Tuesday, 2008-06-17

[2008/06/17 00:08:33] <jamesturnbull> lak: and there goes github again... wasn't all this guithub stuff your idea... :)
[2008/06/17 00:08:42] <lak> yeah, i guess so
[2008/06/17 00:08:51] <hacim> github is having problems?
[2008/06/17 00:09:05] <jamesturnbull> I can't push to my repos
[2008/06/17 00:09:05] <hacim> people keep asking me if it should be used
[2008/06/17 00:09:13] <jamesturnbull> permissions errors
[2008/06/17 00:09:19] <jamesturnbull> though to be fair it's the first problem I've had
[2008/06/17 00:09:48] <lak> i assume there's some way to make sure they know
[2008/06/17 00:10:26] <jamesturnbull> lak: I will log a ticket but it seems intermittent
[2008/06/17 00:10:35] <lak> yeah, i got the same thing
[2008/06/17 00:11:09] <jamesturnbull> lak: why does it appear that there are MORE open 0.24.5 tickets than when I looked yesterday...
[2008/06/17 00:11:21] <lak> because i actually looked through the rest of the db
[2008/06/17 00:11:42] <lak> the majority of them are simple
[2008/06/17 00:11:47] <lak> i'm trying to knock them out right now
[2008/06/17 00:11:48] <jamesturnbull> lak: yeah I saw the flurry of emails :P was just teasing
[2008/06/17 00:11:55] <lak> yeah well
[2008/06/17 00:12:26] <lak> there were quite a few duplicate tickets, and not enough were correctly classified for 0.24.5, unfortunately
[2008/06/17 00:13:08] <lak> we need to get better at fixing bugs right away, rather than waiting for release time :/
[2008/06/17 00:13:43] <lak> jamesturnbull: you just modified that ticket's target version to 0.24.4, rather than 0.24.5
[2008/06/17 00:13:45] <lak> ftr
[2008/06/17 00:15:06] <jamesturnbull> lak: yeah I meant to - it was unplanned - I am going to go through and move all the closed unplanned tickets and assign them to the actual release that had the change - because technically if they were pushed then they aren't "unplanned"
[2008/06/17 00:15:18] <lak> ah
[2008/06/17 00:15:20] <lak> cool
[2008/06/17 00:15:27] <lak> didn't notice it was closed, duh
[2008/06/17 00:15:52] <msf> I have a wee patch here that enables manufacturer.rb on bsd systems
[2008/06/17 00:16:08] <jamesturnbull> lak: yeah re the fixing bugs - it'll be easier when 0.25 comes...
[2008/06/17 00:16:08] <msf> should I mail it out to the list?
[2008/06/17 00:16:17] <jamesturnbull> msf: wiki:DeveloperLifecycle
[2008/06/17 00:16:24] <lak> jamesturnbull: kind of; i mean, there'll always be that conflict
[2008/06/17 00:16:28] <lak> new dev vs. bug fixing
[2008/06/17 00:16:28] <jamesturnbull> wiki:DevelopmentLifecycle
[2008/06/17 00:16:33] <gepetto> jamesturnbull: wiki: wiki:DevelopmentLifecycle is http://reductivelabs.com/trac/puppet/wiki/DevelopmentLifecycle
[2008/06/17 00:16:56] <jamesturnbull> lak: nah I am just telling everyone that world hunger will be cured when 0.25 comes out
[2008/06/17 00:16:59] <jamesturnbull> lak: it's my stock line
[2008/06/17 00:17:01] <lak> thanks
[2008/06/17 00:17:05] <lak> no pressure, right?
[2008/06/17 00:17:13] <jamesturnbull> pressure?
[2008/06/17 00:17:16] * jamesturnbull looks innocent
[2008/06/17 00:17:59] @ randybias joined channel #puppet
[2008/06/17 00:21:04] <fujin> didn't puppetwhen/puppetlast get jammed into ext/?
[2008/06/17 00:21:25] <fujin> oh, it's closed
[2008/06/17 00:21:41] <lak> hey, if you'd fix all these 0.24.5 tickets, i could get back to 0.25 development
[2008/06/17 00:21:49] <lak> where "you" == "the whole channel" :)
[2008/06/17 00:22:18] <fujin> lol
[2008/06/17 00:23:02] <msf> just out of curiosity is there a preferred indent style for submitted code ?
[2008/06/17 00:23:08] <lak> 4 spaces
[2008/06/17 00:23:34] <msf> at least looking through facter I ave been noticing a certain amount of divergence
[2008/06/17 00:23:46] <msf> k. will set my editor to 4 space indents
[2008/06/17 00:24:05] @ edibrac joined channel #puppet
[2008/06/17 00:24:54] <msf> if for example I am working on a file that does not fillow that scheme ?
[2008/06/17 00:25:20] <msf> ie. util/manufacturer.rb uses single tab indent
[2008/06/17 00:25:36] <catdude> there's this fascinating video visualizing all the commits to Python from about '90 to 2005 .. for the first few years, it was all guido all the time
[2008/06/17 00:25:43] <catdude> then other people started to move in
[2008/06/17 00:25:54] <catdude> so luke is just trying to get past those early guido days
[2008/06/17 00:26:07] <catdude> when he can step back and take his rightful place as puppet BDFL
[2008/06/17 00:26:08] <lak> yep
[2008/06/17 00:26:30] <lak> bastard dictator from louisiana?
[2008/06/17 00:26:45] <catdude> close
[2008/06/17 00:27:00] <catdude> guido is benevolent dictator for life, i think he describes himself
[2008/06/17 00:27:04] <lak> ah
[2008/06/17 00:27:08] <lak> life's a long time
[2008/06/17 00:27:23] <lak> i hope puppet isn't as big as python, such that it doesn't require a life commitment
[2008/06/17 00:27:29] <lak> i've got other stuff i want to do before i die
[2008/06/17 00:27:43] <catdude> well that's the point of open source, right?
[2008/06/17 00:27:55] <catdude> if it's good, it's like a kid .. it grows and gets a life on its own
[2008/06/17 00:28:04] <catdude> if not .. it lives in your basement
[2008/06/17 00:28:06] <catdude> ;-)
[2008/06/17 00:28:22] <catdude> and you yell down .. Get a job!
[2008/06/17 00:29:33] <jamesturnbull> lak: I fix as many of them as I can. Though mostly all I get time for is ticket triage and documentation... need more hours in the day
[2008/06/17 00:29:42] <lak> yeah
[2008/06/17 00:29:51] <lak> catdude: heh, yeah
[2008/06/17 00:30:12] <lak> jamesturnbull: yep; it's not you who's not working hard enough
[2008/06/17 00:30:25] <lak> it's obviously a lot of "my fault" for not being a good enough dev to start with
[2008/06/17 00:30:35] <lak> so we've got a lot of technical debt to make up, with bad tests and poorly written code
[2008/06/17 00:30:52] <jamesturnbull> lak: wonder if Anthony Baxter would like a job as the 0.25 release manager...
[2008/06/17 00:30:53] <lak> but at some point we just need more people knuckling down and fixing those problems, like i'm trying to do with 0.25
[2008/06/17 00:30:59] <catdude> there's a great interview on FLOSS weekly with Ward Cunningham I listened to today .. and he makes a great point - how much work you have to put in to actually reviewing patches when people start contributing .. the other side of putting something out there
[2008/06/17 00:31:05] <lak> erm, can he tolerate ruby?
[2008/06/17 00:31:20] <jamesturnbull> lak: not he can't - but it'd be funny - will ask him at OSCON
[2008/06/17 00:31:23] <lak> catdude: we're starting to put a lot more work into it
[2008/06/17 00:31:25] <lak> heh, ok
[2008/06/17 00:31:42] <lak> july is going to be luke's nose => grindstone
[2008/06/17 00:31:52] <catdude> luke, how long since you started puppet?
[2008/06/17 00:32:00] <lak> march 2005 i went full time
[2008/06/17 00:32:09] <jamesturnbull> after 0.25.0 guess we should discuss roadmap
[2008/06/17 00:32:11] <lak> prototype was about a year earlier
[2008/06/17 00:32:16] <catdude> yeah, i remember you flogging it *hard* at LISA conferences
[2008/06/17 00:32:20] <lak> heh
[2008/06/17 00:32:23] <lak> yeah, kind of :)
[2008/06/17 00:32:27] <lak> maybe a little
[2008/06/17 00:32:28] <catdude> getting traction now
[2008/06/17 00:32:30] <catdude> takes a while
[2008/06/17 00:32:34] <lak> yep
[2008/06/17 00:32:47] <lak> jamesturnbull: well, i want to refactor transactions and the ral next
[2008/06/17 00:33:08] <lak> but depending on dev resources, it might make sense to do a language release
[2008/06/17 00:33:11] <lak> then the ral/transactions
[2008/06/17 00:33:15] <lak> which == 1.0
[2008/06/17 00:33:36] <lak> but really, language stuff is kinda extra hard, because the design part is so important
[2008/06/17 00:33:42] <lak> and harder than normal coding design
[2008/06/17 00:33:58] <lak> andrew should definitely be up to speed by the end of july, so we should actually have a dev working near full time, which i can't do
[2008/06/17 00:34:08] <lak> stupid need to eat seems to be taking too much of my time :/
[2008/06/17 00:34:20] <jamesturnbull> lak: and it'd be nice to refactor types and providers too
[2008/06/17 00:34:22] <lak> i've heard people complain that the time i spend running the company is hurting the product
[2008/06/17 00:34:36] <lak> like i could possibly even spend *any* time on the product if i weren't making money
[2008/06/17 00:34:41] <jamesturnbull> lak: I really think if you make it easy for people to prototype them it'll make the product really take off
[2008/06/17 00:34:50] <lak> jamesturnbull: yeah, i agree
[2008/06/17 00:35:08] <lak> similarly to the REST changes, half the difficulty in refactoring the ral stuff will be refactoring the existing types to the new interface
[2008/06/17 00:35:17] <lak> i'll probably, like REST, do a bit of a clean break
[2008/06/17 00:35:40] <edibrac> what well known companies use puppet?
[2008/06/17 00:35:43] <jamesturnbull> yeah it'd be easier than jury-rigging
[2008/06/17 00:35:47] <lak> i think the type interface will look *pretty* similar, but the property/parameter interfaces will be pretty different
[2008/06/17 00:35:50] <jamesturnbull> edibrac: Google
[2008/06/17 00:35:56] <jamesturnbull> edibrac: wiki:WhosUsingPuppet
[2008/06/17 00:35:59] <gepetto> jamesturnbull: edibrac: wiki:WhosUsingPuppet is http://reductivelabs.com/trac/puppet/wiki/WhosUsingPuppet
[2008/06/17 00:36:02] <lak> edibrac: google, sony, red hat, lots more
[2008/06/17 00:36:06] <edibrac> ah thanks
[2008/06/17 00:36:09] <nevyn> edibrac: just between us.. boeing.
[2008/06/17 00:36:16] <lak> :)
[2008/06/17 00:36:29] <waawaamilk> catalyst
[2008/06/17 00:36:31] <lak> the list is far longer than can admit publicly
[2008/06/17 00:36:45] <jamesturnbull> edibrac: Twitter
[2008/06/17 00:36:47] <nevyn> indeed
[2008/06/17 00:36:49] * waawaamilk floats smallcompanyname among bignames in the hope that people might not ask questions
[2008/06/17 00:37:02] <nevyn> waawaamilk: I know who catalyst are..
[2008/06/17 00:37:08] <waawaamilk> orly?
[2008/06/17 00:37:12] <jamesturnbull> which might not be a good recommendation given their service stability right now...
[2008/06/17 00:37:25] <jamesturnbull> waawaamilk: yeah nevyn is another Aussie
[2008/06/17 00:37:28] <nevyn> :)
[2008/06/17 00:37:30] <waawaamilk> durndurndurrrrn
[2008/06/17 00:37:39] <waawaamilk> :)
[2008/06/17 00:37:49] <jamesturnbull> bbiab
[2008/06/17 00:37:51] <waawaamilk> heh, ned had a puppet disaster today
[2008/06/17 00:38:01] <waawaamilk> well, to be fair it was a bit of a pebkac disaster
[2008/06/17 00:38:18] <waawaamilk> but we had a vserver in which a cool search engine for all of our work requests was running
[2008/06/17 00:38:28] <waawaamilk> and it was in git... just not pushed anywhere
[2008/06/17 00:38:42] * waawaamilk gets all tearful
[2008/06/17 00:41:07] @ Demosthenes joined channel #puppet
[2008/06/17 00:42:43] <fujin> christ, 1hr 28minutes installing ri docs for activesupport
[2008/06/17 00:42:45] * fujin cringes
[2008/06/17 00:43:04] <lak> jamesturnbull: i'm claiming that work is pushed to github, but of course it won't be until github works again
[2008/06/17 00:46:03] <randybias> gem install activesupport --no-ri --no-rdoc
[2008/06/17 00:46:12] <fujin> randybias: I want the docs :)
[2008/06/17 00:46:29] <fujin> just a shame that it takes a billion years
[2008/06/17 00:46:46] <randybias> fujin: My current strategy is that all docs should be online, not on the box, but that's me. and I agree, it's horrible how long it takes....
[2008/06/17 00:47:25] <fujin> i wish gem would make some noise "10% complete" or smething
[2008/06/17 00:50:22] @ Quit: andrewcshafer:
[2008/06/17 00:59:18] @ andrewcshafer joined channel #puppet
[2008/06/17 00:59:34] <fujin> 16:59:33 up 1 day, 3:43, 4 users, load average: 4.84, 3.91, 3.43
[2008/06/17 00:59:34] <fujin> USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
[2008/06/17 00:59:35] <fujin> aj pts/1 noc-gw:S.0 Mon13 8.00s 50.16s 50.00s irssi
[2008/06/17 00:59:35] <fujin> aj pts/2 noc-gw:S.1 Mon13 1:57 1:34m 1.05s /bin/bash
[2008/06/17 00:59:35] <fujin> aj pts/3 noc-gw:S.2 Mon14 14.00s 1.12s 1.12s /bin/bash
[2008/06/17 00:59:36] <fujin> aj pts/4 noc-gw:S.3 Mon14 1:34 0.31s 0.31s /bin/bash
[2008/06/17 00:59:40] <fujin> feck
[2008/06/17 00:59:43] <fujin> sorry
[2008/06/17 01:01:59] @ Quit: ski98033: "Leaving."
[2008/06/17 01:09:18] @ Quit: Derailed: "Leaving"
[2008/06/17 01:09:56] <catdude> well I think we should all put in a 'w' just to be in solidarity
[2008/06/17 01:10:29] <nevyn> /exec w
[2008/06/17 01:10:54] <nevyn> information leakage ftw.
[2008/06/17 01:11:39] <catdude> hey, if you haven't dumped your password into irc at least once, you haven't lived
[2008/06/17 01:11:57] <fujin> load of 5 wins
[2008/06/17 01:12:03] <fujin> stupid ass activesupport ridocs
[2008/06/17 01:12:44] <catdude> yeah, even in the days of cheap cycles, there are pigs
[2008/06/17 01:13:40] <fujin> this box aint exactly a beast or antyhing though, no wonder it's grinding
[2008/06/17 01:13:52] <catdude> what kinds specs?
[2008/06/17 01:14:05] <fujin> p4 2.4 / 256mb ddr277
[2008/06/17 01:14:08] <fujin> It's fail. 1u though.
[2008/06/17 01:14:11] <fujin> I get free colo
[2008/06/17 01:14:16] <catdude> well that's not too bad
[2008/06/17 01:14:17] <catdude> considering
[2008/06/17 01:14:18] <nevyn> giant ram fail
[2008/06/17 01:14:19] <catdude> heh
[2008/06/17 01:14:19] <fujin> does the trick for screen+irssi
[2008/06/17 01:14:27] <catdude> virtualized, or real?
[2008/06/17 01:14:29] <fujin> real.
[2008/06/17 01:14:37] <catdude> then that's not too bad then
[2008/06/17 01:14:44] <fujin> yeh, 100mbit port
[2008/06/17 01:14:50] <fujin> <3 NOC job
[2008/06/17 01:15:51] <catdude> i pay for a virtualized server .. my committed perf is (preliminary cringe) a 128mb 128mhz under xen
[2008/06/17 01:16:06] <catdude> but in theory I can burst up to 3ghz
[2008/06/17 01:16:07] <fujin> oh ya
[2008/06/17 01:16:13] <catdude> hey, it's $10 US a month
[2008/06/17 01:16:20] <fujin> that's quite nice
[2008/06/17 01:16:29] <catdude> the perf still beats the box I was running it on, which was a P3 500 dell at home
[2008/06/17 01:16:29] <fujin> wish I had a couple more cycles/ram kicking around
[2008/06/17 01:16:47] <fujin> unfortunate though is this box won't take higher than ddr233 or 277 or whatever it is
[2008/06/17 01:16:48] <catdude> i've been happy, but I don't do dev on it . .. just mail, git, web server
[2008/06/17 01:16:56] <catdude> ah well
[2008/06/17 01:17:23] <catdude> i went mac last fall .. and i've been happy .. but I'm living on a mac mini booting off an external firewire
[2008/06/17 01:17:41] <catdude> it's faster to boot on firewire .. but the throughput is about half
[2008/06/17 01:17:51] <catdude> and if I do something disk intensive .. i suffer
[2008/06/17 01:18:00] <catdude> which I could afford a Mac Pro tower
[2008/06/17 01:18:06] <catdude> :-)
[2008/06/17 01:18:08] <fujin> fuck, I'd give my left nut for a MBP.
[2008/06/17 01:18:21] <fujin> prob wouldn't even run a colobox. do all my dev on that
[2008/06/17 01:18:21] <catdude> yah, that would be good
[2008/06/17 01:18:26] <fujin> hell I'd buy a MBP just for TextMate.
[2008/06/17 01:18:42] @ Quit: Gwayne: "+++ OK ATH OK"
[2008/06/17 01:18:54] <catdude> dang, apple, just give me a reasonable priced cheap tower
[2008/06/17 01:19:33] <catdude> mac is not particular puppet friendly, though
[2008/06/17 01:19:43] <fujin> sure it is lol
[2008/06/17 01:19:47] <fujin> luke developed it on mac didn't he?
[2008/06/17 01:19:53] <catdude> did he?
[2008/06/17 01:19:56] <fujin> aye, think so
[2008/06/17 01:20:00] <catdude> he sure didn't do it on freebsd ;-)
[2008/06/17 01:20:05] <lak> nope
[2008/06/17 01:20:12] <catdude> ohhh
[2008/06/17 01:20:14] <catdude> the man speaks
[2008/06/17 01:20:22] <lak> it was mostly done early-on in debian, but mostly recently on os x
[2008/06/17 01:20:35] <catdude> hmm
[2008/06/17 01:21:03] <catdude> haven't played with it much, but my exp trying to get puppet onto mac was not good
[2008/06/17 01:21:18] <lak> really?
[2008/06/17 01:21:19] <lak> strange
[2008/06/17 01:21:22] * nevyn wants a m9000 and san but then I'm greedy.
[2008/06/17 01:21:35] <catdude> oh i'll have to play with it more
[2008/06/17 01:21:48] <fujin> nevyn: I've been trying to talk $bosses into giving me a VM on our esx farm+san
[2008/06/17 01:21:57] <catdude> the install instructions don't say much of anything about mac imho
[2008/06/17 01:22:00] <fujin> can't really stuff a vm full of disks and run torrents of it though, can we :)
[2008/06/17 01:22:02] <catdude> or maybe I missed something
[2008/06/17 01:22:12] <catdude> ohh that would be nice fujin
[2008/06/17 01:22:53] <fujin> oh, I can dream :}
[2008/06/17 01:22:55] <catdude> I'd love to have a nice big box in a colo running esx
[2008/06/17 01:23:03] <catdude> set up virtual instances by the yard
[2008/06/17 01:23:10] <fujin> gotta have boxes+san to make ESX worth it
[2008/06/17 01:23:19] <catdude> yep
[2008/06/17 01:23:23] <catdude> crazy expensive, too
[2008/06/17 01:23:24] <fujin> vmotion, storage motion and all that
[2008/06/17 01:23:51] <Cuchulain> don't you need multiple sans for storage motion to make sense?
[2008/06/17 01:23:58] <Cuchulain> and sans aren't expensive.
[2008/06/17 01:24:03] <Cuchulain> not any more.
[2008/06/17 01:24:06] <catdude> it all makes me wonder how long vmware can get away with charging what they do for esx
[2008/06/17 01:24:48] <Cuchulain> if you're consolidating 20:1 via virtualisation, then vmware can charge a lot.
[2008/06/17 01:25:10] <Cuchulain> even if you're only consolidating 5:1 or 10:1, if you're talking about hosted space, that is a lot of savings
[2008/06/17 01:25:18] <fujin> Cuchulain: nah, lun<>lun
[2008/06/17 01:25:31] <catdude> i guess
[2008/06/17 01:25:37] <fujin> buying vmware+san off Dell is sweet; they hook us up
[2008/06/17 01:25:42] <Cuchulain> i prefer citrix xenserver to vmware anyway
[2008/06/17 01:25:49] <fujin> lolwut
[2008/06/17 01:25:52] <fujin> ok, detaching
[2008/06/17 01:26:02] <catdude> had a project where I worked last year where they were trying to build out a standardized vmware+server+san for all production deploys
[2008/06/17 01:26:05] <Cuchulain> it used to be a LOT cheaper too, but the prices are normalising a bit
[2008/06/17 01:26:23] <catdude> but the price of vmware was still a factor that stalled the project
[2008/06/17 01:26:44] <catdude> but then, they were buying 20-40 servers at a time .. the vmware hit was considerable
[2008/06/17 01:26:59] <Cuchulain> yeah
[2008/06/17 01:27:03] <catdude> they were being very conservative
[2008/06/17 01:27:08] <catdude> perhaps 3-4 vms per box
[2008/06/17 01:27:12] <catdude> this was at an isp
[2008/06/17 01:27:17] <Cuchulain> right
[2008/06/17 01:27:26] <Cuchulain> that's rather conservative
[2008/06/17 01:27:41] <catdude> well, these were full on services - not customer hosting, but customer facing services
[2008/06/17 01:28:08] <catdude> which were usually rather intensive along some dimension - cpu/io
[2008/06/17 01:28:26] <catdude> virtualization was really all about standardization and getting a box assigned for a new service quickly
[2008/06/17 01:28:57] <catdude> a day or two instead of (typically) 100 days to approve capital purchase, procure, install, etc
[2008/06/17 01:30:26] <catdude> they were trying to use opsware to manage
[2008/06/17 01:30:30] <catdude> that was addfing to the cost too
[2008/06/17 01:30:46] <catdude> but vmware was much more than opsware
[2008/06/17 01:30:52] <Cuchulain> right
[2008/06/17 01:46:39] <waawaamilk> Cuchulain: !
[2008/06/17 01:52:15] <Cuchulain> waawaamilk?
[2008/06/17 01:52:34] <waawaamilk> oh.. you're not donal?
[2008/06/17 01:52:53] <waawaamilk> damn, thought you were someone else
[2008/06/17 01:52:55] <waawaamilk> ;)
[2008/06/17 01:53:01] <Cuchulain> i'm not donal, or donald, no
[2008/06/17 01:56:13] @ jsgotangco joined channel #puppet
[2008/06/17 02:12:59] <jamesturnbull> catdude: wiki:PuppetMacOSX
[2008/06/17 02:13:01] <gepetto> jamesturnbull: catdude: wiki:PuppetMacOSX is http://reductivelabs.com/trac/puppet/wiki/PuppetMacOSX
[2008/06/17 02:13:20] @ Quit: fbe__: Read error: 110 (Connection timed out)
[2008/06/17 02:14:30] <catdude> i did see that, but there was no simple 'download this package' as there was for other architectures .. no clear alternative to installing from source
[2008/06/17 02:14:36] <catdude> or did I miss somethhing?
[2008/06/17 02:14:58] <lak> install from source?
[2008/06/17 02:15:06] <lak> people have variously maintained os x packages
[2008/06/17 02:15:11] <catdude> geeze luke doesn't sleep
[2008/06/17 02:15:14] <lak> but i don't think there's a real platform advocate right now
[2008/06/17 02:15:21] <lak> it's only 1:15
[2008/06/17 02:15:25] <catdude> ahh .. cdt
[2008/06/17 02:15:28] <catdude> that does it
[2008/06/17 02:15:40] <lak> and i'm trying to figure out wtf to do about #1232
[2008/06/17 02:15:41] <gepetto> lak: #1232 is http://reductivelabs.com/redmine/issues/show/1232
[2008/06/17 02:16:38] <lak> it's sad when hunger is what eventually drives one to bed :/
[2008/06/17 02:16:49] <catdude> I certainly didn't poke around a lot - didn't see a puppet package for macports - and given that mac is different, i really wanted a package of some kind
[2008/06/17 02:17:03] <catdude> but ..
[2008/06/17 02:17:07] <catdude> it's really a red herring for me
[2008/06/17 02:17:19] <lak> well, it's not like packages get you, um, a damn thing, on macs
[2008/06/17 02:17:19] <catdude> i would not expect I would manage mac os x with puppet
[2008/06/17 02:17:20] <lak> use gems
[2008/06/17 02:17:25] <lak> google does
[2008/06/17 02:17:29] <lak> 5k+ nodes
[2008/06/17 02:17:36] <catdude> yep, heard that talk
[2008/06/17 02:17:42] <lak> and someone else is in the middle of turning up 7500 mac nodes
[2008/06/17 02:17:59] <catdude> google, of course, is so *generous* with details ;-)
[2008/06/17 02:18:12] <catdude> but in my case, it was just a place to play with puppet
[2008/06/17 02:18:24] <catdude> i ended up using fusion to set up more reasonable virtual machines
[2008/06/17 02:18:31] <lak> cool
[2008/06/17 02:18:32] <catdude> that were easier to install
[2008/06/17 02:18:40] <catdude> so i'm just kevetching
[2008/06/17 02:18:57] <catdude> i understand OS X is a strange hybrid
[2008/06/17 02:19:00] <catdude> unix, but not
[2008/06/17 02:19:11] <nevyn> kinda like.. Aix
[2008/06/17 02:19:24] <lak> at least aix packages can be deleted
[2008/06/17 02:19:39] <nevyn> true.
[2008/06/17 02:19:56] <catdude> i will say that macports have let me pretend it's like normal unix
[2008/06/17 02:20:03] <catdude> i know ports are not ideal vs packages
[2008/06/17 02:20:05] <catdude> but
[2008/06/17 02:20:34] <catdude> (thinking of luke's screed against freebsd here)
[2008/06/17 02:20:58] <lak> ports are theoretically fine as packages
[2008/06/17 02:21:12] <lak> it's just that they've been implemented in a way that makes them nearly impossible to control with software
[2008/06/17 02:21:23] <lak> which is essentially my "good software/bad software" boolean test
[2008/06/17 02:21:29] <catdude> and if you're trying to deploy piles of boxes
[2008/06/17 02:21:31] <catdude> slow
[2008/06/17 02:21:47] <lak> yep
[2008/06/17 02:21:48] <catdude> ah, fair point
[2008/06/17 02:22:31] <catdude> well I think the whole meme of "you should NOT be doing admin on a box by box basis" has not filtered back into OS distributions to the level we'd like
[2008/06/17 02:22:48] <lak> correct
[2008/06/17 02:23:02] <lak> and esp. the bsd guys, because they've all been wizards for 15 years
[2008/06/17 02:23:15] <lak> and that wizardry doesn't translate up a level too well
[2008/06/17 02:23:20] <catdude> i wonder what Yahoo has done with all that
[2008/06/17 02:23:33] <catdude> they must be running large large numbers of servers
[2008/06/17 02:23:45] <catdude> perhaps it's all cloning and simple tools
[2008/06/17 02:23:46] <lak> i'm obviously six shades of biased
[2008/06/17 02:23:49] <catdude> sure
[2008/06/17 02:23:59] <lak> i think "all" and "yahoo" don't have much to do with each other
[2008/06/17 02:24:01] <lak> from what i've heard
[2008/06/17 02:24:25] <catdude> the problem is, in a way ... that the guys who Really have to scale to huge numbers .. google, yahoo, whoever else
[2008/06/17 02:24:36] <catdude> just end up doing it in such a brute force way
[2008/06/17 02:24:38] <catdude> perhaps
[2008/06/17 02:25:00] <catdude> that their experience doesn't produce anything that we could use down in the dozens to thousands of servers range
[2008/06/17 02:25:35] <lak> yep
[2008/06/17 02:25:36] <lak> exactly
[2008/06/17 02:25:40] <lak> not a general solution
[2008/06/17 02:25:46] <catdude> and, of course, for a place like google, being able to operate thousands of servers is their special sauce
[2008/06/17 02:25:48] <lak> mainframe solution, essentially
[2008/06/17 02:25:55] <catdude> did I say thousands?
[2008/06/17 02:25:59] <catdude> i meant millions
[2008/06/17 02:26:12] <catdude> or at least 10e5
[2008/06/17 02:27:05] <catdude> hell, the 1-10 range may be the most critical, arguable
[2008/06/17 02:27:07] <catdude> ly
[2008/06/17 02:27:20] <catdude> that's where you get the individual admin
[2008/06/17 02:27:23] <catdude> to start changing their ways
[2008/06/17 02:28:04] <catdude> cause when they get to hundreds or thousands, it's usually too late
[2008/06/17 02:35:25] <msf> 1-10 is the hardest to get someone to realize they need to change their ways
[2008/06/17 02:36:07] <lak> eh, most wizards can easily handle 25-50 machines by him or herself
[2008/06/17 02:36:27] <msf> not without some form of home grown automation
[2008/06/17 02:36:27] <lak> i was managing a couple hundred largely on my own ten years ago
[2008/06/17 02:36:30] <lak> sure
[2008/06/17 02:36:37] <lak> but those tools don't have any higher concepts
[2008/06/17 02:36:44] <lak> basically just ssh and a for loop
[2008/06/17 02:37:00] <msf> I was using rdist over ssh
[2008/06/17 02:37:01] <lak> and that's kind of the problem with the wizards -- they can get so close that they think it's a sufficient long term solution
[2008/06/17 02:37:12] <lak> heh, basically pretty similar
[2008/06/17 02:37:18] <lak> rsync, rdist, ssh
[2008/06/17 02:37:21] <lak> rdist is more powerful
[2008/06/17 02:37:28] <lak> you can make it almost like a high-level tool
[2008/06/17 02:37:41] <msf> almost but not quite
[2008/06/17 02:37:45] <lak> yeah
[2008/06/17 02:37:45] <catdude> right, so the trick is to make it low-enough mental overhead to get started with small numbers of machines
[2008/06/17 02:37:50] <lak> yep
[2008/06/17 02:38:01] <lak> it has to be much easier to do things the "right" way
[2008/06/17 02:38:08] <catdude> absolutely
[2008/06/17 02:38:09] <catdude> leverage
[2008/06/17 02:38:14] <lak> and, of course, we must laugh uproarously at the people who do it wrong :)
[2008/06/17 02:38:27] <lak> "if you can't be a good example, you'll just have to be a horrible warning" and all that
[2008/06/17 02:38:51] <catdude> image I have is these kinds of tools, done right, are like Ripley from .. which Aliens movie? with the waldos .. ready to do battle
[2008/06/17 02:38:58] <catdude> strap these on
[2008/06/17 02:39:14] <catdude> ah, or like Ironman!
[2008/06/17 02:39:20] <catdude> there's the newer reference
[2008/06/17 02:43:29] <catdude> having said all that, I think the roadblock - for me at least - into understanding puppet was the declarative nature of puppet
[2008/06/17 02:43:41] <catdude> i kept wondering .. ok, where does the work get done?
[2008/06/17 02:43:45] <catdude> how?
[2008/06/17 02:44:10] <lak> magic, man
[2008/06/17 02:44:12] <catdude> and that's the nature of admin - so practical - you want to know how it's done, and you distrust what you can't see
[2008/06/17 02:44:15] <catdude> LOL
[2008/06/17 02:44:24] <lak> kind of
[2008/06/17 02:44:41] <catdude> well, thinking from a more jr admin perspective
[2008/06/17 02:44:49] <lak> i mean, most admins have certain areas where they really want to know everything
[2008/06/17 02:44:57] <catdude> right
[2008/06/17 02:45:01] <lak> but a lot of other areas, they're quite comfortable with ignorance
[2008/06/17 02:45:12] <lak> there's some layer where they want to know, but they don't care above or below
[2008/06/17 02:45:22] <catdude> yeah, and in the podcast you did recently, i appreciated the statement where you said ..
[2008/06/17 02:45:29] <catdude> now I'd just prefer to let puppet do it
[2008/06/17 02:45:31] <lak> "kill all humans"
[2008/06/17 02:45:32] <catdude> it knows the fiddly details
[2008/06/17 02:45:36] <lak> oh, sorry, different podcast :)
[2008/06/17 02:45:41] <catdude> lol
[2008/06/17 02:46:20] <catdude> and that's exactly the right spirit - puppet knows all those little options and details that are different system to system
[2008/06/17 02:46:26] <catdude> adduser does this here, and that there
[2008/06/17 02:46:50] <lak> and in many cases, don't actually know what they do to the disk, only what they do to behaviour
[2008/06/17 02:47:14] <catdude> right
[2008/06/17 02:47:17] * Volcane mostly loves puppet cos senior admins can write classes and defines to impliment policy, and junior admins/developers can do stuff 100% withtin defined admin policies without ever knowing or understanding them
[2008/06/17 02:47:59] <Volcane> make a define to create a vhost the way the senior guys want it, and anyone can go apache::vhost{"www.my.com": }
[2008/06/17 02:48:00] <catdude> so I think puppet does have a transparency issue
[2008/06/17 02:48:11] <catdude> that is mostly a barrier to get over
[2008/06/17 02:48:17] <lak> yeah
[2008/06/17 02:48:19] <catdude> ie, at first you want to know what it does
[2008/06/17 02:48:24] <catdude> so you can understand and trust it
[2008/06/17 02:48:31] <lak> i try to fix that by making it as clear as i can , but any abstraction layer has some opacity
[2008/06/17 02:48:32] <catdude> and after that you go .. fine, let it do the work
[2008/06/17 02:48:39] <msf> I love puppet because I can train juniors to make their own classes and use version control systems make them own a module and then sit back and read diffs and irc all day long :-P
[2008/06/17 02:49:34] <msf> make them do all the messy work ;-)
[2008/06/17 02:49:42] <Volcane> my clients dont generally have sysadmins at all, with puppet providing abstraction developers can do stuff now that they never could and do it the way i like it
[2008/06/17 02:49:50] <lak> nice
[2008/06/17 02:49:56] <catdude> yeah, that is nice
[2008/06/17 02:50:17] <catdude> stuff like ec2 pushes orgs in that direction
[2008/06/17 02:50:23] <catdude> they don't want sysadmins
[2008/06/17 02:50:28] <catdude> and datacenters
[2008/06/17 02:50:57] <catdude> if you're just trying to get a app up
[2008/06/17 02:51:08] <catdude> but then .. of course .. if you're careful, you get twitter.
[2008/06/17 02:51:09] <Volcane> well my main client has 30 odd servers of their own, no sysadmins, they dont even have root on their kit, puppet managest he machines, sudo for them to deploy their code onto puppet managed machines
[2008/06/17 02:51:26] <catdude> excellent.
[2008/06/17 02:52:17] <catdude> maybe it's like anything else .. the folks who hang onto the old ways have to retire
[2008/06/17 02:52:18] <catdude> or die
[2008/06/17 02:52:53] <catdude> or .. these days .. get laid off
[2008/06/17 02:53:19] <lak> nah
[2008/06/17 02:53:29] <lak> there are far too many companies look for clock-punchers
[2008/06/17 02:53:35] <edibrac> i think it's a great time to be a sysadmin ... at the tail end of an era
[2008/06/17 02:53:39] <lak> "just don't ever move from that chair, we'll call if we need you"
[2008/06/17 02:53:44] <edibrac> which will morph into something else
[2008/06/17 02:53:59] <catdude> yeah, clone yourself and put your clone in that chair
[2008/06/17 02:54:13] <catdude> that's the Calvin approach, I think
[2008/06/17 02:54:37] <msf> lol
[2008/06/17 02:54:42] <msf> $ git commit -s -a
[2008/06/17 02:54:42] <msf> Segmentation fault (core dumped)
[2008/06/17 02:54:49] <msf> wheee
[2008/06/17 02:55:33] <catdude> oh yeah, i'm talking about clarity and transparency .. and then there's git
[2008/06/17 02:55:37] <catdude> it's ALL magic
[2008/06/17 02:55:43] <catdude> (at least the implementation)
[2008/06/17 02:55:47] <lak> heh
[2008/06/17 02:56:00] <catdude> Don't Question Linus; it's brilliant
[2008/06/17 02:56:36] <catdude> had to love his Google talk
[2008/06/17 02:56:48] <catdude> essentially: you're an idiot if you use anything but git
[2008/06/17 02:57:51] <lak> g'night all
[2008/06/17 02:57:54] @ Quit: lak:
[2008/06/17 03:00:22] <MrProper_> anyone seen this before v 24.4 "err: Got an uncaught exception of type ArgumentError: Field 'key' is required" when using @@sshkey { $fqdn: type => rsa, key => $sshrsakey }\Sshkey <<||>>
[2008/06/17 03:06:02] @ Quit: MrProper_: "Leaving"
[2008/06/17 03:06:54] @ Quit: Ol_:
[2008/06/17 03:13:27] @ mellen joined channel #puppet
[2008/06/17 03:15:27] @ Quit: catdude:
[2008/06/17 03:16:06] <jamesturnbull> anyone else having issues with reductivelabs.com site?
[2008/06/17 03:17:48] <Volcane> works a charm here
[2008/06/17 03:17:51] <mcbride> seems to be working ok for me.
[2008/06/17 03:17:52] <Volcane> actually fast for a change
[2008/06/17 03:18:38] <kjetilho> here's a nice site: http://downforeveryoneorjustme.com/
[2008/06/17 03:19:08] <Volcane> heh
[2008/06/17 03:31:49] @ Gwayne joined channel #puppet
[2008/06/17 03:31:52] @ roald joined channel #puppet
[2008/06/17 03:32:02] @ Quit: randybias:
[2008/06/17 03:33:14] <jamesturnbull> fujin: so Trac repo browsing should now be disabled
[2008/06/17 03:33:19] @ randybias joined channel #puppet
[2008/06/17 03:33:38] <jamesturnbull> Volcane: yeah seems to have been a hiccup from my end
[2008/06/17 03:34:33] <jamesturnbull> msf: does BSD have the "where" command?
[2008/06/17 03:34:52] <jamesturnbull> msf: I mean "whereis"
[2008/06/17 03:35:13] <mcbride> yes
[2008/06/17 03:35:17] <jamesturnbull> msf: or "which"
[2008/06/17 03:35:36] <mcbride> "HISTORY: The whereis command appeared in 3.0BSD."
[2008/06/17 03:35:54] <jamesturnbull> msf: if it has which you could probably simply that facter patch
[2008/06/17 03:36:19] @ DerekW joined channel #puppet
[2008/06/17 03:36:29] <gepetto> ::puppet:: Ticket #1264 (defect closed): Documentation for pkgdmg provider breaks TypeReference @ http://reductivelabs.com/trac/puppet/ticket/1264#comment:2 (by james@lovedthanlost.net)
[2008/06/17 03:36:30] <gepetto> ::puppet:: Ticket #1267 (defect created): Fix test running in install.rb @ http://reductivelabs.com/trac/puppet/ticket/1267 (by james@lovedthanlost.net)
[2008/06/17 03:36:30] <gepetto> ::puppet:: Ticket #1242 (defect closed): Collect statements should not throw an exception if storeconfigs is no... @ http://reductivelabs.com/trac/puppet/ticket/1242#comment:3 (by james@lovedthanlost.net)
[2008/06/17 03:36:30] <gepetto> ::puppet:: Ticket #1266 (defect closed): tarball release of puppet 0.24.4 lacks test/data directory @ http://reductivelabs.com/trac/puppet/ticket/1266#comment:3 (by james@lovedthanlost.net)
[2008/06/17 03:36:30] <gepetto> ::puppet:: Ticket #1267 (defect closed): Fix test running in install.rb @ http://reductivelabs.com/trac/puppet/ticket/1267#comment:2 (by james@lovedthanlost.net)
[2008/06/17 03:36:30] <gepetto> ::puppet:: Ticket #1271 (defect closed): misleading error message if CA private key can not be decrypted @ http://reductivelabs.com/trac/puppet/ticket/1271#comment:1 (by james@lovedthanlost.net)
[2008/06/17 03:36:31] <gepetto> ::puppet:: Ticket #1271 (defect closed): misleading error message if CA private key can not be decrypted @ http://reductivelabs.com/trac/puppet/ticket/1271#comment:5 (by james@lovedthanlost.net)
[2008/06/17 03:37:52] <mcbride> so whereis has been in BSD since linus was about 11 years old :-)
[2008/06/17 03:38:04] <jamesturnbull> mcbride: :)
[2008/06/17 03:40:34] <jamesturnbull> andrewcshafer: #1373 is fixed - I just hadn't pushed the commit
[2008/06/17 03:40:37] <gepetto> jamesturnbull: andrewcshafer: #1373 is http://reductivelabs.com/redmine/issues/show/1373
[2008/06/17 03:40:56] <andrewcshafer> my bad
[2008/06/17 03:41:17] <jamesturnbull> andrewcshafer: was waiting to see if anyone commented on it - the only place I could find reference was bin/puppet
[2008/06/17 03:41:21] <andrewcshafer> I just fixed it again :/
[2008/06/17 03:42:03] <andrewcshafer> we really need a continuous integration setup
[2008/06/17 03:42:49] * Volcane hates cruisecontrol :P
[2008/06/17 03:43:04] <jamesturnbull> andrewcshafer: word
[2008/06/17 03:43:58] @ Innocenti joined channel #puppet
[2008/06/17 03:44:29] <andrewcshafer> especially since the executables are relatively untestable until they are broken into classes
[2008/06/17 03:49:18] @ zipkid joined channel #puppet
[2008/06/17 03:52:37] @ Quit: randybias: Read error: 110 (Connection timed out)
[2008/06/17 04:04:23] @ mdray joined channel #puppet
[2008/06/17 04:09:26] <fujin> jamesturnbull: oh, nice work
[2008/06/17 04:09:33] <fujin> so pygit is decoupled from it completely?
[2008/06/17 04:09:40] <fujin> I guess we should see less issues with it now
[2008/06/17 04:09:57] @ Quit: Wakko666: Read error: 110 (Connection timed out)
[2008/06/17 04:10:58] @ Wakko666 joined channel #puppet
[2008/06/17 04:12:15] <jamesturnbull> fujin: I hope so - I disabled the repo module and the git plugin
[2008/06/17 04:12:38] <fujin> super
[2008/06/17 04:12:49] <fujin> can you still view old tickets in it?
[2008/06/17 04:13:20] <jamesturnbull> fujin: yes and the wiki
[2008/06/17 04:13:29] <fujin> ah yep
[2008/06/17 04:13:38] <fujin> oh sweet, the tickets button takes you straight to redmine
[2008/06/17 04:14:43] @ Quit: rcoup:
[2008/06/17 04:14:53] <jamesturnbull> fujin: as does the New Ticket button
[2008/06/17 04:15:13] <fujin> awesome :}
[2008/06/17 04:15:49] <fujin> there was some confusion in here the other day regarding tickets
[2008/06/17 04:15:54] <fujin> hopefully that has resolved it
[2008/06/17 04:16:27] <jamesturnbull> fujin: I did it the same tiem I did the cutover
[2008/06/17 04:16:33] <jamesturnbull> fujin: and updated all the odco
[2008/06/17 04:17:20] <fujin> yeah, I saw.. some people blinders on you know
[2008/06/17 04:18:04] <jamesturnbull> fujin: though all old tickets were migrated to Redmine too - should have the same ticket # too
[2008/06/17 04:18:20] <fujin> what's up with the 'admin changed 1 to 7' I see in redmine?
[2008/06/17 04:18:44] <jamesturnbull> fujin: ah yes - minor issue there - we changed our mind about the statuses after we implemented
[2008/06/17 04:18:54] <fujin> ah.
[2008/06/17 04:19:03] <jamesturnbull> fujin: so those are "Old" statuses now deleted and referenced via record #
[2008/06/17 04:19:11] <jamesturnbull> bbiab
[2008/06/17 04:22:32] @ Quit: Gwayne: Remote closed the connection
[2008/06/17 04:22:39] @ |Innocenti| joined channel #puppet
[2008/06/17 04:22:42] @ Quit: Innocenti: Client Quit
[2008/06/17 04:24:17] @ |Innocenti| is now known as Innocenti
[2008/06/17 04:26:11] @ hessml is now known as hessml|away
[2008/06/17 04:38:05] @ Quit: hessml|away: "Leaving..."
[2008/06/17 04:48:37] <kjetilho> anyone here with access to the mail log on Redmine?
[2008/06/17 04:51:18] @ bss joined channel #puppet
[2008/06/17 04:53:18] <fujin> the mail log on redmine?
[2008/06/17 04:53:34] <fujin> for outbound ticket mails?
[2008/06/17 04:53:46] <bss> hey, i have a problem with puppet, i'm trying to get exported resources to work. It works fine when i just define a file without any tags, but doesn't work when i define a tag - has anybody experienced the same?
[2008/06/17 04:53:50] <kjetilho> for activation link mails :)
[2008/06/17 04:54:24] <fujin> kjetilho: can't say I do - I think the only people with access to that box are lak and jamesturnbull
[2008/06/17 04:54:36] <kjetilho> I registered yesterday, but there's no sign of the e-mail
[2008/06/17 04:54:41] <kjetilho> fujin: ok, thanks!
[2008/06/17 04:54:48] <fujin> I didn't think it actually sent out emails?
[2008/06/17 04:54:56] <fujin> for activation
[2008/06/17 04:55:01] <fujin> you tried logging in etc?
[2008/06/17 04:55:16] <mcbride> it also sends emails if you need to reset passwords
[2008/06/17 04:55:22] <mcbride> I did it yesterday, it worked fine.
[2008/06/17 04:56:10] <kjetilho> weird, now it says my username is unknown.
[2008/06/17 04:56:24] <kjetilho> perhaps it doesn't distinguish between "not activated" and "unknown"
[2008/06/17 04:57:30] <kjetilho> hmm! are there two sets? I registered via a link on the bug tracking system
[2008/06/17 04:57:52] <kjetilho> right, trac vs. redmine. trac has no activation
[2008/06/17 04:58:17] <fujin> you registered on trac, or redmine?
[2008/06/17 04:58:56] <kjetilho> Redmine. I just registered on Trac, that was uneventful
[2008/06/17 04:59:15] <kjetilho> tried to re-register on Redmine now, and it says login and e-mail address is taken
[2008/06/17 04:59:44] <fujin> o_0
[2008/06/17 04:59:59] <kjetilho> there doesn't seem to be a "forgot your password" link on Redmine
[2008/06/17 05:00:10] <kjetilho> sorry, I'm blind
[2008/06/17 05:00:29] <fujin> http://reductivelabs.com/redmine/account/lost_password
[2008/06/17 05:00:30] <fujin> lol
[2008/06/17 05:00:43] <fujin> I didn' thave to activate - but mya ccount was copied from trac
[2008/06/17 05:00:46] <fujin> so I dunno sorry :(
[2008/06/17 05:00:52] <fujin> prod james or luke when they're around
[2008/06/17 05:01:17] <kjetilho> ah, lak is Luke, didn't realise
[2008/06/17 05:01:29] <fujin> oh, ya
[2008/06/17 05:25:00] <kjetilho> doh, I found the message. it wasn't addressed to me, but to "undisclosed-recipients : ;"
[2008/06/17 05:25:20] <kjetilho> so it was put in my burgeoning spam folder
[2008/06/17 05:26:35] <fujin> ah.
[2008/06/17 05:26:38] <jamesturnbull> kjetilho: it bcc's things
[2008/06/17 05:27:26] <kjetilho> jamesturnbull: not very convenient :-/
[2008/06/17 05:28:23] @ Quit: jsgotangco: "Ciao"
[2008/06/17 05:28:48] <jamesturnbull> kjetilho: yes well perhaps I'll turn it off and see if anyone compains
[2008/06/17 05:29:27] <kjetilho> but it's not working for me, anyway -- after resetting the password, I still get invalid user or password
[2008/06/17 05:31:52] <jamesturnbull> kjetilho: what user name?
[2008/06/17 05:32:05] <kjetilho> kjetilho
[2008/06/17 05:34:39] <jamesturnbull> kjetilho: try again now
[2008/06/17 05:35:51] <kjetilho> excellent, thanks :)
[2008/06/17 05:43:06] @ sparanjape joined channel #puppet
[2008/06/17 05:43:57] @ Quit: Innocenti: "So Long, Thanks for all the fish"
[2008/06/17 05:52:00] @ Innocenti joined channel #puppet
[2008/06/17 06:00:16] @ Quit: Maliuta: Connection timed out
[2008/06/17 06:00:55] @ Maliuta joined channel #puppet
[2008/06/17 06:10:11] @ yure joined channel #puppet
[2008/06/17 06:13:04] @ exodos joined channel #puppet
[2008/06/17 06:28:51] @ Vivek joined channel #puppet
[2008/06/17 06:35:23] @ Quit: Innocenti: Client Quit
[2008/06/17 06:42:28] @ rcoup joined channel #puppet
[2008/06/17 06:45:01] @ Quit: rcoup: Read error: 104 (Connection reset by peer)
[2008/06/17 06:55:08] @ Quit: kenvandine: "Ex-Chat"
[2008/06/17 07:13:21] @ _zsh joined channel #puppet
[2008/06/17 07:25:13] @ Quit: roald: Read error: 104 (Connection reset by peer)
[2008/06/17 07:27:49] @ roald joined channel #puppet
[2008/06/17 07:40:57] @ Quit: sparanjape: "A day without sunshine is like .... night"
[2008/06/17 07:54:22] @ ashp joined channel #puppet
[2008/06/17 07:59:23] <ashp> http://gizmodo.com/5016555/r2+d2s-in-ur-serverz-monitoring-ur-packetz
[2008/06/17 07:59:28] <ashp> I have a new 'nagios' project :)
[2008/06/17 07:59:33] <ashp> Shoehorning it into r2-d2.
[2008/06/17 07:59:39] @ lomur joined channel #puppet
[2008/06/17 07:59:41] <lomur> puppet
[2008/06/17 07:59:53] <ashp> punch and judy
[2008/06/17 07:59:59] <lomur> puppet
[2008/06/17 08:10:04] @ Quit: duffbeer703:
[2008/06/17 08:14:52] <jamesturnbull> ashp: okay that's seriously weird
[2008/06/17 08:15:09] <lomur> puppet
[2008/06/17 08:15:19] <ashp> jamesturnbull: That's the japanese for you :)
[2008/06/17 08:15:23] <lomur> puppet
[2008/06/17 08:15:30] <jamesturnbull> ashp: I love the Japanese - marvellous people
[2008/06/17 08:15:40] <jamesturnbull> ashp: with the exception of the whale slaughtering I guess
[2008/06/17 08:15:47] <ashp> jamesturnbull: I think that nagios thing would be better if it was a real hologram.
[2008/06/17 08:16:00] <ashp> The whales thing makes me sad, I went to see this whale imax 3d thing with the kids recently, it was awesome.
[2008/06/17 08:16:32] <jamesturnbull> ashp: "Help Me Obi-Wan - I have a CRITICAL Nagios alert - Help me Obi-Wan - you're my only hope..."
[2008/06/17 08:17:02] <lomur> puppet
[2008/06/17 08:17:49] <ashp> ok, lomur is quickly getting annoying.
[2008/06/17 08:17:50] @ kenvandine joined channel #puppet
[2008/06/17 08:17:59] <lomur> poor ashp
[2008/06/17 08:18:24] <jamesturnbull> ashp: just ignore the troll
[2008/06/17 08:18:34] <lomur> jamesturnbull sucks
[2008/06/17 08:19:52] @ Mode +o jamesturnbull by ChanServ
[2008/06/17 08:22:28] @ RichiH joined channel #puppet
[2008/06/17 08:24:30] @ Quit: lomur: K-lined
[2008/06/17 08:24:33] @ RichiH left channel #puppet ()
[2008/06/17 08:25:17] <jamesturnbull> damn was just about to kick him - ah well small pleasues
[2008/06/17 08:28:09] <Volcane> ashp: i have a nagaztag bunny showing y nagios status
[2008/06/17 08:28:14] <Volcane> nabaztag even
[2008/06/17 08:29:18] <Volcane> if anything is in warning state and not acknowledge or silenced, it droops one ear, critical it droops both
[2008/06/17 08:29:32] <Volcane> if nagios alerts as a whole is disabled it'll flash a color to remind me
[2008/06/17 08:38:50] <ashp> :)
[2008/06/17 08:39:04] <ashp> is that's real, I want a picture!
[2008/06/17 08:39:47] @ Mode -o jamesturnbull by ChanServ
[2008/06/17 08:45:04] @ ski98033 joined channel #puppet
[2008/06/17 08:49:58] <ashp> hmmm, now to hunt down lighttpd 1.5 rpms
[2008/06/17 08:51:00] @ Quit: yure: Remote closed the connection
[2008/06/17 08:57:50] <jamesturnbull> ashp: I didn't think 1.5 was released?
[2008/06/17 08:59:32] <ashp> I don't think it is, but my father is telling me he needs it for some 'multi file upload' functionality
[2008/06/17 08:59:45] <ashp> I'm not really sure why, but he might be out of luck as I don't want to maintain it by hand just for wordpress
[2008/06/17 08:59:59] <ashp> I also discovered why he had to reboot the server when it was 'unbearably slow': https://bugzilla.redhat.com/show_bug.cgi?id=444759
[2008/06/17 09:00:08] <ashp> there's a terrible bug with 3ware raid cards :)
[2008/06/17 09:00:34] <ashp> "Not only the
[2008/06/17 09:00:35] <ashp> performance increase is 40 times (!) when it comes to write I/O"
[2008/06/17 09:08:38] @ Quit: magnachef__: Read error: 110 (Connection timed out)
[2008/06/17 09:08:41] <ashp> besides, looks like from svn lighttpd has stalled out
[2008/06/17 09:09:35] <ashp> It's a shame as I refuse to run apache again :(
[2008/06/17 09:10:47] <ashp> I guess I should look at nginx or litespeed next
[2008/06/17 09:11:22] <lazzurs> Hello, I was wondering what people here are doing to 'bootstrap' their puppet deployments. eg a new lab or datacentre has to be installed unattended, do you just add puppet to the install process (eg kickstart) or do something else?
[2008/06/17 09:11:34] <ashp> I add puppet to kickstart
[2008/06/17 09:11:46] <ashp> I have some kickstart stuff that runs puppet during the kickstart so it starts a cert request
[2008/06/17 09:11:52] <ashp> and then i sign that, and it's set to start puppet on boot
[2008/06/17 09:12:21] <lazzurs> yea, I presume there is some way to automatically sign all new clients? or is that going to have to be a script in cron?
[2008/06/17 09:12:45] <ashp> You can turn on autosigning
[2008/06/17 09:12:53] <ashp> I'm not sure how as I only bring up like one box a month at best, but it's possible :)
[2008/06/17 09:14:14] <lazzurs> as long as it can be done I will find it :) the situation I am in is I am going to have monkies installing remote datacentres for me an the deployment of the farm in each datacentre has to be fully automated, while the configuration is mostly static I am using configuration management (puppet) to ensure they don't mess something up and bring down the farm
[2008/06/17 09:14:18] <jamesturnbull> lazzurs: there is an example with Debian at wiki:BootstrappingWithPuppet
[2008/06/17 09:14:22] <gepetto> jamesturnbull: lazzurs: wiki:BootstrappingWithPuppet is http://reductivelabs.com/trac/puppet/wiki/BootstrappingWithPuppet
[2008/06/17 09:14:54] <lazzurs> thanks jamesturnbull, gepetto
[2008/06/17 09:15:10] <lazzurs> while this is a RHEL deployment I am sure that will help
[2008/06/17 09:15:38] <jamesturnbull> lazzurs: post on the list - there are a lot of Red Hat peoples there who use uppet with kickstart
[2008/06/17 09:15:45] <ashp> I'll dig out my bit of fluff for it, hang on
[2008/06/17 09:16:38] <ashp> http://pastebin.com/d2133ac69
[2008/06/17 09:16:48] <ashp> that's what I do to get it going, it's probably not ideal but proven to work :)
[2008/06/17 09:20:18] <jamesturnbull> ashp: be nice to add that to the wiki with some documentation - hint hint
[2008/06/17 09:21:03] <ashp> jamesturnbull: I'll go tweak the bootstrapping with puppet
[2008/06/17 09:23:09] <jamesturnbull> ashp: that'd be great
[2008/06/17 09:32:23] @ lak joined channel #puppet
[2008/06/17 09:35:59] @ muerr joined channel #puppet
[2008/06/17 09:40:19] <ashp> I hate the wiki because of this !rst stuff.
[2008/06/17 09:40:23] <muerr> fyi, ftp://ftp.mozilla.org/pub/firefox/releases/3.0/win32/en-US
[2008/06/17 09:40:29] <muerr> or ftp://ftp.mozilla.org/pub/firefox/releases/3.0/
[2008/06/17 09:41:02] @ Quit: lak: Read error: 104 (Connection reset by peer)
[2008/06/17 09:41:34] @ magnachef_ joined channel #puppet
[2008/06/17 09:41:39] @ lak joined channel #puppet
[2008/06/17 09:43:06] @ Quit: lak: Read error: 104 (Connection reset by peer)
[2008/06/17 09:44:08] @ lak joined channel #puppet
[2008/06/17 09:44:26] @ Quit: lak: Read error: 104 (Connection reset by peer)
[2008/06/17 09:45:24] @ lak joined channel #puppet
[2008/06/17 09:48:35] <Disconnect> muerr: yah but spreadfirefox.com or mozilla.com both push ffx2 as new, download links, etc
[2008/06/17 09:50:33] <ashp> http://reductivelabs.com/trac/puppet/wiki/BootstrappingWithPuppet try that james?
[2008/06/17 09:51:05] <gepetto> ::puppet:: Bootstrapping With Puppet edited by apenney @ http://reductivelabs.com/trac/puppet/wiki/BootstrappingWithPuppet
[2008/06/17 09:51:05] <gepetto> ::puppet:: Bootstrapping With Puppet edited by apenney @ http://reductivelabs.com/trac/puppet/wiki/BootstrappingWithPuppet
[2008/06/17 09:51:41] <Volcane> ah, first little ruby scriptlet written, this is a pretty good book
[2008/06/17 09:51:55] <Volcane> wrote soething to fetch a file off all our static servers, compare md5's etc
[2008/06/17 09:53:02] @ Quit: Demosthenes: "leaving"
[2008/06/17 09:53:05] <jamesturnbull> ashp: looks great
[2008/06/17 09:53:30] @ Quit: lutter: "Leaving."
[2008/06/17 09:54:19] @ glaw joined channel #puppet
[2008/06/17 09:54:21] <duritong> ashp: http://pastie.org/216494 copy netinfo is not really needed
[2008/06/17 09:54:32] <duritong> you can get the stuff from sysconfig
[2008/06/17 09:54:38] <duritong> as I have found out :)
[2008/06/17 09:54:41] @ Quit: leitgebj: Read error: 110 (Connection timed out)
[2008/06/17 09:56:12] <ashp> Oh, hmm
[2008/06/17 09:56:26] <ashp> nice. :)
[2008/06/17 09:56:50] @ Quit: ski98033: Read error: 113 (No route to host)
[2008/06/17 09:56:54] <ashp> I'm tempted to change it to use the PUPPET_SERVER=ip.address format, to get it to bootstrap
[2008/06/17 09:57:15] <ashp> I'll have to play with this one later
[2008/06/17 09:58:04] <duritong> i'm playing with xen since a whole day
[2008/06/17 09:58:17] <duritong> finding out that that network problems are related due to checksum problems
[2008/06/17 09:58:19] <duritong> argh
[2008/06/17 09:58:31] <duritong> we have bought too new hardware
[2008/06/17 09:58:32] <Volcane> chcksum problems?
[2008/06/17 09:58:33] <duritong> :-/
[2008/06/17 09:58:35] <Volcane> ah :(
[2008/06/17 09:58:41] <ashp> I haven't checked out xen as we bought into vmware
[2008/06/17 09:58:44] <ashp> as we have a lot of windows boxes too
[2008/06/17 09:59:05] <Disconnect> last place we used xen for both
[2008/06/17 09:59:06] <duritong> http://lists.xensource.com/archives/html/xen-users/2006-04/msg00032.html
[2008/06/17 09:59:30] <duritong> we have 1 windows box for sw-developers to test their webpass :P
[2008/06/17 09:59:33] <Disconnect> here they're into the vm management console for now, but its expensive enough that the only way to use it is to remote into the one server with the one license..
[2008/06/17 09:59:37] <duritong> webapps
[2008/06/17 10:00:23] <ashp> we have huge site licencing stuff for windows
[2008/06/17 10:01:01] <gepetto> ::puppet:: Bootstrapping With Puppet edited by jamtur01 @ http://reductivelabs.com/trac/puppet/wiki/BootstrappingWithPuppet (by james@lovedthanlost.net)
[2008/06/17 10:01:08] @ kambiz joined channel #puppet
[2008/06/17 10:01:20] <Disconnect> actually ended up using xen for windows video encoders - lost a couple percentage points of performance but uptime went through the roof - just fired up a clean image every morning, did the encodes, then wiped it and replaced it.
[2008/06/17 10:01:27] @ ski98033 joined channel #puppet
[2008/06/17 10:01:45] <Volcane> duritong: and ethtool -K eth0 tx off and sae with rx off doesnt help?
[2008/06/17 10:01:52] <ashp> All our media is encoded on helix on linux
[2008/06/17 10:01:59] @ hessml joined channel #puppet
[2008/06/17 10:02:05] <ashp> Which I found out yesterday isn't supported and doesn't work on rhel 5/centos 5
[2008/06/17 10:02:09] @ muer1 joined channel #puppet
[2008/06/17 10:02:20] <duritong> Volcane: it helps
[2008/06/17 10:02:28] <duritong> but not for bootstrapping (yet)
[2008/06/17 10:02:34] <Volcane> ah
[2008/06/17 10:02:57] <duritong> i think I have to pack ethtool into the intird and turn it off while booting
[2008/06/17 10:04:06] <Volcane> on the dom0?
[2008/06/17 10:04:29] <Volcane> or do you need to disable it on the dumU's?
[2008/06/17 10:04:33] <muer1> Disconnect: they might have fx2, but i'm using 3.0 from the link :)
[2008/06/17 10:05:15] <duritong> Volcane: you need to disable it on both :(
[2008/06/17 10:05:28] <Disconnect> ashp: we needed proper windows-media-encoder encodes for one client. (in addition to the 20-odd other formats we did for them with mencoder/ffmpeg/etc)
[2008/06/17 10:05:32] @ Ol_ joined channel #puppet
[2008/06/17 10:05:43] <duritong> and this is a problem with booting
[2008/06/17 10:05:49] <Volcane> duritong: sux :(
[2008/06/17 10:05:59] <Volcane> duritong: what OS?
[2008/06/17 10:06:32] <duritong> centos
[2008/06/17 10:06:48] <muer1> all our desktops are windows because 90% of the company is not very technical.
[2008/06/17 10:07:03] <Volcane> so a relatively clean way to fix it is to make a small rpm that installs an init script just before or just after network initialisation and does that for you?
[2008/06/17 10:07:09] <Volcane> pop that into your repos and into your kickstarts?
[2008/06/17 10:08:08] @ jY joined channel #puppet
[2008/06/17 10:08:18] <jY> is there anyway to put multiple hostnames in a node {} entry
[2008/06/17 10:09:10] <duritong> Volcane: well from where does anaconda get the rpm?
[2008/06/17 10:09:15] <duritong> I have to pack it into initrd
[2008/06/17 10:09:17] <ashp> jY, it can only match on one name I believe.
[2008/06/17 10:09:23] <jY> ok
[2008/06/17 10:09:40] <Volcane> duritong: ah yes, serious chicken and egg proble :(
[2008/06/17 10:10:03] <duritong> ay
[2008/06/17 10:10:19] <Volcane> not sure if you really can execute stuff in initrds?
[2008/06/17 10:10:56] <duritong> why not?
[2008/06/17 10:11:20] <muer1> correct, you can only use one hostname per node{} statement.
[2008/06/17 10:11:23] <Volcane> because they're just ramdisks where the kernel look for modules and such?
[2008/06/17 10:11:41] <duritong> but it also sets up networks mounts, stuff and so on
[2008/06/17 10:11:47] <muer1> jY: you can, however, use a class that multiple nodes are going to need that includes the other classes you want, or has the other statements to configure multiple nodes.
[2008/06/17 10:11:50] <duritong> and you can even pack a whole live system into an initrd
[2008/06/17 10:12:39] <Volcane> yeah i suppose
[2008/06/17 10:13:00] <muer1> oh my bad.
[2008/06/17 10:13:11] <muer1> you can use multiple nodes, apparently, from the wiki: http://reductivelabs.com/trac/puppet/wiki/LanguageTutorial#nodes
[2008/06/17 10:13:14] <muer1> I just never do :-)
[2008/06/17 10:13:45] <muer1> it might be a best practice or style guideline i followed using one node per node {}
[2008/06/17 10:13:54] <muer1> jY: sorry for adding to any confusion :-x
[2008/06/17 10:14:03] * muer1 drinks more coffee.
[2008/06/17 10:14:10] <jamesturnbull> jY: node hostname,hostname,hostname {}
[2008/06/17 10:14:36] <ashp> oh wow, you can?
[2008/06/17 10:14:45] <jamesturnbull> ashp: yes... :P
[2008/06/17 10:14:48] <ashp> Interesting, I'll never do it, but clever that you can :)
[2008/06/17 10:15:04] <ashp> besides I put my network details into the node so, I can't :)
[2008/06/17 10:15:35] <jamesturnbull> ashp: useful for a lot of similar hosts
[2008/06/17 10:16:00] <ashp> Maybe if I did my network different, but I have $ip =, $gateway = etc in the node statement
[2008/06/17 10:16:05] <muer1> jamesturnbull: thats what i use a 'node class' for: class nodetype::webserver inherits nodetype { } :)
[2008/06/17 10:16:57] @ brenton joined channel #puppet
[2008/06/17 10:18:47] @ Quit: muerr: Read error: 113 (No route to host)
[2008/06/17 10:18:53] @ shenson_not_here is now known as shenson
[2008/06/17 10:20:24] @ mikepea joined channel #puppet
[2008/06/17 10:20:50] <mikepea> .
[2008/06/17 10:22:10] @ hessml is now known as hessml|away
[2008/06/17 10:26:15] @ Quit: jvanzyl:
[2008/06/17 10:28:27] <ashp> So how do the rest of you handle your network configuration across the board if you use nodes
[2008/06/17 10:29:50] @ Quit: magnachef_: Read error: 110 (Connection timed out)
[2008/06/17 10:29:51] <Volcane> i dont config network through puppet
[2008/06/17 10:30:11] <Volcane> but dont have enough nodes where thats a problem yet
[2008/06/17 10:30:20] <Volcane> soethings I'd rather not break by accident :)
[2008/06/17 10:31:05] <ashp> I considered static DHCP but there's a definite agreement we don't want it here. :)
[2008/06/17 10:31:35] <jamesturnbull> ashp: I don't use internal nodes generally - LDAP or external nodes
[2008/06/17 10:31:40] <Volcane> yeah, dhcp i only use during bootstrap
[2008/06/17 10:31:50] <Volcane> kickstart configures static network info
[2008/06/17 10:31:56] <muer1> fyi, ftp.mozilla.org is overloaded for 3.0 i imagine, but mirrors can be found from: http://download.mozilla.org/?product=firefox-3.0&os=win&lang=en-US (change os=linux or os=osx if required)
[2008/06/17 10:31:58] <jamesturnbull> Volcane: it'd be nice to have a few more network config types
[2008/06/17 10:31:58] <muer1> :)
[2008/06/17 10:32:14] <Volcane> jamesturnbull: yes, i will soon need ot do bonds, and afaik the redhat one doesnt do that yet
[2008/06/17 10:32:42] <ashp> I did consider just having kickstart do the network stuff
[2008/06/17 10:32:49] <ashp> but it's nice that when we resize subnets, like we did earlier
[2008/06/17 10:32:57] <ashp> I can just change puppet and let the changes distribute out
[2008/06/17 10:33:11] * Volcane has simple networks on purpose
[2008/06/17 10:33:11] <muer1> If anyone needs to configure red hat system network interfaces, check out my network module, it'll do single interface, or bonded interfaces, and will do trunk and carp interfaces on openbsd.
[2008/06/17 10:33:18] <jamesturnbull> Volcane: log a feature ticket
[2008/06/17 10:33:20] @ Quit: hessml|away: "Leaving..."
[2008/06/17 10:33:22] <muer1> http://github.com/jtimberman/puppet/tree/master
[2008/06/17 10:33:34] <jamesturnbull> Volcane: or muer1's modle seemingly... :P
[2008/06/17 10:33:39] <Volcane> jamesturnbull: nods, well i also know the redhat interface type code and I'll rather not use it :P
[2008/06/17 10:33:44] @ muer1 is now known as muerr
[2008/06/17 10:33:55] <muerr> jamesturnbull: :)
[2008/06/17 10:34:00] <Volcane> unless its been *completely* redone since it was introduced
[2008/06/17 10:34:03] <jamesturnbull> Volcane: yeah I know what you mean
[2008/06/17 10:34:08] <jamesturnbull> Volcane: and no it hasn't
[2008/06/17 10:34:30] <jamesturnbull> alright night all - some sleep needed
[2008/06/17 10:34:40] <Volcane> c'ya jamesturnbull
[2008/06/17 10:34:41] <muerr> my new employer might see more value in me extending the network interface type within puppet itself than my current employer, but i'm not sure yet - depends on how much their clients use puppet for networks
[2008/06/17 10:35:15] @ Quit: BMDan: "Leaving."
[2008/06/17 10:35:41] <muerr> s/networks/network interfaces/
[2008/06/17 10:35:53] @ leitgebj joined channel #puppet
[2008/06/17 10:37:11] <kjetilho> muerr: why not rely on DHCP?
[2008/06/17 10:37:38] @ Quit: glaw: "I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to nor
[2008/06/17 10:38:10] <muerr> kjetilho: because we need static IPs.
[2008/06/17 10:38:52] <Volcane> kjetilho: one less thing to make highly available
[2008/06/17 10:38:56] <muerr> and entering mac address -> ip address specifics in dhcp is a pain.
[2008/06/17 10:39:40] <muerr> we also do a default-deny firewall policy, so specific IPs have to be allowed for specific ports, and thats a complete nightmare with dhcp.
[2008/06/17 10:40:23] <ashp> relying on DHCP is a real pain, I just don't trust it
[2008/06/17 10:40:33] <ashp> it's fine for desktops and stuff, but for servers..
[2008/06/17 10:40:50] <Volcane> though my dhcp server is on y linux-ha cluster, i ake no effort to share leases files etc, cos i only use it during bootstrap
[2008/06/17 10:40:50] <muerr> ashp: that too :)
[2008/06/17 10:40:59] <Volcane> its mostly on linux-ha cluster so i can reinstall the dhcpserver too
[2008/06/17 10:41:04] <muerr> we /do/ use dhcp for our build vlan.
[2008/06/17 10:41:08] <ashp> yeah, same here
[2008/06/17 10:41:14] <muerr> kind of required there for pxe boot and all :-)
[2008/06/17 10:42:08] * muerr drives to work.
[2008/06/17 10:43:44] @ flakrat joined channel #puppet
[2008/06/17 10:43:58] @ hessml joined channel #puppet
[2008/06/17 10:44:10] @ muerr left channel #puppet ()
[2008/06/17 10:44:16] <kjetilho> relying on DHCP is a *real* win when you need to renumber your network :)
[2008/06/17 10:44:38] <ashp> well, that's why i statically assign the network via puppet
[2008/06/17 10:44:40] * Volcane renumbers on his firewalls :P
[2008/06/17 10:44:47] <ashp> it's like dhcp, except permanent unless changed :D
[2008/06/17 10:45:08] <kajtzu> renumbering is a pita
[2008/06/17 10:45:17] <kjetilho> perhaps more relevant: moving servers physically. shut it down, update DHCP config, move it, switch it on
[2008/06/17 10:46:19] <ashp> it's nice, but i hate relying on another device to get the network config
[2008/06/17 10:46:20] <hessml> i've been searching for a document on how to debug your .pp file; is there such a thing?
[2008/06/17 10:46:28] <kjetilho> Volcane: oh, and if you assign addresses statically, making dhcpd HA is trivial
[2008/06/17 10:46:53] <hessml> btw, good morning :-)
[2008/06/17 10:47:18] <kjetilho> dynamic addresses are painful for Unix hosts, so that's a non-starter IMHO.
[2008/06/17 10:47:20] @ jvanzyl joined channel #puppet
[2008/06/17 10:48:01] @ Quit: bss: "using sirc version 2.211+KSIRC/1.3.12"
[2008/06/17 10:48:04] @ magnachef__ joined channel #puppet
[2008/06/17 10:50:02] <jamesturnbull> hessml: wiki:VersionControlPuppet has some information
[2008/06/17 10:50:05] <gepetto> jamesturnbull: hessml: wiki:VersionControlPuppet is http://reductivelabs.com/trac/puppet/wiki/VersionControlPuppet
[2008/06/17 10:54:48] @ brscott joined channel #puppet
[2008/06/17 10:56:10] <hessml> that looks like a way to find syntax errors in your code, which sounds good; but how about logic errors? does the client keep a log of actions? is there a place to log print statements to? Is there a prefered way of spewing diagnostics from your code in puppet?
[2008/06/17 10:56:58] <lak> all actions are logged to syslog by default
[2008/06/17 10:57:16] <lak> and you can do --evaltrace to watch it traverse the resource graph
[2008/06/17 10:57:35] <lak> and you can look at generated graphs of the resources and their relationships in .dot format, with omnigraffle or visio
[2008/06/17 10:57:40] <kjetilho> is there a way to catch stuff like double assignments to variables?
[2008/06/17 10:58:47] @ hiffy_ is now known as hiffy
[2008/06/17 10:59:27] <kjetilho> e.g. $foo = "bar"; case $lsbdistid { "Debian": { $foo = "zot" } }
[2008/06/17 10:59:35] @ brscott left channel #puppet ()
[2008/06/17 11:02:19] <Volcane> ok, debian ppl, how to get a doainname onto my debian box other than putting fqdn into /etc/hostname (which the docs says not to do)
[2008/06/17 11:04:09] <kjetilho> ignore the docs.
[2008/06/17 11:04:41] <kjetilho> not using FQDN is a recipe for disaster IMHO
[2008/06/17 11:04:53] <hessml> so 3 debugging techniques: 1) hook svn to catch ruby syntax errors, 2) inspect syslog (and log to syslog), 3) dump resource graph and inspect in omnigraffle or visio
[2008/06/17 11:04:57] <ashp> yeah, I always put my fqdn into /etc/hosts
[2008/06/17 11:05:02] <hessml> any more?
[2008/06/17 11:05:03] <ashp> whatever doc said not to is wrong and stupid :/
[2008/06/17 11:05:26] <bda> dnsdomainname seems to use /etc/hosts.
[2008/06/17 11:05:31] <bda> So it can't be *that* stupid and wrong.
[2008/06/17 11:05:53] <Volcane> http://www.debian.org/doc/manuals/reference/ch-gateway.en.html#s-name-host
[2008/06/17 11:06:06] <Volcane> like, there is no more authoritive docs than that
[2008/06/17 11:06:52] <zipkid> hello.... is the 'latest' option for package with rug as provider supposed to ever have worked?
[2008/06/17 11:08:10] <Volcane> no the docs is right
[2008/06/17 11:08:21] <Volcane> the file /etc/hostname should just have hostname
[2008/06/17 11:08:23] <lak> zipkid: not sure; not many people use the rug provider
[2008/06/17 11:08:31] <Volcane> then /etc/hosts should have x.x.x.x fqdn hostnae
[2008/06/17 11:08:38] <Volcane> and thats how it works
[2008/06/17 11:08:53] @ Quit: magnachef__: Read error: 110 (Connection timed out)
[2008/06/17 11:09:06] @ magnachef_ joined channel #puppet
[2008/06/17 11:09:08] <zipkid> lok i have a new nagios in a self-made yum source and puppet fails to update
[2008/06/17 11:09:27] <Volcane> zipkid: yum clean all
[2008/06/17 11:09:33] <zipkid> the code says rug install can be used to update but it does not
[2008/06/17 11:09:34] <Volcane> else it only updates now and then
[2008/06/17 11:10:01] <zipkid> Volcane: i can update using the rug commands manualy....
[2008/06/17 11:10:13] <ashp> Volcane: Oh I thought you meant the fqdn in hosts, not hostname :)
[2008/06/17 11:10:26] <mdray> hey hey
[2008/06/17 11:10:49] <mdray> is there a handy rhel rpm for yum-priorities
[2008/06/17 11:10:57] <mdray> i feel like there should be but i can't find one
[2008/06/17 11:11:28] <Volcane> yum-priorities.noarch 1.0.4-3.el5.centos.2 centos-base
[2008/06/17 11:11:39] <Volcane> in base, should be in base for rhel too
[2008/06/17 11:12:02] @ muerr joined channel #puppet
[2008/06/17 11:12:28] <kjetilho> unqualified hostname probably works for small installations
[2008/06/17 11:12:46] <Volcane> kjetilho: the *right* way is to not put fqdn in /etc/hostname and then do fqdn in /etc/hosts
[2008/06/17 11:12:54] <Volcane> kjetilho: thats how you get a fqdn
[2008/06/17 11:13:09] <kjetilho> "uname -n" should always return FQDN
[2008/06/17 11:13:10] <ashp> lak: why not just set the facter timeout to 5 only on AIX
[2008/06/17 11:13:14] <ashp> lak: problem solved!
[2008/06/17 11:13:24] <kjetilho> there is no other portable way of getting a FQDN on Unix
[2008/06/17 11:13:36] <Volcane> kjetilho: not according to the man page :P
[2008/06/17 11:13:37] <kjetilho> short of writing your own Perl script to do it
[2008/06/17 11:14:00] <Volcane> anyway not on shitty debian
[2008/06/17 11:14:05] <kjetilho> if you have a homogoneous network with only Debian, feel free.
[2008/06/17 11:14:07] <Volcane> on redhat machines nodename -n gives fqdn
[2008/06/17 11:14:27] <Volcane> i mean uname -n
[2008/06/17 11:14:37] <Volcane> -n is for nodename, which should just be the horname surely?
[2008/06/17 11:15:02] <kjetilho> uname -n returns the hostname, yes. which should be FQDN to keep sane
[2008/06/17 11:15:26] <Volcane> well, to keep sane and to be true to docs is apparently 2 things :)
[2008/06/17 11:16:23] @ Quit: jvanzyl: Read error: 110 (Connection timed out)
[2008/06/17 11:17:02] <lak> ashp: you think so, until the next one crops up :/
[2008/06/17 11:20:05] <ashp> lak: It's the edge cases that always ruin nice elegant clean code :(
[2008/06/17 11:20:16] <lak> and *nix is all edge cases
[2008/06/17 11:20:42] <ashp> I do wish it was possible to just shove all the people involved in a room and say
[2008/06/17 11:20:47] <ashp> PICK ONE WAY TO DO IT AND STOP IT ALL
[2008/06/17 11:21:00] <ashp> 2008 and we still can't standardise where to put things on the filesystem
[2008/06/17 11:21:08] <gepetto> ::puppet:: Frequently Asked Questions edited by hessml @ http://reductivelabs.com/trac/puppet/wiki/FrequentlyAskedQuestions (by martinhess@mac.com)
[2008/06/17 11:21:13] <kjetilho> yeah, LSB was supposed to fix that
[2008/06/17 11:21:15] <ashp> I'm tired of /etc/apache2/, /etc/httpd/ /etc/apache/, /usr/local/apache/etc/
[2008/06/17 11:21:29] <Volcane> heh
[2008/06/17 11:21:31] <ashp> well it's the rule of the bike shed
[2008/06/17 11:21:38] <ashp> the less important a decision is, the more people will invest in it
[2008/06/17 11:21:57] <ashp> want to change the scheduler for linux? You'll get two comments. Want to rename a file, 20,000 people will write diatribes about it
[2008/06/17 11:23:05] <kajtzu> I'm perfectly happy with /etc/httpd to be honest ;)
[2008/06/17 11:23:15] <ashp> me too, I just want everyone to standardize and stop messing around
[2008/06/17 11:23:19] <ashp> i don
[2008/06/17 11:23:26] <ashp> i don't really care what things are called, as long as it's always the same
[2008/06/17 11:23:42] <mdray> volcane: aha, RH don't build yum-priorities in their standard yum-utils RPM :~(
[2008/06/17 11:23:44] <ashp> I'm lucky we're just moving towards a single distribution
[2008/06/17 11:23:50] <Volcane> mdray: ah sux
[2008/06/17 11:23:59] @ Quit: roald: Read error: 104 (Connection reset by peer)
[2008/06/17 11:24:01] <ashp> because I couldn't handle supporting four different package managers/package formats and all the rest it entails.
[2008/06/17 11:26:01] @ martha joined channel #puppet
[2008/06/17 11:26:37] <Volcane> get rid of all the debian and debian derivatives and 95% will be the same :P
[2008/06/17 11:27:22] <ashp> :)
[2008/06/17 11:27:47] @ AimanA joined channel #puppet
[2008/06/17 11:28:13] <Volcane> then you're left with gentoo, and convincing minions who likes gentoo to change their mind is easy, thats what 486 desktops are for
[2008/06/17 11:36:19] <muerr> ashp: don't forget /opt/httpd/etc, /opt/IBMHTTPD/etc, /opt/apache/etc,
[2008/06/17 11:36:46] <Volcane> heh
[2008/06/17 11:36:53] <muerr> thx ibm httpd, which is apache :-x
[2008/06/17 11:37:24] <muerr> but thats why you set up your http configs with a define and have a parameter "httpd_config_dir"
[2008/06/17 11:45:22] @ Quit: erikh: Client Quit
[2008/06/17 11:45:34] @ erikh joined channel #puppet
[2008/06/17 11:49:42] @ Quit: Wicky656: Remote closed the connection
[2008/06/17 11:51:11] @ Wicky656 joined channel #puppet
[2008/06/17 11:55:41] <hessml> any idea why puppet creates the directory /var/log/puppet/ but logs to var/log/syslog?
[2008/06/17 11:56:01] <hessml> nothing seems to show up in /var/log/puppet/
[2008/06/17 12:01:40] @ Quit: zipkid: Read error: 110 (Connection timed out)
[2008/06/17 12:03:11] <lak> hessml: well, i figured people would want the ability to log to a file, and there's no way right now to say "i might use this directory" without always creating it
[2008/06/17 12:04:02] <hessml> ah
[2008/06/17 12:04:09] <ashp> If the developers don't start documenting their hideous php mess i'm going to punch someone in the head
[2008/06/17 12:04:34] <ashp> to make a database connection I have to fix odbc.ini, freetds.conf, a credential file that's hashed with a massive random name so I can't find it, some apache configuration..
[2008/06/17 12:04:53] <ashp> none of this is documented or listed anywhere of course :)
[2008/06/17 12:05:16] <erikh> what app is this?
[2008/06/17 12:05:32] <erikh> oh; probably internal
[2008/06/17 12:05:37] <erikh> that's par :)
[2008/06/17 12:06:12] <erikh> heh, I worked at one place where our 'app' couldn't be installed on new machines without rsync
[2008/06/17 12:06:51] <erikh> (a Fortune 100 company)
[2008/06/17 12:10:29] <ashp> it's internal, yeah
[2008/06/17 12:10:33] <ashp> well here's my favourite bit
[2008/06/17 12:10:44] <ashp> they use this library they wrote to hide username/passwords from the scripts
[2008/06/17 12:10:58] <ashp> so it requests them via this library, which then looks in /share/credentials/ (which is available all over the place)
[2008/06/17 12:11:19] <ashp> the files are called things like afc905b994fc861cc20bd1b5b9af0531c459ed4fc9af64987ec66140ce25566aa0c276c272c53f7e9bcab99f0a0d9f64c01ad94bdc27c61d9cd976c05dd191eb
[2008/06/17 12:11:29] <ashp> But if you know roughly what kind of data you might want (like say, SFS data)
[2008/06/17 12:11:35] <ashp> you just grep the entire damn directory for sfs
[2008/06/17 12:11:43] <ashp> and find a file with the username of SFSDATA
[2008/06/17 12:11:51] <ashp> and it doesn't take a genius to unravel this insane security scheme
[2008/06/17 12:12:19] <muerr> lol
[2008/06/17 12:13:29] <Volcane> thats quite comon
[2008/06/17 12:13:35] <Volcane> seen small variations of that everywhere
[2008/06/17 12:13:45] <ashp> i suppose it keeps them out of the scripts
[2008/06/17 12:13:52] <ashp> but we have no documented way to find a credential
[2008/06/17 12:14:00] <ashp> except one developer has a script in his home directory that I know can do it
[2008/06/17 12:15:15] <Volcane> I've seen <service>,<username>,<password>,<callinguser> in a huge txt file
[2008/06/17 12:15:26] <Volcane> so app can ask for access to mysql
[2008/06/17 12:15:30] <Volcane> with callinguser=app
[2008/06/17 12:15:34] <Volcane> service=mysql
[2008/06/17 12:15:41] <Volcane> and it would get username/password back
[2008/06/17 12:15:43] <Volcane> yuk
[2008/06/17 12:15:49] <erikh> ashp: are those files marshalled sessions?
[2008/06/17 12:15:51] <ashp> the odbc/freetds stuff isn't our fault as they use mssql databases
[2008/06/17 12:15:57] <erikh> naming them that way would certainly make sense
[2008/06/17 12:15:57] <ashp> erikh: I.. I don't even know what that means, sorry
[2008/06/17 12:16:12] <ashp> i can't unravel the viper nest of php to find out how they are created
[2008/06/17 12:16:14] <Volcane> they look like md5's that are concat'd
[2008/06/17 12:16:18] <ashp> and the developers don't answer questions
[2008/06/17 12:16:21] <ashp> they are 'too busy' for that
[2008/06/17 12:16:25] <erikh> Volcane: or a SHA-512 sum
[2008/06/17 12:17:03] <Volcane> yeah
[2008/06/17 12:17:39] <erikh> ashp: if it's anything like the places I've programmed for, they probably are :) use your wits and give them some incentive to talk... start suggesting to them how they write their application
[2008/06/17 12:17:44] @ Quit: hessml: "Leaving..."
[2008/06/17 12:18:09] <erikh> that'll get any developer to spend an hour spewing condescending drivel at you all justifying their design decisions, which will be detailed clearly
[2008/06/17 12:18:45] <ashp> they aren't really that busy, we work for a university
[2008/06/17 12:18:49] <ashp> they just don't want to admit they wrote a monster
[2008/06/17 12:18:57] <ashp> a lot of this is going away as we move to a cms system
[2008/06/17 12:19:11] <ashp> no more homegrown templating system that requires you to pre/post append files via apache <directory> stanzas!
[2008/06/17 12:19:13] <erikh> heh. yeah, definitely tell them how to write their application if you want to work with it, then.
[2008/06/17 12:19:45] <ashp> They got a new development manager who is horrified
[2008/06/17 12:19:53] <ashp> they don't even have a svn tree or anything
[2008/06/17 12:20:04] * Volcane goes
[2008/06/17 12:20:11] <ashp> it's all just random people with gentoo doing random stuff all over production (they've been removed from production now)
[2008/06/17 12:20:59] <erikh> yeah, the company I described above... I had to call a 2 hour meeting to describe a version control system to the other devs after spending about 9 months pushing for it
[2008/06/17 12:21:32] <erikh> not svn, or git, just version control in general. `cp` was the preferred version control before then
[2008/06/17 12:21:51] <erikh> and they refused to use branching for any reason. didn't see the point in it.
[2008/06/17 12:22:05] <ashp> haha, i'm glad everyone has to fight the same battles
[2008/06/17 12:22:08] <ashp> even if i'm dismayed
[2008/06/17 12:22:16] <ashp> they must hate the new guy, he's all about 'enterprise'
[2008/06/17 12:22:23] <ashp> the latest thing is this enterprise service bus
[2008/06/17 12:22:33] <ashp> he also tore out like 5 years of work on 'webtools', this monster ldap management horror
[2008/06/17 12:22:38] <ashp> and is replacing it with sun identity manager
[2008/06/17 12:22:40] <erikh> well, when 80% of your IT workforce is in India you lose a significant chunk of experience when you start handing out pink slips
[2008/06/17 12:22:55] <ashp> webtools is black magic at the finest and NOBODY except the guy who wrote it has any idea what it does, to where, or how
[2008/06/17 12:23:01] <muerr> heh heh.
[2008/06/17 12:23:13] <ashp> the guy who wrote it has something like 200 scripts that run as him on production
[2008/06/17 12:23:16] <ashp> to do 'things'
[2008/06/17 12:23:24] <muerr> Contrast the above conversation with my experience with developers at IBM... one word: Ugh.
[2008/06/17 12:23:38] <ashp> it does all kinds of cross checking between parts of ldap and some mysql stuff and deletes things that don't conform, it works but is terrifying
[2008/06/17 12:23:52] <erikh> well, lots of dev shops have a tool guy
[2008/06/17 12:24:11] <barnum> most shops have a few 'tools'
[2008/06/17 12:24:19] <erikh> someone to get the shit done when all the other developers are holding meetings to figure out how to do i