[23:49:38] * jeremyb  starts hacking on gerrit again

[23:56:46] <jeremyb>	 gerrit is sooooo slow
[23:57:23] <jeremyb>	 s/gerrit/java/ ? ;-P
[23:57:45] <TimStarling>	 java is slow?
[23:57:56] <jeremyb>	 sometimes!
[23:57:59] <TimStarling>	 you write your web apps in assembly no doubt
[23:58:41] <jeremyb>	 in this particular case mysqld is slow too: starting via an init script. or i guess starting up at all