[02:09:34] * jeremyb  wonders if Ryan_Lane's around

[02:09:41] <jeremyb>	 i don't see a labs-storage-wm here
[02:09:52] <jeremyb>	 is it supposed to be?
[02:10:11] <jeremyb>	 (i see it in a manifest but haven't checked where that manifest is used)
[02:11:54] <Ryan_Lane>	 yeah. supposed to be
[02:12:01] <Ryan_Lane>	 likely just needs to reconnect
[02:21:06] <jeremyb>	 k
[02:21:14] * jeremyb  is fighting with git and packet loss

[02:21:20] <jeremyb>	 at least i have a local clone ;)
[03:27:46] <jeremyb>	 Ryan_Lane: still there?
[03:27:50] <jeremyb>	 >                 return [os.path.join(hookconfig.logdir, log) for log in logs]
[03:28:11] <jeremyb>	 is that understandable? should be broken down somehow?
[09:35:50] <labs-home-wm>	 09/05/2012 - 09:35:50 - Creating a home directory for ajentzsch at /export/keys/ajentzsch
[09:40:51] <labs-home-wm>	 09/05/2012 - 09:40:51 - Updating keys for ajentzsch at /export/keys/ajentzsch
[09:41:24] <hashar>	 paravoid: do you have any idea how Gerrit handle authentication ? 
[09:41:50] <hashar>	 ajentzsch ( Anja Jentzsch ) has been created and I can find her in the LDAP 
[09:42:02] <hashar>	 but Gerrit does not find the account when I am trying to add her to a group :/
[09:46:15] <paravoid>	 she has to login once first iirc
[09:46:22] <paravoid>	 to gerrit
[09:46:38] <hashar>	 ahh
[09:47:10] <hashar>	 paravoid: thanks :)
[09:47:34] <hashar>	 so the chain is roughly:  paravoid --> hashar --> Abraham_WMDE1 > AnjaJ_WMDE
[09:47:35] <hashar>	 ;)
[09:48:05] <paravoid>	 heh
[09:48:22] <hashar>	 at least we are all sitting in Europe
[09:50:00] <Abraham_WMDE1>	 Hi guys
[09:50:05] <Abraham_WMDE1>	 *wave
[09:51:09] <paravoid>	 tell that to my broken circadian clock :)
[09:51:44] <Abraham_WMDE1>	 hashar: now Anja logged in to gerrit
[09:51:51] <Abraham_WMDE1>	 I think that was the problem
[09:51:53] <hashar>	 \O/
[09:52:00] <hashar>	 I have added her to the wikidata project
[09:52:06] <Abraham_WMDE1>	 awesome
[09:52:08] <Abraham_WMDE1>	 thank you!
[09:52:11] <hashar>	 AnjaJ_WMDE: welcome to the Wikidata team :-)))))))))))))
[09:52:58] <hashar>	 and I have finished my javascript in ant :)
[09:53:00] <hashar>	 \O/
[09:53:11] <paravoid>	 why do you all have _WMDE in your nicknames? :)
[09:53:27] <hashar>	 viral advertising of WMDE ? 
[09:53:52] <Platonides>	 I find that a bit silly, too
[09:53:56] <Platonides>	 but not their fault
[09:54:20] <Abraham_WMDE1>	 paravoid: that's policy
[09:55:14] <Abraham_WMDE1>	 again, thank you guys for your support!
[09:55:16] <Abraham_WMDE1>	 :-)
[09:55:35] <AnjaJ_WMDE>	 thanks, hashar !
[09:55:53] <labs-home-wm>	 09/05/2012 - 09:55:53 - Updating keys for ajentzsch at /export/keys/ajentzsch
[09:56:45] <hashar_WMF-fr>	 AnjaJ_WMDE: you are welcome
[09:57:10] <hashar_WMF-fr>	 AnjaJ_WMDE: if you get any issue with git / Gerrit  and your coworkers are too busy, let me know
[09:57:20] <hashar_WMF-fr>	 AnjaJ_WMDE: I don't really understand German unfortunately :(
[09:57:46] * Platonides  sets mode +b *_WM*

[09:57:59] <AnjaJ_WMDE>	 hashar_WMF-fr: thanks for that offer. I will get back to you if needed
[09:59:41] <hashar_WMF-fr>	 Platonides: I have just discoverd I can sneak javascript in ant build script :-))
[10:01:26] <Platonides>	 that looks even more awful :)
[10:02:03] <hashar_WMF-fr>	 http://dpaste.org/vfT0x/
[10:04:22] <Platonides>	 doesn't break work as an 'early return' ?
[10:04:28] <hashar_WMF-fr>	 ahh good idea
[10:05:10] <hashar_WMF-fr>	 Platonides: does not :(
[10:05:17] <hashar_WMF-fr>	 will find out one deay
[10:05:28] <hashar_WMF-fr>	 change is https://gerrit.wikimedia.org/r/22712 if you are curious
[10:05:47] <hashar_WMF-fr>	 I got an InjectExtensions.php script that load ext in order as it found them in extensions/ directory
[10:05:51] <hashar_WMF-fr>	 but that does not play nice
[10:06:01] <hashar_WMF-fr>	 since some extension can be loaded before its dependency
[10:06:02] <hashar_WMF-fr>	 :(
[10:13:30] <Platonides>	 do we have such extensions?
[10:13:57] <Platonides>	 <script language="javascript"> is space-indented
[10:14:04] <Platonides>	 while the rest of the file seems to be tab-indented
[11:02:42] <hashar_WMF-fr>	 arf
[11:02:44] <hashar_WMF-fr>	 wrong copy paste
[11:20:07] <hashar>	 grmblbl
[11:20:08] <hashar>	 http://stackoverflow.com/questions/12280295
[11:20:12] <hashar>	 does not work
[11:20:33] <hashar>	 script does not play well with antcall :D
[11:24:58] <hashar>	 lunnch time
[13:23:39] <hashar>	 I have fixed it ahaha
[13:23:42] <hashar>	 \O/
[13:23:45] <hashar>	 wholoualalaouuaou
[13:23:46] <paravoid>	 fixed what?
[13:23:52] <hashar>	 http://stackoverflow.com/questions/12280295
[13:23:57] <hashar>	 got an issue with ant and javascript :)
[13:24:42] <hashar>	 was mostly self cheering
[13:26:39] <hashar>	 I hate it when it takes me like 3 hours to figure out a problem
[13:40:25] <hashar>	 a oneline fix https://gerrit.wikimedia.org/r/#/c/22720/2/jobs/_shared/build.xml,unified
[14:13:33] <TomDaley>	 hashar: I see your oneline fix and raise you a 28 line feature: https://gerrit-review.googlesource.com/#/c/37711/ :)
[14:15:33] <TomDaley>	 37711 will make replication much easier for me to configure :)
[14:15:41] <TomDaley>	 No need for that silly authGroup nonsense.
[14:22:45] <hashar>	 TomDaley: need tests!!!!!!!!! ;-]
[14:23:03] <hashar>	 I like the feature
[14:23:26] <hashar>	 + the plugins could probably be released more often than the gerrit app
[14:23:51] <TomDaley>	 Yeah, potentially. That's part of the draw :)
[14:24:48] <hashar>	 TomDaley: weren't you working on having Gerrit nightly build by our Jenkins installation ?
[14:28:24] <TomDaley>	 We're gonna do something with it, not entirely sure if it'll be jenkins-based or something else. Gonna talk with Ryan about it next week.
[14:29:05] <TomDaley>	 What I really want is an easy way to build the .debs.
[14:29:52] <hashar>	 checkout the repo
[14:29:59] <hashar>	 run the magic command like debuilder
[14:30:01] <hashar>	 copy to nightly 
[14:30:03] <hashar>	 done ?
[14:30:22] <hashar>	 I have no idea how to properly run debuilder
[14:30:24] <hashar>	 paravoid would know
[14:30:34] <TomDaley>	 Well it doesn't have the control files and such
[14:30:41] <hashar>	 each time I want to build a debian package I end up rereading the debian doc and getting lost in it
[14:30:42] <TomDaley>	 (getting that upstreamed would be nice, if possible)
[14:30:47] <TomDaley>	 Me too!
[14:32:20] <TomDaley>	 There's a guy working on packaging gerrit for debian. I don't plan on using the actual package (we want something closer to master), but the work for it (control files, dependencies being packaged rather than maven things) would be super-useful to me.
[14:32:23] <TomDaley>	 I need to reach out to him.
[14:36:42] <hashar>	 TomDaley: I am pretty sure you could reuse the control file and just make them point to master
[14:36:45] <hashar>	 instead of some git tag
[14:37:03] <TomDaley>	 Right, that's what I'm thinking.
[14:37:04] <hashar>	 http://wiki.debian.org/PackagingWithGit
[14:37:49] <hashar>	 I am off for this afternoon will be back later on hopefully
[14:37:57] <hashar>	 ohh
[14:37:58] <hashar>	 last thing
[14:38:03] <hashar>	 we definitely need a gerrit irc plugin :)
[14:41:51] <TomDaley>	 I e-mailed repo-discuss about this :)
[14:42:03] <hashar>	 \O/
[14:42:36] <hashar>	 see you later!
[17:09:21] <matanya>	 bug triage now in #wikimedia-dev
[17:53:58] <TomDaley>	 Ryan_Lane: So I'm pretty sure the random timeouts people have been getting on gerrit are because of the repack issue.
[17:54:22] <TomDaley>	 Repo gets way too big, gerrit implodes.
[17:54:30] <TomDaley>	 http://p.defau.lt/?r1gMue0Z1Oktk_z6tJZJxg - case in point
[17:55:18] <Ryan_Lane>	 likely
[17:55:39] <TomDaley>	 However, it's been pointed out that running a repack on a live server can be a Bad Idea.
[17:55:45] <Ryan_Lane>	 oh?
[17:55:50] <TomDaley>	 Something about upload-pack expecting some object that got moved or somesuch.
[17:55:58] <Ryan_Lane>	 oh
[17:55:59] <TomDaley>	 I don't really get it. So I'm afraid it'll cause downtime :\
[17:56:21] <TomDaley>	 I'm going to dig into this today. We need a way to deal with this w/o causing downtime.
[17:57:09] <Ryan_Lane>	 yeah
[17:58:57] <TomDaley>	 A `git gc` is probably sane, but since it won't prune immediately there's the possibility of it making the repo even larger.
[17:59:22] <TomDaley>	 So I figured `git gc --prune=now` but Martin from Qualcomm said "You should probably never run that on a live Gerrit server since you could cause new uploads to fail with it."
[19:24:47] <TomDaley>	 Ryan_Lane: So turns out, repack is fine. gc is fine. It's prune=now that's evil.
[19:25:03] <TomDaley>	 Just need to come up with a sane way to iterate over all the repos and we can cron this up.
[19:25:11] <TomDaley>	 (Weekly would be fine)
[19:25:47] <Ryan_Lane>	 find /var/lib/gerrit2/revew_site - name "*.git" -exec git repack {} \;
[19:26:18] <TomDaley>	 You want to do all of them every time?
[19:26:24] <TomDaley>	 Core will typically take an ~hour
[19:26:33] <Ryan_Lane>	 a cron
[19:26:35] <Ryan_Lane>	 run every minute
[19:26:57] <Ryan_Lane>	 ;)
[19:27:49] <TomDaley>	 Tooooo often can run into its own problems as well. I was given an example of daily :)
[19:27:55] <TomDaley>	 I figured weekly would suffice for us.
[19:28:21] <TomDaley>	 Also, that find is way too broad, you don't wanna list all the files in the subdirectories :p
[19:29:24] <TomDaley>	 Want -type d
[19:31:20] <Ryan_Lane>	 EVERY MINUTE
[19:31:46] <TomDaley>	 YOU CRAZY
[19:31:48] <Ryan_Lane>	 heh
[19:31:53] <Ryan_Lane>	 it's fun trolling you
[19:32:02] <Ryan_Lane>	 come up with a decent cron and push it into the puppet repo
[19:32:09] <Ryan_Lane>	 I'll review it
[19:32:15] <Ryan_Lane>	 and put it on high priority
[19:32:30] <RoanKattouw>	 Ah, WTF
[19:32:33] <RoanKattouw>	 Getting 503s from Gerrit?
[19:33:08] <RoanKattouw>	 Back now
[19:33:16] <Ryan_Lane>	 stop breaking things roan
[19:33:26] <TomDaley>	 503s are from huge repos.
[19:33:33] <TomDaley>	 I'm pretty sure.
[19:33:51] <RoanKattouw>	 I was refreshing the change view for a mediawiki/core change
[19:33:53] <Ryan_Lane>	 TomDaley: github wouldn't 503
[19:33:54] <RoanKattouw>	 Waiting for Jenkins
[19:34:16] <Ryan_Lane>	 TomDaley: neither would phabricator
[19:34:19] <Ryan_Lane>	 :)
[19:34:21] <RoanKattouw>	 The build succeeded but Jenkins didn't post back for some reason
[19:47:46] <Ryan_Lane>	 TomDaley: when are we upgrading to 2.5?
[19:47:54] <Ryan_Lane>	 RoanKattouw_away: I have a question for you
[19:47:56] <TomDaley>	 When it at least hits rc1.
[19:47:59] <Ryan_Lane>	 TomDaley: or maybe you
[19:48:05] <TomDaley>	 And/or we talk about a build system next week.
[19:48:14] <Ryan_Lane>	 it's a mediawiki group/rights question
[19:48:18] <Ryan_Lane>	 https://gerrit.wikimedia.org/r/#/c/22692/
[19:48:29] <Ryan_Lane>	 https://gerrit.wikimedia.org/r/#/c/22693/
[19:48:32] <Ryan_Lane>	 those two changes....
[19:48:41] <Ryan_Lane>	 it doesn't do exactly what I want
[19:49:27] <Ryan_Lane>	 what I want is for OpenStackManager to know that OATH is enabled for a user
[19:49:43] <Ryan_Lane>	 so, I add the user to a group that has a right
[19:49:59] <Ryan_Lane>	 but, i require it for users that have userrights right
[19:50:17] <Ryan_Lane>	 people with userrights can just add/remove themselves from the oath group
[19:50:28] <Ryan_Lane>	 is there some way to disallow that?
[19:50:35] <Ryan_Lane>	 or is there a better way to do it?
[19:56:37] <Ryan_Lane>	 talked with roan. I think I got it worked out
[19:58:50] <TomDaley>	 asjoicaojassaoianc
[19:58:53] <TomDaley>	 Stupid find
[19:59:09] <Ryan_Lane>	 that does indeed look like a stupid find
[19:59:09] <Ryan_Lane>	 LD
[19:59:11] <Ryan_Lane>	 err
[19:59:12] <Ryan_Lane>	 :D
[19:59:12] <TomDaley>	 `find /var/lib/gerrit2/review_site/git -type d -name "*.git"` -execdir git gc {} \;` isn't working
[19:59:30] <Ryan_Lane>	 of course it doesnt
[19:59:42] <Ryan_Lane>	 you are using backquotes
[19:59:55] <Ryan_Lane>	 and why are you using execdir?
[19:59:56] <TomDaley>	 Oh, that was just from copy+pasting it elsewhere.
[20:00:10] <TomDaley>	 git gc needs to be run from within the directory, not passed as a parameter to git gc.
[20:00:17] <Ryan_Lane>	 ah
[20:01:17] <TomDaley>	 find /var/lib/gerrit2/review_site/git -type d -name "*.git" -execdir git gc --quiet \;
[20:01:33] <TomDaley>	 ^ Was exactly what I pasted, but I got a spew of "fatal: Not a git repository (or any of the parent directories): .git"
[20:02:36] <TomDaley>	 I could do it in like 5 lines in a bash script, but I'm convinced this can be one-lined.
[20:03:31] <mutante>	 TomDaley: are you missing the {} part now?
[20:03:42] <TomDaley>	 Tried it with and without
[20:03:44] <TomDaley>	 Same result
[20:04:45] <Ryan_Lane>	 TomDaley: you could use pushd popd
[20:04:50] <Ryan_Lane>	 or cd; command; cd -
[20:05:08] <Ryan_Lane>	 (if you didn't know about cd -, it's really useful)
[20:05:35] <TomDaley>	 I thought you can't use cd in find -exec?
[20:05:51] <Ryan_Lane>	 heh
[20:05:53] <Ryan_Lane>	 probably can't
[20:06:05] <TomDaley>	 Google said no when I tried about 10 mins ago ;-)
[20:06:21] <mutante>	 "any given invocation of command will only list files that exist in the same subdirectory.
[20:06:58] <TomDaley>	 http://p.defau.lt/?9rgFIpP2z0eoO4rKxUqNDA - i mean we could just add a file to the gerrit box and execute that ;-)
[20:07:26] <Ryan_Lane>	 TomDaley: that's a one liner too
[20:07:51] <Ryan_Lane>	 for d in `find /var/lib/gerrit2/review_site/git -type d -name "*.git"`; do cd $d; git gc; done
[20:08:32] <TomDaley>	 Ah, yes
[20:08:34] <TomDaley>	 That'll work
[20:08:54] <Ryan_Lane>	 that should work in a cron entry. it won't work in a puppet exec
[20:08:55] <Ryan_Lane>	 fucking puppet
[20:10:34] <mutante>	 is git-gc = git gc ?
[20:12:16] <TomDaley>	 Yes.
[20:12:44] <Ryan_Lane>	 it would be nice if gerrit did periodic gc
[20:12:49] <TomDaley>	 Ryan_Lane: https://gerrit.wikimedia.org/r/#/c/22824/
[20:12:56] <TomDaley>	 It's jgit's fault.
[20:13:27] <TomDaley>	 gerrit would have to shell out to normal git to do it.
[20:13:44] <Ryan_Lane>	 failed lint ;)
[20:14:06] <mutante>	 just reading about the --aggressive option. it says it "only needs to be used occasionally, every few hundred changesets or so"..
[20:14:20] <TomDaley>	 Ryan_Lane: fixed in PS2.
[20:14:35] <Ryan_Lane>	 doubt it
[20:14:42] <TomDaley>	 mutante: --aggressive is really just shorthand for "do a repack and drop every delta you already have and recompute"
[20:14:47] <Ryan_Lane>	 you're using double quotes inside of double quotes
[20:14:48] <Damianz>	 You put a "" in a ""
[20:14:53] <TomDaley>	 Oh.
[20:14:54] <TomDaley>	 That.
[20:15:28] <TomDaley>	 mutante: `git gc --aggressive` is really just `git repack -a -d -f --depth=250 --window=250` more or less.
[20:16:00] <mutante>	 aha
[20:16:09] <TomDaley>	 -a on git-repack carries some weird warnings about public repos
[20:16:19] <TomDaley>	 (more specifically, it suggests being used on "private" repos)
[20:16:40] <TomDaley>	 mutante: http://metalinguist.wordpress.com/2007/12/06/the-woes-of-git-gc-aggressive-and-how-git-deltas-work/ - good reading on the topic :)
[20:16:53] <mutante>	 and --auto ? " checks whether any housekeeping is required; if not, it exits without performing any work."
[20:17:22] <mutante>	 thanks TomDaley
[20:17:38] <TomDaley>	 In my experience, --auto isn't really all that smart about it.
[20:17:43] <mutante>	 i see, "a big no-no".ok:)
[20:17:49] <mutante>	 about --aggressive
[20:19:00] <TomDaley>	 Ryan_Lane: PS3.
[20:20:48] <Ryan_Lane>	 merged
[20:20:54] <Ryan_Lane>	 and force running on manganese
[20:21:02] <Ryan_Lane>	 I'm also going to run the command
[20:21:55] <Ryan_Lane>	 doesn't work
[20:22:11] <Ryan_Lane>	 ah
[20:22:25] <Ryan_Lane>	 you need to wrap the command in single quotes, rather than double
[20:22:39] <Ryan_Lane>	 TomDaley: $d is being evaluated by puppet
[20:23:41] <Ryan_Lane>	 I'm getting errors like: error: packfile ./objects/pack/pack-c82c47b6bf990a2e33ef892263820742d8898611.pack cannot be accessed
[20:23:51] <Ryan_Lane>	 and: error: refs/tags/1.17.0rc1 does not point to a valid object!
[20:24:58] <Ryan_Lane>	 and: fatal: object 9416c7950f23b51069806cf1c2855086da975232 is corrupted
[20:28:12] <TomDaley>	 Oh man.
[20:30:14] <Ryan_Lane>	 ?
[20:30:20] <Ryan_Lane>	 it's not happening a lot
[20:30:21] <Ryan_Lane>	 just some
[20:30:31] <TomDaley>	 I want to figure out why though :\
[20:30:38] <Ryan_Lane>	 yeah
[20:30:46] <TomDaley>	 Could you abort, and run a `git fsck` on mediawiki core?
[20:31:01] <TomDaley>	 It might take awhile.
[20:31:19] <TomDaley>	 `git fsck --full` even
[20:31:54] <wm-bot>	 Change on 12mediawiki a page Developer access was modified, changed by !Silent link https://www.mediawiki.org/w/index.php?diff=580433 edit summary: /* User:!Silent */ Changed
[20:43:36] <wm-bot>	 Change on 12mediawiki a page Developer access was modified, changed by Jeremyb link https://www.mediawiki.org/w/index.php?diff=580442 edit summary: /* User:!Silent */ done
[21:59:26] <Ryan_Lane>	 who wants a nicer labsconsole interface? who wants to do a few fairly minor reviews for me? :) 22690, 22692, and 22833
[21:59:39] <Ryan_Lane>	 I'll just keep spamming channels until someone does it. heh
[22:00:41] <Damianz>	 Only if we can have a pink unicorn :D
[22:06:37] <Damianz>	 Urgh, I hate the way php lets you use vars in strings... <3 python's sprintf style handling
[22:07:21] <Ryan_Lane>	 yeah. agreed
[22:07:29] <Ryan_Lane>	 though isn't that deprecated in python?
[22:10:36] <Damianz>	 "Hello %s" % name? Don't think so, unless you mean in 3 where I think it's like print("Hello %s", name) or something *shrug*.
[22:10:45] <Damianz>	 22692 looked evil, though it actually wasn't that bad.
[22:11:14] * Damianz  is still trying to get out the habbit of using print for debugging because the logging module is freaking awesome

[22:14:58] <Ryan_Lane>	 Damianz: ok. check out the labsconsole interface now
[22:15:03] <Ryan_Lane>	 and force-refresh the page
[22:15:07] <Ryan_Lane>	 I've changed the sidebar
[22:15:32] <Ryan_Lane>	 I've broken the actions into sections, and the sections are only shown if you have rights to use them
[22:15:38] <Damianz>	 Oooh that's shiny
[22:15:52] <Ryan_Lane>	 I wrote this extension ages ago.
[22:16:16] <Ryan_Lane>	 better, eh? :)
[22:16:28] <Damianz>	 That's actually a lot nicer to use.
[22:17:05] <Damianz>	 https://labsconsole.wikimedia.org/wiki/Server_Admin_Log that actually looks like hell
[22:17:18] <Ryan_Lane>	 yeah
[22:17:19] <Ryan_Lane>	 it does
[22:17:41] <Ryan_Lane>	 I'd really like to break it down further
[22:17:53] <Ryan_Lane>	 so that each project only shows the past x days
[22:18:11] <Ryan_Lane>	 that will require changes to the bot and to the semantic functions, though
[22:18:14] <Damianz>	 Like Special:Recent or w/e it is.
[22:18:30] <Ryan_Lane>	 well, it would really be ideal for this to be an extension
[22:19:16] <Damianz>	 Imo it would be nice to get events from openstack in there as well like 'damian rebooted <instance name>' or such.
[22:19:19] <Ryan_Lane>	 hm. it would be better if anons didn't see most of the actions on the sidebar too
[22:19:29] <Ryan_Lane>	 yeah
[22:19:33] <Ryan_Lane>	 in the SAL
[22:19:37] <Ryan_Lane>	 or the recent changes?
[22:19:50] <Ryan_Lane>	 adding it to recent changes would be doable
[22:19:58] <Ryan_Lane>	 it's harder to add to the SAL
[22:19:58] <Damianz>	 SAL, though machine state in recent changes could be interesting
[22:20:42] <Damianz>	 Considering the amount we abuse pages (and thus make it easy to use recent) we could probably do some weird transcluding things for the project pages hmm, or that could look horrid
[22:21:46] <Ryan_Lane>	 thanks for the reviews, btw :)
[22:22:46] <Damianz>	 Btw is the host entry being blank due to the semantic search being wrong now or just a field we can't/don't update until such time there's a nice triggers system in openstack?
[22:23:40] <Damianz>	 &lt;openstackmanager-novakey-key&gt; < heh
[22:24:35] <Ryan_Lane>	 host entry is blank?
[22:24:45] <Ryan_Lane>	 where do you see that message?
[22:24:54] <Damianz>	 On https://labsconsole.wikimedia.org/wiki/Special:Ask/-5B-5BResource-20Type::instance-5D-5D/-3FInstance-20Name/-3FInstance-20Type/-3FProject/-3FImage-20Id/-3FFQDN/-3FLaunch-20Time/-3FPuppet-20Class/-3FModification-20date/-3FInstance-20Host/-3FNumber-20of-20CPUs/-3FRAM-20Size/-3FAmount-20of-20Storage/searchlabel%3Dinstances/offset%3D0 the field is blank
[22:25:00] <Damianz>	 The above is on the manage your ssh keys page
[22:25:08] <Ryan_Lane>	 uuuuggghhh
[22:25:16] <Ryan_Lane>	 how did that message disappear? :(
[22:25:26] <Damianz>	 Dunno, it wasn't removed in the diff
[22:25:41] <Ryan_Lane>	 I was complaining about this just today in the i18n channel
[22:26:24] <Damianz>	 Well you know, localization isn't important for wikipedia :D
[22:26:28] <Ryan_Lane>	 -_-
[22:27:44] <mutante>	 ah nice, i don't want to worry about localization of planet any longer :p.. just 3 words that are in English ..:p
[22:27:58] <Ryan_Lane>	 I just removed the section :)
[22:28:01] <Ryan_Lane>	 no need for that
[22:29:09] <Damianz>	 I swear my chrome hates rendering mediawiki...
[22:29:20] <Damianz>	 Randomly just loose table cells that re-appear on resizing -.-
[22:29:27] <Ryan_Lane>	 hm
[22:29:34] <Ryan_Lane>	 I don't like the sidebar item being called "openstack"
[22:29:41] <Ryan_Lane>	 it doesn't really make much sense
[22:30:11] <Ryan_Lane>	 3 of those items should only be shown to logged in users, too
[22:30:17] <Damianz>	 It would make more sense to be called labs probably, seen as the avg user (who it's really aimed at) doesn't care what's running on the backend
[22:31:33] <Ryan_Lane>	 hm
[22:31:36] <Ryan_Lane>	 * doesn't work as a group
[22:31:57] <Ryan_Lane>	 ah
[22:31:59] <Ryan_Lane>	 yeah it does
[22:32:09] <Ryan_Lane>	 stupid case sensitivity
[22:32:48] <Damianz>	 Hmm, I kinda hate that all the tabs are either named 'Special page' or 'Nova recourse' but then I guess we don't really define project vs instance so the wiki doesn't know what the pages are.
[22:33:11] <Ryan_Lane>	 there we go
[22:33:12] <Ryan_Lane>	 that's cleaner
[22:34:04] <Ryan_Lane>	 hm
[22:34:06] <Ryan_Lane>	 * isn't right
[22:34:08] <Damianz>	 Should I still be able to see 'Labs Users' logged out?
[22:34:09] <Ryan_Lane>	 I should likely use user
[22:34:47] <Ryan_Lane>	 that did it
[22:35:19] <Ryan_Lane>	 I should likely add a "request shell" link
[22:35:20] <Ryan_Lane>	 hm
[22:35:32] <Ryan_Lane>	 that doesn't really need to be in the sidebar, eh?
[22:35:51] <Ryan_Lane>	 nah
[22:36:37] <Damianz>	 I'd rather put a link there like 'getting started' or such, though that could also just be on the front page. Try and construct some flow of docs that prevent repetition of annoyingly simple questions. *shrug*
[22:36:47] <Ryan_Lane>	 yeah
[22:37:10] <Ryan_Lane>	 "Contents" in help doesn't really draw people's attention
[22:38:52] <Ryan_Lane>	 really I should only show "Manage Projects" to people in the "shell" group, but meh
[22:41:19] <Ryan_Lane>	 the main page could definitely use some work
[22:41:23] <Damianz>	 Heh is there such thing as a 'cloud admin' anymore? Or did they just drop project wide access and kept the role for creating tenants etc
[22:41:24] <Ryan_Lane>	 the unicorn logo is a little too big
[22:41:34] <Ryan_Lane>	 there's a wiki group called cloudadmin
[22:42:21] <Ryan_Lane>	 I should really remove  bureaucrat
[22:42:21] <Damianz>	 ah
[22:42:41] <Ryan_Lane>	 bureaucrat basically implies cloudadmin
[22:43:00] <Ryan_Lane>	 and no one should ever be granted that right, unless they are going to be a cloudadmin
[22:43:35] <Ryan_Lane>	 I can kill off accountcreators soon
[22:43:57] <Ryan_Lane>	 I can technically give people administrator rights now
[22:45:13] <Damianz>	 #progression
[22:45:29] <Ryan_Lane>	 yeah, it's nice, right? :D
[22:45:49] <Ryan_Lane>	 I'm about to take all of the policies out of draft as well
[22:46:25] <Damianz>	 I should probably get around to making my 1 login page with a link to the end user one :P
[22:46:49] <Damianz>	 And yeah progression is nice, but sometimes I could easily spend 20hours stabbing issues and get no where :P
[22:46:49] <Ryan_Lane>	 what do you mean?
[22:46:57] <Ryan_Lane>	 yeah
[22:48:21] <Damianz>	 https://www.mediawiki.org/wiki/Wikimedia_Labs/Agreement_to_disclosure_of_personally_identifiable_information is for end users, right? Though technically a username isn't personal info but I probably should stick it on there since I've got 10% to rewriting the damn thing to be logically.
[22:48:36] <Ryan_Lane>	 yep
[22:48:48] <Ryan_Lane>	 you should definitely stick it there
[22:49:49] <Ryan_Lane>	 well, seems I'm speaking at puppet conf
[22:50:04] <Damianz>	 puppet has a conference?
[22:50:08] <Ryan_Lane>	 yeah
[22:50:23] <Damianz>	 can you tell them how much using a dsl sucks :D
[22:50:32] <Ryan_Lane>	 heh
[22:50:36] <Ryan_Lane>	 I don't mind a dsl
[22:50:43] <Ryan_Lane>	 I like salt's quite a bit
[22:50:47] <Ryan_Lane>	 though it's really basically yaml
[22:50:49] <Damianz>	 I mind having to crazy inline ruby stuff :(
[22:51:00] <Damianz>	 I do *love* facter though, oh god that's useful
[22:51:01] <paravoid>	 you can write puppet manifests in ruby for quite a while
[22:51:15] <Ryan_Lane>	 then you have to use ruby, though
[22:51:17] <Ryan_Lane>	 :)
[22:52:05] <paravoid>	 well, yeah, but the chef people were the ones that were being vocal about the DSL
[22:52:08] <paravoid>	 so, there, fixed :-)
[22:52:23] <Ryan_Lane>	 salt lets you pick or implement your own dsl ;)
[22:52:40] <Ryan_Lane>	 and you can do it on a state by state basis
[22:52:49] <paravoid>	 what could possibly go wrong!
[22:52:51] <Damianz>	 I'm not sure if I like Cfengine or Puppet more config wise... Puppet is higher level but sometimes it's trying too hard and just fails to provide the abstraction enough so you end up doing crazy bash/exec things like you would in cfengine anyway.
[22:52:55] <Ryan_Lane>	 people generally use the jinja state engine, though
[22:53:05] <Ryan_Lane>	 and use python when they *must*
[22:53:07] <Damianz>	 Jinja is awesome
[22:53:10] <Ryan_Lane>	 generally the jinja one is perfect
[22:53:26] <paravoid>	 like a perl6 talk I was in a fosdem a few years back, they were saying how awesome perl6 is because you can now define your own syntax
[22:53:32] <Ryan_Lane>	 heh
[22:53:45] <Ryan_Lane>	 in perl that sounds like disaster
[22:53:52] <Ryan_Lane>	 in a dsl it's occasionally nice
[22:54:02] <Damianz>	 Is perl 6 even out? I thought they where still well stuck on 5.
[22:54:15] <Ryan_Lane>	 perl 6 is like duke nukem forever
[22:54:20] <Ryan_Lane>	 it *may* come out, it may not
[22:54:32] <Damianz>	 heh
[22:54:40] <Damianz>	 5.16.1 is stable IIRC, which works pretty well
[22:57:01] <Thehelpfulone>	 Ryan_Lane, https://labsconsole.wikimedia.org/wiki/Help:Contents#Requesting_Shell_Access who is "We" in that context? If someone wanted a request to be done more quickly, who would they go about poking?
[22:58:01] <Ryan_Lane>	 Thehelpfulone: lemme see
[22:58:36] <Damianz>	 Ryan_Lane: Thinking about opening up signups, could you setup an rc feed if you didn't already before we get spammed to hell and all the docs look wonky?
[22:59:09] <Ryan_Lane>	 I guess I should add a captcha extension
[22:59:20] <Ryan_Lane>	 and other anti-vandalism tools
[22:59:48] <Ryan_Lane>	 I haven't set one up yet
[23:00:44] <Ryan_Lane>	 Thehelpfulone: is it better now?
[23:01:13] <Ryan_Lane>	 there must be some way to internal link to that page
[23:02:34] <Ryan_Lane>	 yep. there is
[23:04:39] <Damianz>	 Is https://www.mediawiki.org/wiki/Wikimedia_Labs/Terms_of_use going to move or stay at that url?
[23:04:48] <Ryan_Lane>	 I think we should move it
[23:04:54] <Ryan_Lane>	 I need to ask legal where it should go
[23:05:13] <Damianz>	 It is in a rather obscure place
[23:05:36] <Damianz>	 Though we do call labs labsconsole and beta deployment-prep :P
[23:08:30] <Ryan_Lane>	 labsconsole will eventually be renamed to wikitech
[23:09:35] <Ryan_Lane>	 I want to keep all the logos and such there, though
[23:09:42] <Ryan_Lane>	 which makes things weird
[23:10:13] <Damianz>	 Hmm
[23:11:11] <Damianz>	 So for when someone eats the fiber a static copy offsite/mysql replication under a subdomain then merge not 100years out of date bits of wikitech into labs docs?
[23:12:10] <Damianz>	 http://themetapicture.com/warning-sign-done-right/ < I wonder if we can get that cc-by-sa licensed for use on labs docs :D
[23:13:07] <Thehelpfulone>	 Ryan_Lane, yeah that's better thanks - but I'm unclear on what "shell access in Labs" means - is that access to Bastion? I thought anyone could add anyone else to that?
[23:13:18] <Ryan_Lane>	 right, that's changing now, though
[23:13:35] <Ryan_Lane>	 now to be a member of any project, you need to be in the shell mediawiki group first
[23:14:10] <Ryan_Lane>	 otherwise we couldn't stop someone from creating a bunch of sockpuppets and re-granting themselves access
[23:14:30] <Ryan_Lane>	 this still doesn't fully solve that problem, but it makes it a lot harder
[23:14:33] <Damianz>	 Though the only benifit they'd get would be being able to proxy connections via bastion
[23:14:44] <Ryan_Lane>	 not really
[23:14:55] <Ryan_Lane>	 you can be added to projects without bastion
[23:15:13] <Ryan_Lane>	 if people have public ips and make their security groups permissive enough
[23:15:18] <Damianz>	 yeah, but unless some twonk was in a project then added themselves again by default they would be limited to bastion
[23:15:21] <Ryan_Lane>	 this stops them from being added to any project at all
[23:16:26] <Ryan_Lane>	 this unfortunately makes it one step more annoying to add someone
[23:16:40] <Ryan_Lane>	 but we're removing a step of creating the user, so it's not that bad
[23:17:16] <sumanah>	 Thanks for the code review, Damianz 
[23:17:20] <Damianz>	 Can you catch a hook of when they're added to that group and auto add them to bastion at that point? So we don't have the usual of being added to project but not bastion
[23:18:00] <Damianz>	 sumanah: np
[23:18:27] <Ryan_Lane>	 hm
[23:18:31] <Ryan_Lane>	 I probably could.
[23:18:46] <Thehelpfulone>	 Ryan_Lane, removing the step of creating the user? so how do they get an account?
[23:18:55] <Ryan_Lane>	 Thehelpfulone: self-registration
[23:19:00] <Ryan_Lane>	 we're coming out of beta
[23:19:22] <Thehelpfulone>	 ok, that's going to get messy with some spam accounts etc too
[23:19:28] <Ryan_Lane>	 yep
[23:19:34] * RoanKattouw  sings "We're out of beta and we're releasing on time"

[23:19:36] <Damianz>	 Hmm, actaully do people *need* to be in the project? Couldn't we just change the script that makes the homedir/dumps the keys to check against the shell group, if that's accessible in ldap. Though breaking from 'standard' for one projec tis probably bad
[23:19:46] <RoanKattouw>	 Hmm without the "and" I suppose
[23:19:49] <Thehelpfulone>	 have you made some more admins too Ryan_Lane?
[23:20:01] <Ryan_Lane>	 Damianz: I could do that, yeah
[23:20:06] <Ryan_Lane>	 Thehelpfulone: no. we can't
[23:20:18] <Ryan_Lane>	 anyone who is a cloud admin has the ability to add themselves to any project
[23:20:27] <sumanah>	 yay self-registration!
[23:20:41] <Thehelpfulone>	 you can't? for what reason?
[23:20:41] <Ryan_Lane>	 same with bureaucrat 
[23:20:43] <Thehelpfulone>	 hmm, looking at that cloudadmin list, I see a potential bottleneck - all of those people are likely to be very busy with other stuff too - how much time would they be able to spending doing shell access requests?
[23:21:06] <Ryan_Lane>	 the shell access requests are really quick and easy
[23:21:13] <Ryan_Lane>	 so, it shouldn't bottleneck badly
[23:21:17] <Ryan_Lane>	 no worse than account creation
[23:21:26] <Ryan_Lane>	 and we remove a bottleneck from gerrit by doing this
[23:21:28] <Damianz>	 If needed we could probably have some script with limited sudo in the same style as creations to enable access anyway
[23:21:44] <Ryan_Lane>	 I'd prefer people not have shell at all
[23:22:16] <Ryan_Lane>	 local escalation vulnerabilities make me worry about people signing up for immediate shell
[23:22:36] <Ryan_Lane>	 oh
[23:22:43] <Ryan_Lane>	 shell isn't an ldap group
[23:22:53] <Ryan_Lane>	 so we can't easily map that to bastion
[23:23:07] <Damianz>	 That sucks
[23:23:13] <Ryan_Lane>	 I can likely make a hook to modify bastion based on shell, though
[23:23:26] <Ryan_Lane>	 hm
[23:23:41] * Damianz  should actually revise his nagios code as ldap's probably changed since the other day

[23:23:56] <Ryan_Lane>	 maybe I could make project membership for other projects be based off access to bastion, and make adding/removing membership to bastion restricted
[23:23:58] <Ryan_Lane>	 that's harder, though
[23:24:55] <Ryan_Lane>	 roles in keystone are arbitrary now...
[23:24:57] <Damianz>	 Either (unless implimented in hooks) renders the code partly un-usable outside of labs which is bad in some ways
[23:25:04] <Ryan_Lane>	 I could base this on a keystone role
[23:25:28] <Ryan_Lane>	 then it could affect things on a project-per-project basis
[23:25:36] <Ryan_Lane>	 which I need for private data in labs, anyway
[23:25:40] <Ryan_Lane>	 lemme open a bug
[23:25:47] <Ryan_Lane>	 for now we'll continue on with using the shell group
[23:26:09] <Thehelpfulone>	 Ryan_Lane, shell access is quick and easy - so what criteria are you using to give it to people?
[23:26:27] <Ryan_Lane>	 Thehelpfulone: they write a justification in the shell justification field
[23:26:31] <Ryan_Lane>	 they need to have a reason to have shell
[23:26:32] <Damianz>	 Is the members being able to add other members an admin thing we've tacked on?
[23:26:51] <Ryan_Lane>	 Damianz: we manage project/role access directly
[23:27:01] <Ryan_Lane>	 completely bypassing keystone
[23:27:04] <Damianz>	 As an ldap admin?
[23:27:07] <Damianz>	 ah
[23:27:08] <Ryan_Lane>	 yes
[23:27:19] <Ryan_Lane>	 we directly modify ldap
[23:27:23] <Ryan_Lane>	 keystone reads that
[23:27:39] <Damianz>	 I was thinking if/when we can use nova from the command line for awesome it would be awesome but get around stuff like that... but as keystone proxies ldap in effect that's not an issue.
[23:28:01] <Ryan_Lane>	 yeah, can't get around that
[23:28:18] <Damianz>	 Though I dislike using any admin credentials for the user stuff, apart from where critical.
[23:28:18] <Ryan_Lane>	 cli issues are same as always. dns, puppet, wiki modification
[23:28:39] <Ryan_Lane>	 well, the interface uses your credentials to keystone
[23:28:48] <Damianz>	 mhm
[23:28:57] <Damianz>	 dns etc will be admin I assume though
[23:28:57] <Ryan_Lane>	 and cli tools would do the same
[23:29:00] <Damianz>	 since it's still outside
[23:29:06] <Ryan_Lane>	 dns is implemented in nova now
[23:29:12] <Damianz>	 oooh :D
[23:29:16] <Ryan_Lane>	 wiki editing is implemented as a plugin
[23:29:19] <Ryan_Lane>	 in nova
[23:29:28] <Ryan_Lane>	 and I think puppet is also, or will be
[23:29:47] <Damianz>	 Is that hidden on leappad somewhere under andrewbogott?
[23:29:51] <Ryan_Lane>	 I think we'll wait until folsom or grizzly to concentrate on cli, though
[23:30:04] <Ryan_Lane>	 dunno where the plugins are
[23:30:08] <Ryan_Lane>	 dns is in essex
[23:30:13] <Ryan_Lane>	 we'll switch to that soon enough
[23:30:31] <Ryan_Lane>	 we're still directly editing ldap for that right now
[23:30:56] <Ryan_Lane>	 Thehelpfulone: so, basically, any legitimate reason will get someone shell
[23:31:06] <Ryan_Lane>	 "I want to run a bot" - granted
[23:31:14] <Ryan_Lane>	 "I want to write a tool" - granted
[23:31:14] <Ryan_Lane>	 heh
[23:31:17] <andrewbogott>	 I'm messing with instance-info-updating for essex right now.  It's a little messy without the folsom stuff but will mostly work.
[23:31:35] <Damianz>	 folsom is the release at the end of this month then grizzly in another 6months or so?
[23:31:39] <Ryan_Lane>	 yep
[23:31:44] <Damianz>	 Cool
[23:31:56] <Damianz>	 Legtimate reason example; I LOVE UNICORNS.
[23:32:41] <Ryan_Lane>	 heh
[23:32:43] <Ryan_Lane>	  - denied
[23:33:23] <Thehelpfulone>	 who changed the colour of the unicorn? it's ugly now!
[23:33:26] <Ryan_Lane>	 that form really needs a way to have comments too
[23:33:30] <Ryan_Lane>	 Thehelpfulone: apparently we had to
[23:33:40] <Damianz>	 I like the pink one better
[23:33:52] <Thehelpfulone>	 we had to? that sounds ominous
[23:33:54] <Ryan_Lane>	 Thehelpfulone: some religious organization uses it
[23:34:12] <Ryan_Lane>	 http://en.wikipedia.org/wiki/Invisible_Pink_Unicorn
[23:34:25] <Damianz>	 We totally couldn't be seen to be related with any org ever, legal might have a heart attack
[23:34:27] <andrewbogott>	 'religious'
[23:34:34] <Ryan_Lane>	 yeah
[23:35:34] <Thehelpfulone>	 can't we choose a darker green that matches the site logo?
[23:35:44] <Ryan_Lane>	 Thehelpfulone: have at it
[23:35:53] <Damianz>	 (and make it smaller so you don't have to scroll half a page down to read the text)
[23:36:06] <Ryan_Lane>	 actually, I can probably give you the ability to change the interface however you'd like
[23:36:20] <Ryan_Lane>	 I'm nearly positive I took all other rights away from sysop
[23:36:45] <Ryan_Lane>	 that I did
[23:36:52] <Damianz>	 <!-- This is a modified source of wikipedia main page, I stole it :) --Petrb --> < LOL
[23:36:58] <Ryan_Lane>	 you don't need sysop to edit the color of the unicorn, though
[23:37:23] <Damianz>	 I keep debating making the homepage prettier than realise my thoughts of pretty generally don't agree with anyone elses :P
[23:37:29] <Ryan_Lane>	 heh
[23:37:34] <Thehelpfulone>	 I'll have a look on photoshop tomorrow
[23:37:41] <Ryan_Lane>	 cool
[23:37:42] <Ryan_Lane>	 thanks
[23:37:55] <Thehelpfulone>	 sure, Damianz by all means have a go - if we don't like it we'll just undo :P
[23:38:08] <Ryan_Lane>	 yep
[23:38:25] <Ryan_Lane>	 the help section could seriously use some love
[23:38:32] <Ryan_Lane>	 it's just a chopped down version of contents
[23:38:34] <Thehelpfulone>	 it's relatively clean at the moment though in terms of content, but yeah some people think that enwiki home page format is ugly
[23:38:36] <Damianz>	 The help section needs to DIAF
[23:38:43] <Ryan_Lane>	 hahahaha
[23:38:47] <Ryan_Lane>	 same with logging
[23:38:50] <Ryan_Lane>	 logging should die too
[23:38:58] <Damianz>	 Thehelpfulone: I was thinking of moving the bottom section to the right, making it obuvious where to find help etc.
[23:38:58] <Ryan_Lane>	 it's explained in the help contents
[23:39:06] <Ryan_Lane>	 the unicorn is too large, too
[23:39:17] <Damianz>	 The help section needs an overview page then seperate pages for everything all inter-linked
[23:39:30] <Damianz>	 Like 'I have no idea what I'm doing here, what's the internet'
[23:39:33] <Thehelpfulone>	 Ryan_Lane, labsconsole isn't going to joining SUL is it? also, if we are allowing open registration then maybe the main page should be sysop only editing?
[23:39:42] <Ryan_Lane>	 ah. right
[23:39:43] <Thehelpfulone>	 Damianz ++
[23:39:43] <Ryan_Lane>	 it should
[23:39:45] <Ryan_Lane>	 good call
[23:39:48] <Ryan_Lane>	 Thehelpfulone: no SUL
[23:39:51] <Ryan_Lane>	 this isn't in the cluster
[23:40:08] <Ryan_Lane>	 also, it requires LDAP auth
[23:40:24] <Thehelpfulone>	 ah of course
[23:40:31] <Ryan_Lane>	 I have ideas on how to kill that off over time, but keystone and mediawiki need to support oauth first
[23:40:47] <Damianz>	 Everything should support oauth
[23:40:48] <Ryan_Lane>	 and we'd need to support openid as a provider on wikimedia projects
[23:40:51] <Damianz>	 oauth makes my life better
[23:41:00] <Ryan_Lane>	 I do have plans in my head, though ;)
[23:41:14] <Damianz>	 I swaer there's a line that says it's not a plan unless it's written down ;)
[23:41:34] <Ryan_Lane>	 I probably should write it down
[23:41:48] <Damianz>	 Btw, can we kill m1.tiny in a pit of fire? You can't even run puppet on it
[23:41:50] <Ryan_Lane>	 I can finally move all of the proposal pages over to labsconsole when we open registration! :)
[23:42:02] <Ryan_Lane>	 let me open a bug for that
[23:42:13] <Ryan_Lane>	 because that flavor is indeed crap
[23:43:16] <Thehelpfulone>	 ok the whole MP/body MP templates are slightly annoying
[23:43:20] <Ryan_Lane>	 agreed
[23:43:23] <Damianz>	 very*
[23:43:32] <Ryan_Lane>	 feel free to fix that :)
[23:44:15] <Damianz>	 Some of the wikipedia templates make me cringe, it's like hey it's a templating including 4 templates that also include 3 other templates that do some crazy huge 1 line thing that no one can read.
[23:44:31] <Ryan_Lane>	 maybe we should add Lua to this :)
[23:44:55] <Thehelpfulone>	 Ryan_Lane, so shall I bring it all back to the Main Page - if so you'll need to protect it and then I'll need sysop please
[23:45:01] <Damianz>	 That would take the labs mailing list from low volume to spam :P
[23:45:01] <Ryan_Lane>	 sure
[23:45:10] <Ryan_Lane>	 Thehelpfulone: gimme a min
[23:45:15] <Thehelpfulone>	 thanks
[23:45:41] <Ryan_Lane>	 I'm giving you importer rights too
[23:45:45] <Ryan_Lane>	 just incase you need to import stuff
[23:46:33] <Ryan_Lane>	 protected
[23:46:37] <Ryan_Lane>	 and added you
[23:47:27] * Damianz  notes Thehelpfulone is now the volunteered person for all bugs relating to the homepage :D

[23:49:56] <Ryan_Lane>	 well, it's way easier to hand out admin now :)
[23:50:06] <Thehelpfulone>	 heh
[23:50:44] <Thehelpfulone>	 if you make people crats Ryan_Lane, it'd be a nice way of delegating chores.. I could delegate to Damianz as an example :P
[23:50:53] <Ryan_Lane>	 can't do that :(
[23:51:00] <Damianz>	 :P
[23:51:04] <Ryan_Lane>	 crats can grant themselves cloud admin
[23:51:52] <Ryan_Lane>	 I'm likely to kill off the bureaucrat group completely, and only have cloud admin, so there's no confusion there
[23:52:45] <Damianz>	 I totally don't understand the permissions system in mediawiki, solution TICK EVERYTHING (for self)
[23:52:53] <Ryan_Lane>	 heh
[23:53:13] <Ryan_Lane>	 admin lets you move/delete pages, edit mediawiki namespace, edit interface
[23:53:21] <Ryan_Lane>	 crat lets you manage groups
[23:53:29] <Ryan_Lane>	 bot lets you be ignored in recent changes
[23:53:40] <Ryan_Lane>	 cloud admin lets you manage everything in openstack
[23:53:43] <Ryan_Lane>	 :)
[23:53:44] <Damianz>	 Ah.. I thought bureaucrat was under admin not above it.
[23:54:27] <Damianz>	 Bot has a whole load of policy around it :P Even by 'bot' account isn't technically a bot on enwiki
[23:55:13] <Ryan_Lane>	 yeah
[23:55:23] <Ryan_Lane>	 I mean from mediawiki's perspective :)
[23:56:01] <Ryan_Lane>	 I really <3 how stable the virt boxes have been since the upgrade: http://ganglia.wikimedia.org/latest/?c=Virtualization%20cluster%20pmtpa&h=virt6.pmtpa.wmnet&m=load_one&r=hour&s=by%20name&hc=4&mc=2
[23:56:40] <Ryan_Lane>	 virt6 is the most heavily loaded node, too
[23:56:52] <Damianz>	 We totally need to use more ram
[23:57:24] <Damianz>	 The io wait doesn't look like a mountain range either :D
[23:58:18] <Ryan_Lane>	 exactly
[23:58:37] <Ryan_Lane>	 it actually looks sane
[23:59:09] <Damianz>	 It would be interesting if we could graph the active number of instances on each host
[23:59:25] <Ryan_Lane>	 yeah
[23:59:38] <Ryan_Lane>	 I put in some bugs about improving ganglia and nagios for the openstack stuff
[23:59:51] <Ryan_Lane>	 that should be a really easy plugin to write, though