[01:43:20] <anomie>	 Coren: a mysterious file?
[01:59:31] <Coren>	 I don't know.  Maybe it's an omen or something.
[02:08:24] <scfc_de>	 Unfortunately no obvious mysterious instances :-).
[02:10:51] <Coren>	 They're not accessible yet, annoyingly enough.  Annoying because that is, literally, the only thing missing.
[02:11:32] <Coren>	 If you could connect, you could query.
[02:16:31] <scfc_de>	 I think I will be able to wait a few hours :-) (or days?  Is tomorrow a holiday in the US?).
[02:16:44] <jeremyb>	 i wonder if the DYK in the banner (on login) rotates
[02:16:53] <jeremyb>	 > Did you know that your home isn't protected by default?
[02:16:55] * Coren  has little knowledge of USian holidays.

[02:17:18] <Coren>	 jeremyb: It does, though I think Petan's list is shortish.
[02:17:28] <jeremyb>	 i think it's not a holiday...
[02:17:33] <jeremyb>	 it's versioned?
[02:17:41] <Cyberpower678>	 Tomorrow is memorial day.
[02:17:57] <Cyberpower678>	 A holiday that means nothing to me.
[02:18:01] <anomie>	 Memorial day is May 27, not tomorrow.
[02:18:08] <carl-cbm>	 tomorrow is not a US holiday
[02:18:22] <jeremyb>	 it's not memorial day
[02:18:26] <jeremyb>	 Coren:  en.usa#holiday@group.v.calendar.google.com
[02:18:35] <Cyberpower678>	 Crap.  I thought today was next week already.
[02:18:39] <Cyberpower678>	 Oops.
[02:18:56] <anomie>	 I don't know whether tomorrow is a holiday, but if it is it's not one that WMF US staff get off.
[02:19:44] <jeremyb>	 US staff are certainly off on memorial day. which ain't tomorrow
[02:19:46] * Cyberpower678  is dumb.

[02:20:21] <Cyberstupid678>	 Der
[02:20:32] <Cyberstupid678>	 Durr
[02:20:33] <anomie>	 For example, May 17 was "National Defense Transportation Day". But I don't think anyone had it off for that reason.
[02:24:46] <scfc_de>	 To correct myself: I meant *today* :-) (at least in UTC; May 20th, "Whit Monday" this year).
[02:27:39] <jeremyb>	 what's that day?
[02:28:29] <scfc_de>	 jeremyb: http://en.wikipedia.org/wiki/Pentecost
[02:29:30] <scfc_de>	 http://en.wikipedia.org/wiki/Pentecost#Public_holiday suggests no holiday in the US.
[02:29:48] <scfc_de>	 (Or Canada :-).)
[02:44:14] <Coren>	 So yeah.  Those credentials will be useful once the network actually lets the labs talk to the databases.  :-)
[02:45:56] <scfc_de>	 Coren: And what's (who's) that hinging on?
[02:46:47] <Coren>	 Leslie, mostly.  Technically, I have all the bits to go fiddle on her networking gear.  In practice, especially since she's going to be within arm's reach in a few days, doing so would be profoundly unwise.  :-)
[02:50:10] <scfc_de>	 Well, so long as Tools still has to be properly puppetized, I think you would have some protection, but as I think you worked on this very intensely over the past few days it's a good idea to let someone else with enough sleep handle the scissors :-).
[03:02:57] <jeremyb>	 Coren: so what hostnames will we use in labs?
[03:03:47] <jeremyb>	 (for DBs)
[03:04:48] <jeremyb>	 scfc_de: also, there's the juniper vs. cisco issue
[03:05:58] <jeremyb>	 i see the mysterious file now!
[03:06:43] <jeremyb>	 Coren: i think you need to prepend a line on that file. '[mysql]'
[03:06:53] <jeremyb>	 or '[client]' is even better
[03:06:57] <jeremyb>	 iirc
[03:07:09] <Coren>	 It's not meant to be used directly, actually, although I might as well.
[03:09:47] <Coren>	 jeremyb: I dunno yet; we have an extra complication that we have varying ports.  LIke I said, I might actually fiddle something with NAT to simplify.
[03:10:35] <jeremyb>	 Coren: could have different bind-addresses
[03:10:54] <jeremyb>	 in my.cnf
[03:11:14] <Coren>	 I'll figure something out.  :-)
[03:11:20] <jeremyb>	 i'm sure :)
[03:12:51] <jeremyb>	 !ping
[03:12:51] <wm-bot>	 pong
[03:12:57] <jeremyb>	 wow, you slow
[03:13:04] <wm-bot>	 You are root identified by name .*@wikimedia/jeremyb
[03:41:05] <Coren>	 BBL
[07:56:01] <petan>	 jeremyb after restart the bot is always slow... at least first 2 minutes
[07:56:09] <petan>	 freenode is spamming it with lot of information to parse
[08:04:02] <petan>	 jeremyb why did you restart it btw? :o
[08:21:37] <hashar>	 !log deployment-prep removing thumbnails from the Gluster shared directory: cd /data/project/upload7 && find -maxdepth 3 -wholename '*/thumb'|xargs -n1 -P4 rm -v -fR
[08:21:41] <labs-morebots>	 Logged the message, Master
[09:20:31] <petan>	 !rq test
[09:20:32] <wm-bot>	 https://wikitech.wikimedia.org/wiki/Shell_Request/test?action=edit https://wikitech.wikimedia.org/wiki/User_talk:test?action=edit&section=new&preload=Template:ShellGranted https://wikitech.wikimedia.org/wiki/Special:UserRights/test
[10:37:33] <wm-bot>	 Warning: There is 1 user waiting for shell: Gilfriedman (waiting 0 minutes) 
[10:51:02] <wm-bot>	 Warning: There is 1 user waiting for shell: Gilfriedman (waiting 13 minutes) 
[11:04:32] <wm-bot>	 Warning: There is 1 user waiting for shell: Gilfriedman (waiting 27 minutes) 
[11:18:06] <wm-bot>	 Warning: There is 1 user waiting for shell: Gilfriedman (waiting 40 minutes) 
[11:31:31] <wm-bot>	 Warning: There are 2 users waiting for shell, displaying last 2: Gilfriedman (waiting 54 minutes) אור שפירא (waiting 5 minutes) 
[11:45:01] <wm-bot>	 Warning: There are 2 users waiting for shell, displaying last 2: Gilfriedman (waiting 67 minutes) אור שפירא (waiting 18 minutes) 
[11:58:30] <wm-bot>	 Warning: There are 2 users waiting for shell, displaying last 2: Gilfriedman (waiting 81 minutes) אור שפירא (waiting 32 minutes) 
[12:11:56] <wm-bot>	 Warning: There are 2 users waiting for shell, displaying last 2: Gilfriedman (waiting 94 minutes) אור שפירא (waiting 45 minutes) 
[12:25:17] <wm-bot>	 Warning: There are 2 users waiting for shell, displaying last 2: Gilfriedman (waiting 107 minutes) אור שפירא (waiting 58 minutes) 
[12:33:34] <Coren>	 Moo!
[12:34:26] <valhallasw>	 Coren: I think I have found the problem with jsub and in-script resource demands - command line parameters to qsub have a higher priority than the in-script ones, and qsub thus overwrites some of them
[12:34:54] <Coren>	 Wait, what problem with jsub?
[12:35:16] <Coren>	 Nobody told me about problems.
[12:35:17] <valhallasw>	 Coren: jsub ignores, for example,  #$ -l v_hmem = 500MB
[12:35:21] <valhallasw>	 in a script file
[12:35:25] <Coren>	 Aha!
[12:36:13] <Coren>	 Well, actually, if you use -continuous, any such would be ignored entirely -- it's not invoking the script directly but wrapping it.  That's not a bug, it's by design.
[12:36:40] <Coren>	 Honestly, if you use #$ in a qsub script, you're no newbie and can cope with qsub directly.  :-)
[12:36:58] <Coren>	 But I see how that could be seen to be confusing.
[12:36:59] <valhallasw>	 Coren: not really. #$ was the suggested way of doing things on the toolserver
[12:37:11] <Coren>	 Oh, it was?
[12:37:13] <valhallasw>	 'use #$ and qcronsub <script>'
[12:37:16] * Coren  hrms.

[12:37:37] <Coren>	 Well, I could check for and parse those in jsub as well.
[12:37:43] <Coren>	 Wouldn't be hard, really.
[12:37:47] <valhallasw>	 Coren: on the TS the parameters are parsed after submitting
[12:38:10] <valhallasw>	 the jsv script Merlissmo mentioned
[12:38:47] <valhallasw>	 I'm not sure whether that would allow things like -mem, though :/
[12:39:31] <Coren>	 It would, if jsub parses them.
[12:39:53] <Coren>	 I can simply feed them into ArGV
[12:40:03] <valhallasw>	 ah, yes, and then leave the default settings to the post-submit script
[12:40:10] * Coren  nods.

[12:40:23] <Coren>	 Do you have one that has #$ settings in it already?
[12:40:58] <valhallasw>	 cat /data/project/nlwikibots/bin/tvpupdater
[12:42:39] <valhallasw>	 I think the name also gets overwritten by jsub
[12:42:45] <valhallasw>	 and the in- and output files :-)
[12:44:12] <Coren>	 valhallasw: Give me a minute and see what I can do.  This is trivial enough to parse.
[12:44:20] <valhallasw>	 Coren: cool :-)
[12:49:04] <Cyberpower678>	 Coren, when I create a program to put in the crontab, what command in jsub will allow me to extend the memory usage?
[12:49:35] <Coren>	 Cyberpower678: -mem xxx where xxx is of the form 100m or 1g and so on.
[12:50:04] <Cyberpower678>	 Thanks.  I'm writing a script that's going to consume a bit of memory.
[12:51:38] <petan>	 Cyberpower678 bit of memory isn't really much, but it's hard to actually allocate 1 bit as 1 byte is lowest possible value which you can allocate address for ;)
[12:51:53] <Coren>	 petan: Smartass.  :-)
[12:52:59] <petan>	 you can use some bit operations actually to store bits into / from 1 byte - but despite this eats small amount of ram, it eats more cpu :( hence bool in c has 1 byte, instead of 1 bit
[12:53:21] <petan>	 in c# and java it has 4 bytes... bleh
[12:53:46] <Cyberpower678>	 petan, :p
[12:54:17] <Cyberpower678>	 petan, what about PHP?
[12:54:23] <petan>	 php is using GC?
[12:54:32] <petan>	 it depends on design...
[12:54:55] <petan>	 as long as you need to allocate pointer for your bool, it needs more memory than it actually /needs/
[12:55:19] <petan>	 in c# everything is object, even bool, so you need to create a reference for it, which is larger than whole object
[12:55:30] <petan>	 in php it's probably same
[12:56:35] * Cyberpower678  loves PHP

[12:56:55] <petan>	 well, I like some things about it at least :)
[12:57:24] <Cyberpower678>	 I like how easy it is to use it.
[12:58:36] <petan>	 in programming it is usually: easier it is to program something, more the produced binary code suck...
[12:59:36] <Cyberpower678>	 The sentence is too off.  I don't what you're saying.
[12:59:43] <valhallasw>	 Cyberpower678: it's easy to deploy, at least. I wouldn't call programming in it necessarily 'easy'.
[12:59:51] <Cyberpower678>	 Sorry petan.
[13:01:58] <valhallasw>	 Cyberpower678: petan's point is that harder programming languages generate faster/better programs
[13:02:32] <valhallasw>	 it's not really true, though :-)
[13:03:54] <petan>	 valhallasw no it's not, there are like 0.00001% of applications which are denying this :P
[13:04:11] <petan>	 but, for rest 99.9999% this apply
[13:05:27] <Cyberpower678>	 valhallasw, petan: Depends on the compiler.
[13:06:11] <petan>	 not really... it depends on the code as well
[13:06:33] <Cyberpower678>	 petan, true.
[13:07:14] <petan>	 easy-to-write languages let you write a crap you would never write in other languages if you actually saw the background which needs to be executed in order to do what you need
[13:07:16] <Cyberpower678>	 But if you write a program as efficient as possible and have compiler that as efficient as it can be, then you'll have one kickass program.
[13:07:45] <valhallasw>	 petan: except that's not true again. C++ is not easy to write, but still allows you to do that.
[13:07:48] <petan>	 if this was true for all languages, linux kernel wouldn't need to be written in c and assembler in order to be efficient
[13:08:11] <petan>	 valhallasw I consider c++ easy-to-write language :P
[13:08:40] * Cyberpower678  is learning C++ soon.

[13:08:46] <Cyberpower678>	 or is itC#
[13:08:48] <petan>	 definitely c++ is member of category of languages that allow you to write "beautiful crap"
[13:08:51] <valhallasw>	 C++ is not easier to write than C
[13:09:03] <petan>	 depends
[13:09:07] <Cyberpower678>	 I think C++.
[13:09:14] <sumanah>	 I think different people might have different aptitudes
[13:09:17] <petan>	 c++ allows you to do some things easier than you would need to in c
[13:09:27] <valhallasw>	 That is true, yes.
[13:09:35] <Cyberpower678>	 Of got compen and compsci next semester. :D
[13:09:39] <Cyberpower678>	 Can't wait.
[13:09:43] <petan>	 which is reason why Linus hates it
[13:10:20] <Cyberpower678>	 petan, Sinus?
[13:10:26] <Cyberpower678>	 Uh... Linus?
[13:10:33] <petan>	 c++ allows you to hide "evil code" and wrap it so that it looks nice even if it's crap
[13:10:34] <valhallasw>	 Torvalds.
[13:16:04] <jeremyb>	 petan: no special reason. you told me last time to use @restart instead. so i figured i should actually try it before i need it. i kinda expected it to reply to me...
[13:16:17] <petan>	 aha
[13:16:22] <jeremyb>	 (and tell me it was restarting or it was finished)
[13:16:34] <petan>	 jeremyb well, it replies in #wm-bot :P
[13:16:41] <jeremyb>	 yeah, i figured
[13:17:21] <petan>	 it doesn't send much to irc, as irc is first part that shutdown during shutdown... if you really need to watch details, you need to tail -f logfile
[13:17:27] <jeremyb>	 hashar: you know you can just give `find` a -delete ?
[13:17:46] <petan>	 jeremyb that is too simple for such a hacker like hashar
[13:18:18] <hashar>	 jeremyb: but then it serialize
[13:18:25] <jeremyb>	 hashar: possibly :)
[13:18:31] <hashar>	 jeremyb:  I used xargs -n1 -P<XXX>
[13:18:41] <hashar>	 to launch several parallel process :-]
[13:18:51] * jeremyb  goes back to scrollback

[13:19:41] <Pyfisch>	 Is gerrit.wikimedia.org the right place for bot source code or is it only for bigger projects?
[13:20:22] <petan>	 Pyfisch I recommend you to use github, but that is just because I don't like gerrit :P
[13:20:42] <petan>	 otherwise, it is probably a correct place for any kind of code
[13:25:04] <hashar>	 jeremyb: so yeah Gluster has a high latency, so I ran the deletions in // to speed up things.  xargs -n1 -P4   : pass 1 argument, run 4 parallel processes.
[13:36:52] <Coren>	 valhallasw: Should options in the script override explicit commandline arguments or vice-versa?
[13:37:49] <valhallasw>	 vice versa
[13:38:06] <Coren>	 That's what I'd though too.
[13:38:14] <valhallasw>	 thats what qsub does
[14:03:28] <gerrit-wm>	 New patchset: coren; "Tool Labs: Allow jsub to parse script qsub opts" [labs/toollabs] (master) - https://gerrit.wikimedia.org/r/64575
[14:04:12] <gerrit-wm>	 New review: coren; "Tested to work." [labs/toollabs] (master); V: 2 - https://gerrit.wikimedia.org/r/64575
[14:04:44] <Coren>	 valhallasw: Win.
[14:08:50] <valhallasw>	 Coren: seems to be working!
[14:13:30] <valhallasw>	 Coren: although reading Perl feels like I'm reading Haskell every time I do it.
[14:13:51] <Coren>	 Perl is teh winz.
[14:13:56] <valhallasw>	 "I should be able to understand what's going on here, but... err..."
[14:14:20] <valhallasw>	 If it makes you productive, by all means, use it :-)
[14:14:23] <Coren>	 The thing that makes perl have a steep learning curve is that it's highly idiomatic.
[14:15:41] * Coren  likes his Swiss Army Chainsaw.

[14:42:47] <greenrosetta>	 Eddie Haskell?
[14:43:54] <valhallasw>	 greenrosetta: http://en.wikipedia.org/wiki/Haskell_(programming_language) :-)
[14:44:07] <greenrosetta>	 so many languages
[14:54:52] <greenrosetta>	 Coren are you going to Amsterdam?
[14:55:03] <Coren>	 greenrosetta: Of course.  :-)
[14:55:40] <greenrosetta>	 More motivation for you to get the replication working.  Or else the developers might punish you
[14:56:10] <Coren>	 What do you mean, "get the replication working"?
[14:56:10] <greenrosetta>	 Dutch Oven'
[14:56:21] <greenrosetta>	 wait, you have DB replication in place?
[14:56:23] <Coren>	 It works very nicely, tyvm.
[14:56:31] <greenrosetta>	 wth... when did this happen
[14:56:50] <greenrosetta>	 checking my email
[14:56:58] <Coren>	 Oh, /you/ don't get to use it yet.  Network issues that will be fixed today.  :-)
[14:57:02] <Coren>	 But it's working.
[14:57:06] <greenrosetta>	 ah...
[14:57:25] <greenrosetta>	 so will each project get its own login?
[14:57:30] <greenrosetta>	 I'
[14:57:43] <greenrosetta>	 I'm assuming it will be the same login for each db
[15:01:09] <Coren>	 If you carefully inspect homes you might find some interesting stuff.
[15:01:45] <greenrosetta>	 replica?
[15:02:48] <greenrosetta>	 of course the host is missing... and where would we get the db names?
[15:06:06] <petan>	 Krinkle|detached ping
[15:06:09] <petan>	 @notify Krinkle
[15:06:09] <wm-bot>	 I'll let you know when I see Krinkle around here
[15:06:23] <petan>	 Coren are you familiar with public transport in amsterdam? :D
[15:06:38] <petan>	 last time I asked someone I was told I need to rent a bike there in order to get somewhere
[15:06:39] <valhallasw>	 petan: bikes without locks everywhere!
[15:06:41] <valhallasw>	 ;-)
[15:07:00] <petan>	 RoanKattouw you are local aren't you
[15:07:03] <petan>	 :P
[15:07:19] <valhallasw>	 I don't think RoanKattouw wants to be called local to Amsterdam :D
[15:07:28] <petan>	 is he not?
[15:08:21] <anomie>	 hashar: Unless plans have changed, later on today I'll be toggling $wmgUseCodeEditorForCore to true on Beta Labs for testing. I know there are issues with that file being cached, what's the appropriate way to work around that?
[15:08:29] <valhallasw>	 He's Dutch (or Frisian, really, I think), but not from Amsterdam
[15:08:40] <hashar>	 anomie: that file ? 
[15:08:59] <anomie>	 hashar: Ugh, I forgot to say which file. wmf-config/InitialiseSettings-labs.php
[15:09:04] <hashar>	 ha
[15:09:13] <anomie>	 The mtime of wmf-config/InitialiseSettings.php gets checked, but the -labs version doesn't.
[15:09:16] <hashar>	 anomie: I never heard of any cache trouble for the configuration
[15:09:34] <hashar>	 anomie: the mediawiki-config is deployed on beta by Jenkins  and it takes care of touching the IS.php file :-)
[15:09:55] <anomie>	 Oh, good. I don't have to worry about it then because Jenkins does it for me (:
[15:10:11] <hashar>	 anomie: https://integration.wikimedia.org/ci/view/Beta/job/beta-mediawiki-config-update/436/console
[15:10:11] <hashar>	 there is a touch at the end
[15:10:44] <anomie>	 Thanks!
[15:11:14] <RoanKattouw>	 petan: Where do you want to go?
[15:11:22] <petan>	 probably to venue :D
[15:11:32] <RoanKattouw>	 And valhallasw is right that I'm not local to Amsterdam; I used to be local to the Netherlands, and am now local to San Francisco
[15:11:39] <RoanKattouw>	 petan: From where?
[15:11:48] <petan>	 airport
[15:11:52] <RoanKattouw>	 Right
[15:12:13] <RoanKattouw>	 So you'll need to take the train to Amsterdam Muiderpoort station, it's a 15-min walk from there
[15:12:21] <hashar>	 petan: have you considered renting a bike ?
[15:12:21] <petan>	 aha cool
[15:12:31] <RoanKattouw>	 There are no direct trains from Schiphol to Muiderpoort, you can change at either Amsterdam Centraal or Duivendrecht
[15:12:43] <RoanKattouw>	 ns.nl has a travel planner that shows you all the trains
[15:12:50] <petan>	 hashar well, yes... but... :D I don't know if I want to travel in bike clothes
[15:13:12] <RoanKattouw>	 Hah, who wears bike clothes
[15:13:24] <petan>	 idk... like everyone here :P
[15:13:39] <valhallasw>	 Let me guess, you also wear an helmet? ;-)
[15:13:44] <petan>	 of course lol
[15:13:49] <RoanKattouw>	 Bike clothes are for people that take cycling seriously and have expensive bikes and wear helmets and stuff
[15:13:57] <petan>	 I got expensive bike :D
[15:14:04] <RoanKattouw>	 i.e. recreational cyclists
[15:14:22] <petan>	 mhmm... ok
[15:14:25] <RoanKattouw>	 For utilitarian cycling (cycling simply because you want to go somewhere) you don't need any of that :)
[15:14:44] <petan>	 we have nothing like "utilitarian cycling" in here or I don't know about it...
[15:14:49] <RoanKattouw>	 I wouldn't recommend riding from the airport, but riding around the city should be fun
[15:14:51] <valhallasw>	 So what about bike messengers? ;-)
[15:15:13] <petan>	 bike messengers here in prague really do wear helmets ;) and if you were one of them, you would do that as well...
[15:15:18] <RoanKattouw>	 petan: Yeah it's not really a thing outside .nl and .dk really. Just spend a day in AMS and you'll see what I'm talking about soon enough :)
[15:15:32] <petan>	 aha ok
[15:15:39] <valhallasw>	 RoanKattouw: in .dk everyone does wear helmets, from what I've heard.
[15:15:44] <petan>	 so... I will probably take a train from airport and then walk
[15:16:10] <valhallasw>	 I'm still somewhat disappointed Amsterdam doesn't have the Paris/Bordeaux/etc bike rental system
[15:16:17] <valhallasw>	 http://en.wikipedia.org/wiki/Velib et al
[15:16:32] <RoanKattouw>	 petan: http://goo.gl/maps/8rajj
[15:16:40] <petan>	 ok
[15:17:08] <petan>	 !venue is http://goo.gl/maps/8rajj
[15:17:09] <wm-bot>	 Key was added
[15:17:11] <RoanKattouw>	 valhallasw: A bike rental system implies that there exist people that don't own their own bikes. That notion is laughable over here ;)
[15:17:22] <valhallasw>	 RoanKattouw: that is true
[15:17:33] <petan>	 I can see myself with laptop in centre of Amsterdam typing !venue to irc :D
[15:17:38] <RoanKattouw>	 There is a bike rental system at the train stations but joining it isn't really convenient
[15:18:13] <petan>	 RoanKattouw wait. so I need to get my own bike? no possibility to rent it?
[15:18:26] <RoanKattouw>	 Oh there are plenty of rental places
[15:18:28] <valhallasw>	 petan: there are a few bike rental places
[15:18:36] <RoanKattouw>	 It's just that there isn't a short-term rental thing like in Paris or DC
[15:18:39] <petan>	 plenty... few... well :P
[15:18:41] <RoanKattouw>	 But daily rentals are plenty
[15:18:51] <RoanKattouw>	 I don't know about the hostel, but the hotel I
[15:18:55] <RoanKattouw>	 'm staying at has bike rentals
[15:19:43] <anomie>	 RoanKattouw: http://goo.gl/maps/MdpgH ?
[15:19:44] <petan>	 I heard that in Amsterdam they only accept cash on most places?
[15:20:07] <petan>	 anomie what is that?
[15:20:14] <valhallasw>	 petan: credit cards are not widely accepted, EMV debit cards are
[15:20:15] <RoanKattouw>	 anomie: Oh right, that train to Centraal continues to Muiderpoort, so I suppose that is a direct train :)
[15:20:20] <RoanKattouw>	 I stand corrected
[15:20:22] <anomie>	 petan: Train from the airport to Muiderpoort
[15:20:32] <RoanKattouw>	 Also, https://www.mediawiki.org/wiki/Amsterdam_Hackathon_2013/Muiderpoort-venue_directions
[15:20:37] <anomie>	 RoanKattouw: I was checking whether Google was telling me wrong ;)
[15:20:41] <RoanKattouw>	 No, it's right
[15:20:44] <petan>	 aha
[15:20:46] <RoanKattouw>	 I just misremembered
[15:21:13] <petan>	 !airport-centre is http://goo.gl/maps/MdpgH :)
[15:21:13] <wm-bot>	 Key was added
[15:21:52] <RoanKattouw>	 Hah, so I'm apparently giving a workshop
[15:22:13] <RoanKattouw>	 I wonder when they were going to tell me
[15:23:09] <valhallasw>	 RoanKattouw: they just expect you to improvise, obviously ;-)
[15:23:15] <RoanKattouw>	 Clearly
[15:23:19] <Krenair>	 Are the deployment-prep instances restricted to WMF shell users?
[15:23:50] <petan>	 Krenair no
[15:25:55] <Krenair>	 They have copies of the mediawiki-config and everything set up so you could look through the final values of config there with the maint scripts?
[15:26:15] <petan>	 well, probably yes...
[15:29:04] <Krenair>	 Is it possible to get access to one of those instances then?
[15:32:09] <anomie>	 Krenair, petan: "mwscript eval.php --wiki=enwiki" and the like work there, yes. I don't know who handles granting access, maybe hashar (or else I'm sure he knows).
[15:32:56] <Krenair>	 Yes that's basically what I had in mind
[15:35:53] <Krenair>	 Ryan_Lane or andrewbogott maybe?
[15:38:54] <andrewbogott>	 Krenair, I have no idea what their policy is for giving out project membership (if there is any policy).  Probably best to ask hashar when he appears.
[15:39:24] <hashar>	 there is no policy :-D
[15:39:35] <hashar>	 Krenair: ping me by email please, I am leaving my coworking plcae
[15:39:43] <hashar>	   hashar@free.fr 
[15:39:45] <hashar>	 *wave
[15:42:29] <Cyberpower678>	 petan, your attention is needed on #wikipedia-en-accounts-devs
[15:42:39] <Krenair>	 sent
[15:42:52] <Cyberpower678>	 wm-bot doesn't recognize me as an admin there either.
[15:45:03] <valhallasw>	 Coren: I have added two functions to jsub: one to set once=true if called as qcronsub, one function that makes sure h_vmem >= virtual_free, which means toolserver users with badly configured virtual_free values will get h_vmem=256M, while users with correctly configures virtual_free values (and >= 256M) will get v_hmem = 1.1*virtual_free. 
[15:45:21] <valhallasw>	 you can find it in /home/valhallasw/jsub
[15:45:59] <gerrit-wm>	 Change merged: coren; [labs/toollabs] (master) - https://gerrit.wikimedia.org/r/64575
[15:46:12] <Coren>	 valhallasw: Can you push this to gerrit, please?
[15:46:25] <valhallasw>	 Coren: I think so, yes
[15:47:57] <valhallasw>	 Coren: shouldnt jstart be a symlink to jsub?
[15:48:45] <Coren>	 It is, by design
[15:52:06] <valhallasw>	 Coren: that's currently not the case in git
[15:52:46] <Coren>	 valhallasw: That's done by the package.
[15:55:08] <gerrit-wm>	 New patchset: Merlijn van Deen; "Added .gitreview" [labs/toollabs] (master) - https://gerrit.wikimedia.org/r/64585
[15:56:17] <valhallasw>	 Coren: ...ehm, ok? there are now two distinct files in jobutils/usr/local/bin, that have different contents
[15:56:18] <Coren>	 Ah, yes, I meant adding that .gitreview and keep forgetting.  :-)
[15:57:09] <Coren>	 valhallasw: Can this wait until after Amsterdam?  The next 48h are... hectic for me.  :-)
[15:57:19] <valhallasw>	 sure
[16:01:44] <gerrit-wm>	 New patchset: Merlijn van Deen; "Add (parts of) qcronsub functionality into jsub:   - To ease transition from the toolserver, set h_vmem=1.1*virtual_free     if h_vmem<virtual_free and virtual_free is given;   - set -once if started as qcronsub" [labs/toollabs] (master) - https://gerrit.wikimedia.org/r/64586
[16:02:06] <Coren>	 As a reminder:  There will be a Labs cluster-wide reboot today that will impact Tool Labs.  Grid jobs will be interrupted (then restarted if started as continuous), and interactive sessions will be broken.
[16:27:29] * anomie  needs admin permission on en.wikipedia.beta.wmflabs.org. User Anomie.

[16:42:07] <petan|bla>	 Coren why
[16:42:33] <Coren>	 petan|bla: Why what?
[16:42:37] <petan|bla>	 reboot
[16:42:59] <jeremyb>	 Coren: rules link in ssh banner seems to be non-existant page
[16:43:15] * jeremyb  runs away

[16:43:16] <Coren>	 jeremyb: They're being written.  :-)
[16:43:49] <jeremyb>	 i mean redlink
[16:44:58] <anomie>	 chrismcmahon: Can I get admin on en.wikipedia.beta.wmflabs.org, for testing CodeEditor in the MediaWiki namespace?
[16:49:36] <greenrosetta>	 what is the rev_user field for in the revision table?
[16:51:16] <Coren>	 The user who made the edit.
[16:53:07] <Coren>	 petan|bla: Security kernel upgrade.  Doing the instances too, but the actual cluster servers need it.
[16:53:21] <Coren>	 petan|bla: Ryan announced it on labs-l last Friday.
[16:53:37] <greenrosetta>	 is there a table where I can get the textual representation of page.page_namespace ?
[16:54:19] <scfc_de>	 greenrosetta: No, it's set in PHP.  You can query the API.
[16:55:17] <greenrosetta>	 that seems odd
[16:56:13] <greenrosetta>	 I haven't used the API yet, is there a primer for getting started on lbas?
[16:58:06] <scfc_de>	 lbwhat?
[16:58:14] <greenrosetta>	 oops... "labs"
[16:58:37] <greenrosetta>	 using the API with labs that is
[17:00:00] <scfc_de>	 The API is a web interface of the wiki in question.  Try http://de.wikipedia.org/w/api.php?action=query&meta=siteinfo&siprop=namespaces for the namespaces for the German wikipedia.  Complete documentation is at https://www.mediawiki.org/wiki/API.  https://www.mediawiki.org/wiki/API:Client_code has some libraries.
[17:01:29] <greenrosetta>	 oh, so each wiki has its own namespace set (i guess tha makes sense)
[17:01:33] <Krenair>	 greenrosetta, rev_user is the user's id
[17:02:08] <greenrosetta>	 so if I want the namespace for the enwiki its simply this: http://en.wikipedia.org/w/api.php?action=query&meta=siteinfo&siprop=namespaces
[17:02:15] <Krenair>	 greenrosetta, you may be interested in https://bugzilla.wikimedia.org/show_bug.cgi?id=48625
[17:02:46] <Krenair>	 (obviously this is a feature request, does not exist yet, and you will have to use the API at the moment)
[17:03:50] <greenrosetta>	 are the namespace id's the same across all wikis for article, talk, etc.?
[17:04:16] <Krenair>	 For the default mediawiki ones? Should be.
[17:04:16] <greenrosetta>	 it looks that way, but  I guess it doesnt have to be
[17:04:21] <Krenair>	 Emphasis on should
[17:04:34] <Krenair>	 Historical wikis might be different etc.
[17:07:45] <Betacommand>	 greenrosetta: what did you need?
[17:07:54] <greenrosetta>	 nvm.. figured it out on my own
[17:08:21] <petan|bla>	 Coren aha what is reason
[17:08:27] <petan|bla>	 so whole labs will be down?
[17:10:54] <scfc_de>	 petan|bla: Didn't you discuss the reboots on labs-l or wikitech-l with Ryan?! :-)
[17:11:08] <petan|bla>	 no
[17:11:19] <petan|bla>	 I discussed reboots of production with some guy
[17:12:45] <scfc_de>	 Ah, yeah, Greg Grossmeier.  I think I smell a connection, though :-).
[17:13:39] <Krenair>	 Wasn't this the linux security fix?
[17:14:05] <Krenair>	 oh yes: <Coren> petan|bla: Security kernel upgrade.  Doing the instances too, but the actual cluster servers need it.
[17:14:19] <petan|bla>	 it's not really a big security issue from my POV, so I am surprised they are patching the host systems
[17:14:54] <petan|bla>	 Coren nvm I see you already said that just didn't notice, thans Krenair
[17:15:34] <petan|bla>	 basically - it seems to me that only systems where untrusted users have shell access are affected, and tbh similar issues are being found quite often
[17:16:23] <scfc_de>	 petan|bla: Your link http://www.h-online.com/open/news/item/Exploit-for-local-Linux-kernel-bug-in-circulation-Update-1863892.html says: "after an exploit was released that proves that normal, logged-in users can gain root access this way."  -- I definitely wouldn't want some jerk to mess up Coren's work of the past weeks :-).
[17:16:50] <petan|bla>	 scfc_de I am not talking about guests but hosts
[17:17:00] <petan|bla>	 I understand why instances needs to be rebooted
[17:17:21] <petan|bla>	 but basically only Ryan and trusted folks can access the hosts
[17:18:19] <scfc_de>	 I think it's *much* safer to push the reboot button once instead of maintaining a list of dangers: "If someone ever gets unprivileged access to the host, we're fucked."
[17:19:57] <Coren>	 scfc_de: More to the point, a privilege escalation exploit transforms any "can get shell" exploit in /other/ software into "can get root"
[17:28:17] <scfc_de>	 Coren: The ETA for alpha-public of the replicas, will it be more a matter of hours, or tomorrow, or in the course of the week?
[17:30:48] <Coren>	 scfc_de: Hours, likely, but probably as late as this evening.
[17:31:39] <Coren>	 (GMT+5)
[17:31:51] <Coren>	 (GMT-4)*
[17:32:00] <Coren>	 Not sure where +5 came from there.  :-)
[17:32:25] <scfc_de>	 Probably wishful thinking :-).
[17:34:14] <Krinkle>	 petan: What's up?
[17:39:47] <gerrit-wm>	 New patchset: Merlijn van Deen; "jstart is now a symlink to jsub" [labs/toollabs] (master) - https://gerrit.wikimedia.org/r/64596
[17:41:09] <valhallasw>	 hm, is gerrit reviewer bot broken? :/
[17:41:49] * valhallasw  checks

[18:30:35] <hashar>	 !log deployment-prep Added Krenair to the project
[18:30:38] <labs-morebots>	 Logged the message, Master
[18:58:04] <Coren>	 *** As a reminder:  There will be a Labs cluster-wide reboot in 1H that will impact Tool Labs.  Grid jobs will be interrupted (then restarted if started as continuous), and interactive sessions will be broken.
[19:00:48] <Ryan_Lane>	 (It will affect all of labs, in fact ;) )
[19:02:38] <Coren>	 Ryan_Lane: Sure, but Tool Labs are my favorite.  :-P
[19:04:30] <Ryan_Lane>	 heh
[19:04:34] <Ryan_Lane>	 hm. nova has a stop command
[19:04:45] <Ryan_Lane>	 but it looks like it just does a virsh destroy :D
[19:04:49] <Ryan_Lane>	 that's not helpful
[19:05:01] <Ryan_Lane>	 I guess I'll be nice and run a shutdown -h now via salt to all the instances
[19:05:27] <Ryan_Lane>	 ok. anyway. food. for reals now
[19:22:41] <wm-bot>	 Warning: There is 1 user waiting for shell: Taha (waiting 0 minutes) 
[19:36:10] <wm-bot>	 Warning: There is 1 user waiting for shell: Taha (waiting 13 minutes) 
[19:46:56] <Coren>	 *** As a reminder:  There will be a Labs cluster-wide reboot in 15m that will impact Tool Labs.  Grid jobs will be interrupted (then restarted if started as continuous), and interactive sessions will be broken.
[19:47:48] <scfc_de>	 The credentials for replicated databases and tools-db are different.  Is there a way to specify both in ~/.my.cnf?  (Not seeing anything obvious on http://dev.mysql.com/doc/refman/5.0/en/option-files.html.)
[19:49:44] <wm-bot>	 Warning: There is 1 user waiting for access to tools project: Pouyan (waiting 10 minutes) 
[19:50:45] <Coren>	 scfc_de: Not that I know of, but you can use the replicated one with --defaults-file=replica.my.cnf
[19:50:59] <Coren>	 scfc_de: That'll obviously be the default in that commandline utility you filed a bz for.  :-)
[19:51:34] <Coren>	 That said, I'm probably going to switch the credentials around for the local DB too to simplify things.
[19:51:47] <Coren>	 In my Copious Free Time™
[19:52:53] <scfc_de>	 Coren: :-)  Unification of credentials would be *very* useful.  I assume there are some scripts/libraries that read ~/.my.cnf and anything else causes a headache.
[19:53:28] <scfc_de>	 (Even better would of course be per-host settings in mysql.)
[19:53:37] <Coren>	 scfc_de: That's beyond my powers.  :-)
[19:54:09] <scfc_de>	 It's written in C, I believe :-).
[19:55:21] <Coren>	 Heh.  Sure, I'll submit a patch to the mariadb peeps once I'm done with my tool labs tasklist.  How does 3Q 2014 sound? :-)
[19:57:56] <scfc_de>	 :-)
[20:00:15] <Krenair>	 ooh, I have a replica.my.cnf
[20:00:37] <Coren>	 Krenair: Magic!
[20:00:39] <Krenair>	 Does that mean we can connect and run queries on some dbs now?
[20:00:58] <Coren>	 Krenair: Soon.  Missing DNS and some network level holes to connect through.
[20:03:13] <wm-bot>	 Warning: There is 1 user waiting for access to tools project: Pouyan (waiting 24 minutes) 
[20:03:21] <Ryan_Lane>	 ok. time for maintenance
[20:03:27] <Ryan_Lane>	 everything is going down....