Friday, 2008-06-13

[2008/06/13 00:01:03] @ plathrop joined channel #puppet
[2008/06/13 00:20:40] @ Quit: a-priori:
[2008/06/13 00:46:05] @ martha left channel #puppet ()
[2008/06/13 01:01:13] @ ski98033 joined channel #puppet
[2008/06/13 01:05:23] @ shadoi joined channel #puppet
[2008/06/13 01:34:05] <nwp> is there a variable set in puppet config to the IP of the node we're configuring?
[2008/06/13 01:34:14] <nwp> could there be even?
[2008/06/13 01:34:34] <nwp> it would be a bloody good idea, anyway
[2008/06/13 01:35:27] <nwp> actually, is the node name even available for use within the config?
[2008/06/13 01:36:25] <shadoi> ipaddress
[2008/06/13 01:36:43] <shadoi> anything in facter is available as a variable
[2008/06/13 01:40:27] @ Quit: jvanzyl:
[2008/06/13 01:41:36] @ ski98033 left channel #puppet ()
[2008/06/13 01:41:55] @ bartc_ joined channel #puppet
[2008/06/13 01:42:20] <nwp> shadoi: right, wasn't sure it would be from facter
[2008/06/13 01:43:05] <nwp> ah, no
[2008/06/13 01:43:09] <nwp> that gets the wrong one
[2008/06/13 01:43:20] <shadoi> wrong one?
[2008/06/13 01:43:46] <shadoi> it gets the facts for the node that the configuration is being compiled for.
[2008/06/13 01:43:49] <nwp> that was the point - I need the IP address of the node referred to by the node name
[2008/06/13 01:44:00] <nwp> facter doesn't know which IP to return
[2008/06/13 01:44:08] <nwp> the machine has several
[2008/06/13 01:44:17] <nwp> I need the one which puppetmaster is talking to
[2008/06/13 01:44:38] <nwp> (which is the one that you would get if you did a dns lookup on the node name)
[2008/06/13 01:44:56] <shadoi> I'm not sure if the newer version of facter does it correctly, but I know they added alias support
[2008/06/13 01:45:08] <nwp> I don't see how it could
[2008/06/13 01:45:27] <shadoi> I think it currently just gives you eth0
[2008/06/13 01:45:35] <nwp> without being provided with extra information
[2008/06/13 01:45:46] <nwp> if the node name itself available in a variable?
[2008/06/13 01:45:55] <nwp> I can look that up if I must
[2008/06/13 01:46:11] <nwp> sorry, "is the node name..."
[2008/06/13 01:47:13] <shadoi> hostname
[2008/06/13 01:47:27] <shadoi> just run facter on a client to see everything you can get
[2008/06/13 01:47:38] <shadoi> you can create a custom fact to figure it out very easily
[2008/06/13 01:47:42] <shadoi> wiki:AddingFacts
[2008/06/13 01:47:44] <gepetto> shadoi: wiki: wiki:AddingFacts is http://reductivelabs.com/trac/puppet/wiki/AddingFacts
[2008/06/13 01:49:20] <nwp> I think hostname can be made to work with some extra hackery
[2008/06/13 01:49:45] <nwp> but really it would still be good to have the node name rather than anything actually calculated
[2008/06/13 01:50:19] <nwp> the issue is that we have vservers running on the puppet clients
[2008/06/13 01:50:45] <nwp> and they are not really compartmentalized enough, so things have a tendency to bind to interfaces that they shouldn't
[2008/06/13 01:50:58] <shadoi> *nod*
[2008/06/13 01:50:58] <nwp> so e.g. zabbix-agent needs to be told in its config which IP to bind to
[2008/06/13 01:51:18] <nwp> which should be the IP associated with the nodename in the puppet config
[2008/06/13 01:51:24] <nwp> but facter can't know what that is
[2008/06/13 01:51:38] <shadoi> yeah
[2008/06/13 01:51:41] <nwp> although like I said, I might be able to find it indirectly via the facter hostname
[2008/06/13 01:52:13] <shadoi> custom facts will probably be your best bet.
[2008/06/13 01:52:28] <nwp> yep, for now
[2008/06/13 01:53:02] <nwp> ugly though
[2008/06/13 01:53:25] <shadoi> really?
[2008/06/13 01:53:48] @ Quit: bartc: Read error: 110 (Connection timed out)
[2008/06/13 01:54:02] <shadoi> file a bug if you think there's a sane fix for it.
[2008/06/13 01:56:58] @ morfoh_ joined channel #puppet
[2008/06/13 01:57:31] @ bio____ joined channel #puppet
[2008/06/13 01:59:07] @ Quit: plathrop: Remote closed the connection
[2008/06/13 02:09:12] @ Quit: morfoh: Read error: 110 (Connection timed out)
[2008/06/13 02:09:14] @ Quit: bio___: Read error: 110 (Connection timed out)
[2008/06/13 02:19:19] @ aymerick joined channel #puppet
[2008/06/13 02:24:14] <nwp> shadoi: thing is I'm not sure there is a completely sane fix
[2008/06/13 02:24:37] <nwp> other than "dump vserver and use xen" or something like that
[2008/06/13 02:41:02] @ sparanjape joined channel #puppet
[2008/06/13 02:48:06] @ jsgotangco joined channel #puppet
[2008/06/13 03:02:59] @ Quit: mistat: "www.communitynet.net.au"
[2008/06/13 03:08:29] @ Quit: MrProper_: "Leaving"
[2008/06/13 03:08:34] @ DerekW joined channel #puppet
[2008/06/13 03:11:37] @ \ask_ joined channel #puppet
[2008/06/13 03:37:20] @ Quit: kolla: Remote closed the connection
[2008/06/13 03:40:43] @ Quit: shadoi: "leaving"
[2008/06/13 04:01:14] <fujin> nwp: do the interfaces facts not suffice for working out which IP to bind to?
[2008/06/13 04:01:41] <fujin> ipmess et al
[2008/06/13 04:01:57] @ aymerick left channel #puppet ()
[2008/06/13 04:03:27] <fujin> anyone running facter master HEAD?
[2008/06/13 04:03:41] @ marcoecc joined channel #puppet
[2008/06/13 04:05:27] <fujin> jamesturnbull: who's the maintainer for facter now? is it still @ reductivelabs.com Git?
[2008/06/13 04:05:48] <jamesturnbull> fujin: me
[2008/06/13 04:05:57] <jamesturnbull> fujin: master is still at rl.com
[2008/06/13 04:07:31] <fujin> jamesturnbull: is your github facter up to date?
[2008/06/13 04:07:50] <jamesturnbull> fujin: yeah I think so
[2008/06/13 04:08:08] <fujin> cool
[2008/06/13 04:08:11] <fujin> ah yes, that works
[2008/06/13 04:08:15] <fujin> fetched, checked out and reinstalled
[2008/06/13 04:08:42] <fujin> lukes was out of date
[2008/06/13 04:09:31] @ Quit: jsgotangco: Read error: 104 (Connection reset by peer)
[2008/06/13 04:13:13] @ Quit: rcoup:
[2008/06/13 04:15:41] <jamesturnbull> fujin: really?
[2008/06/13 04:15:54] <jamesturnbull> fujin: which one of luke's?
[2008/06/13 04:16:12] <fujin> lak/master @github
[2008/06/13 04:17:39] <jamesturnbull> fujin: ah
[2008/06/13 04:19:06] * Volcane 's puppetmaster really is becoming rather random in behavious
[2008/06/13 04:19:13] <Volcane> run a client, error cant find some class/define
[2008/06/13 04:19:17] <Volcane> run it again, compiles fine
[2008/06/13 04:22:32] @ Quit: marcoecc: "Quitting"
[2008/06/13 04:24:58] @ jsgotangco joined channel #puppet
[2008/06/13 04:29:35] @ mdray joined channel #puppet
[2008/06/13 04:29:59] @ Quit: jsgotangco: "This computer has gone to sleep"
[2008/06/13 04:32:53] @ jsgotangco joined channel #puppet
[2008/06/13 04:44:39] @ Quit: jsgotangco: "Ciao"
[2008/06/13 04:53:00] @ strerror_work joined channel #puppet
[2008/06/13 05:07:40] @ dsch04 joined channel #puppet
[2008/06/13 05:21:18] @ Innocenti joined channel #puppet
[2008/06/13 05:33:39] @ Quit: johnf: Read error: 110 (Connection timed out)
[2008/06/13 05:44:26] @ johano joined channel #puppet
[2008/06/13 06:02:32] @ kolla joined channel #puppet
[2008/06/13 06:07:45] <nwp> fujin: no, I don't think so
[2008/06/13 06:08:05] <nwp> not without some fiddly reorganisation of the type which using puppet is supposed to rid you
[2008/06/13 06:11:15] <fujin> jamesturnbull: odd. I can't seem to call `facter ipaddress_eth0` like I should.
[2008/06/13 06:11:28] <fujin> It spits it out when I run `facter`, but refues to output anything when I specify it
[2008/06/13 06:11:37] @ gileswork joined channel #puppet
[2008/06/13 06:11:38] <fujin> oddly enough, `facter ipaddress` works
[2008/06/13 06:11:54] <fujin> this is your HEAD @ github/jamtur01/master
[2008/06/13 06:12:29] <fujin> pastie:
[2008/06/13 06:12:46] <pastie> http://pastie.org/214385 by fujin.
[2008/06/13 06:12:51] <fujin> jamesturnbull: ^
[2008/06/13 06:13:36] <jamesturnbull> fujin: that won't work
[2008/06/13 06:13:47] <jamesturnbull> fujin: the fact isn't pre-fetched like that
[2008/06/13 06:13:57] <jamesturnbull> fujin: it is constructed
[2008/06/13 06:13:59] <jamesturnbull> fujin: oh
[2008/06/13 06:14:05] <jamesturnbull> fujin: hmmmm maybe that should work
[2008/06/13 06:16:50] <Volcane> 'facter ipaddress_eth0' defo used to work
[2008/06/13 06:24:31] <jamesturnbull> Volcane: really?
[2008/06/13 06:24:37] * jamesturnbull looks confused
[2008/06/13 06:29:44] <Volcane> % facter ipaddress_eth0
[2008/06/13 06:29:44] <Volcane> 78.47.195.198
[2008/06/13 06:29:45] <Volcane> :)
[2008/06/13 06:29:59] <Volcane> centos 5.1 facter-1.3.8
[2008/06/13 06:30:04] @ jgonzalez joined channel #puppet
[2008/06/13 06:31:46] @ Quit: Kindy: Read error: 104 (Connection reset by peer)
[2008/06/13 06:31:52] @ Kindred joined channel #puppet
[2008/06/13 06:44:04] * Volcane once again wishes the puppet book had an index
[2008/06/13 06:44:17] <Volcane> jamesturnbull: how about making a downloadable pdf index that we can print out? :)
[2008/06/13 06:45:42] @ Quit: \ask_:
[2008/06/13 06:49:49] <jamesturnbull> Volcane: indexing is actuaqlly quite hard
[2008/06/13 06:50:24] <f3ew> TeX
[2008/06/13 06:51:00] <Volcane> so is finding the section dealing with autosigning :P
[2008/06/13 06:51:05] * Volcane gave up and searched the slow wiki
[2008/06/13 06:52:35] @ johnf joined channel #puppet
[2008/06/13 06:53:20] @ morfoh_ is now known as morfoh
[2008/06/13 06:58:14] @ babysimon joined channel #puppet
[2008/06/13 06:59:31] <babysimon> Am I correct in thinking at a machine cannot be both a client of a puppet master, and its own separate master?
[2008/06/13 06:59:55] <babysimon> The use case is my workstation, which is a client of the real master but I also want it to be its own master for development.
[2008/06/13 07:00:29] <babysimon> i.e. it's the master for some clients but not itself.
[2008/06/13 07:02:02] <Volcane> babysimon: you can, just run it --server=127.0.0.1
[2008/06/13 07:02:23] <Volcane> babysimon: but rather check out environments and make a dev environment for your puppet dev which is hosted on the same master but restricted to your desktop
[2008/06/13 07:02:30] <babysimon> I think there are problemns with the CA though
[2008/06/13 07:02:35] <Volcane> much less hackery
[2008/06/13 07:02:50] <Volcane> babysimon: ah, true that, use environments :)
[2008/06/13 07:05:04] <babysimon> the wiki page says environments need 0.24. Does that mean the master or the master+client?
[2008/06/13 07:05:18] <Volcane> both
[2008/06/13 07:05:22] <babysimon> (I have lots of clients running etch and so having 0.20)
[2008/06/13 07:05:26] <babysimon> ah, right
[2008/06/13 07:05:52] <duritong> well 0.20 ist heavily outdated, you should usebackports for etch
[2008/06/13 07:06:02] <Volcane> sux :(
[2008/06/13 07:06:12] <Volcane> 0.20 thats like the dark ages
[2008/06/13 07:06:23] <babysimon> :(
[2008/06/13 07:08:39] <babysimon> But there is no backport on backports.org - is there one available or do I have to make my own?
[2008/06/13 07:10:04] <Volcane> i pull in from testing
[2008/06/13 07:10:12] <Volcane> it works on 3.0 machines
[2008/06/13 07:10:41] <Volcane> like use the pinning features so it only pulls in puppet and facter from the testing branch
[2008/06/13 07:10:44] <Volcane> but
[2008/06/13 07:10:45] <Volcane> you also then dont use odules
[2008/06/13 07:10:54] <Volcane> and environments isnt that hot with the normal fileserver
[2008/06/13 07:11:03] <Volcane> so its probably not *the* solution for u
[2008/06/13 07:11:13] <babysimon> hmm, OK
[2008/06/13 07:11:36] <Volcane> if u just wanted to have different sets of classes for your dev then environments would work a charm
[2008/06/13 07:11:42] <duritong> you can use the sid one
[2008/06/13 07:11:52] <duritong> there was recently a discussion on the lists about that
[2008/06/13 07:13:23] <babysimon> I'm sure when I tried that last it wanted to pull in a new libc. Something like puppet -> libshadow-ruby -> libc
[2008/06/13 07:13:38] <Volcane> do you need to manage passwords?
[2008/06/13 07:14:04] <babysimon> that would be nice too! we mostly have ldap, but some machines in a rack for which passwords would be good
[2008/06/13 07:14:19] <Volcane> if you dont need password management now, you dont need the shadow thing
[2008/06/13 07:14:43] <Volcane> it will automagically manage passwords where its available (like your etch machines) and ignore iton the rest
[2008/06/13 07:15:45] <babysimon> how do you avoid the dependency though? puppet on lenny depends on libshadow-ruby1.8?
[2008/06/13 07:16:19] <Volcane> does it? I've not seen that
[2008/06/13 07:16:31] <babysimon> http://packages.debian.org/lenny/puppet
[2008/06/13 07:16:57] <babysimon> maybe it didn't used to and you picked up an older version?
[2008/06/13 07:16:57] <Volcane> maybe i was blind :)
[2008/06/13 07:17:20] <Volcane> i have libshadow on my debian 3.1 machine
[2008/06/13 07:17:31] <Volcane> and libc 2.3.2.ds1-22sa
[2008/06/13 07:17:40] <babysimon> I'll get my own backports repo eventually, it's on my todo list anyway
[2008/06/13 07:17:51] <Volcane> nods
[2008/06/13 07:18:25] <babysimon> I'm just curious though: do you guys edit directly on the master when you're developing a new class?
[2008/06/13 07:18:52] <Volcane> svn
[2008/06/13 07:19:32] <babysimon> yeah, we use hg, but then I make lots of commits saying "maybe it'll work this time"
[2008/06/13 07:19:54] <babysimon> "no, maybe this time!"
[2008/06/13 07:20:01] <Volcane> yeah :)
[2008/06/13 07:20:13] <Volcane> know the feeling
[2008/06/13 07:20:35] <jamesturnbull> babysimon: pin and upgrade - it's harmless and 0.24.x is incredibly more stable and featured
[2008/06/13 07:20:42] <babysimon> OK, I wondered if there was a good answer for that :)
[2008/06/13 07:21:26] <babysimon> jamesturnbull: I will get a backport going, honest, but I don't want to pull in a new libc and I don't see how I can avoid that.
[2008/06/13 07:21:40] <jamesturnbull> babysimon: fair enough
[2008/06/13 07:22:16] @ Quit: _zsh: Read error: 104 (Connection reset by peer)
[2008/06/13 07:22:59] <babysimon> Anyway, thanks everyone, you've given me some useful stuff to think about...
[2008/06/13 07:24:09] <mdray> hey. besides o'reilly velocity are there any upcoming conferences with puppet talks, labs or content?
[2008/06/13 07:25:37] @ Quit: johnf: Read error: 113 (No route to host)
[2008/06/13 07:30:18] @ Quit: dysinger:
[2008/06/13 07:46:27] @ Quit: babysimon: "Ex-Chat"
[2008/06/13 07:52:39] @ Quit: Innocenti: Read error: 110 (Connection timed out)
[2008/06/13 07:58:15] @ Quit: kenvandine: Read error: 110 (Connection timed out)
[2008/06/13 08:11:08] @ johnf joined channel #puppet
[2008/06/13 08:17:21] @ Quit: fsweetser: Remote closed the connection
[2008/06/13 08:28:02] @ kenvandine joined channel #puppet
[2008/06/13 08:29:30] @ kambiz joined channel #puppet
[2008/06/13 08:35:04] @ fsweetser joined channel #puppet
[2008/06/13 08:49:41] <fujin> jamesturnbull: should I file a ticket re that behaviour earlier?
[2008/06/13 08:50:33] @ keithlard joined channel #puppet
[2008/06/13 08:51:22] @ Quit: johnf: Read error: 113 (No route to host)
[2008/06/13 09:03:33] <jamesturnbull> fujin: yep
[2008/06/13 09:04:49] <jamesturnbull> mdray: OSCON
[2008/06/13 09:04:58] <jamesturnbull> mdray: Luke and I both speaking there
[2008/06/13 09:16:57] <mdray> jamesturnbull: okay, thanks
[2008/06/13 09:17:17] <mdray> (rats, can't make it :( )
[2008/06/13 09:18:02] @ Quit: sparanjape: "OUCH!!!"
[2008/06/13 09:21:01] * Volcane will defo plan to attend one or two next year
[2008/06/13 09:23:24] @ markl_ joined channel #puppet
[2008/06/13 09:27:22] <Volcane> does the autosign.conf accept network ranges? like 1.2.3.0/24 etc?
[2008/06/13 09:27:29] <Volcane> testing suggests it doesnt
[2008/06/13 09:29:07] <Volcane> but i might be doing something wrong
[2008/06/13 09:29:18] @ raasdnil joined channel #puppet
[2008/06/13 09:29:21] <raasdnil> evening all
[2008/06/13 09:29:42] <raasdnil> simple question. I am trying to figure out the best way to define a website for virtual hosting
[2008/06/13 09:30:14] <raasdnil> each site has a few variables, servername, aliases (can be more than one), admin email, directory, etc. Some of it I can infer (like the dir name from the host name)
[2008/06/13 09:30:24] <raasdnil> but what is the best way to do this in puppet?
[2008/06/13 09:30:51] <raasdnil> I want to be able to iterate over a collection of arrays, but can't quite see how to do this in puppet.
[2008/06/13 09:31:18] <raasdnil> Or would it be better to define a function / method and then call it each time for each website?
[2008/06/13 09:31:55] <fsweetser> I believe people typically create a define, and call it for each website
[2008/06/13 09:32:12] <raasdnil> that list could get fairly long is all.
[2008/06/13 09:32:23] <raasdnil> I guess tuck that away in a file and then include it in the node definition?
[2008/06/13 09:32:51] <fsweetser> if you already have the list somewhere else, like in a database, you could look into the external node method
[2008/06/13 09:33:38] <Volcane> i can show you an example of my define
[2008/06/13 09:33:45] <Volcane> wont work for all your cases, but you will probably learn fro it
[2008/06/13 09:34:00] <raasdnil> I have vhosts working fine. I am now setting up rails sites... and there is a bit of work in that :)
[2008/06/13 09:34:10] @ mikepea joined channel #puppet
[2008/06/13 09:34:12] <raasdnil> Volcane: yeah, that would be good
[2008/06/13 09:34:30] <Volcane> i just do apache::vhost{"blah.com": }
[2008/06/13 09:34:33] <Volcane> and it does it all
[2008/06/13 09:34:36] <raasdnil> fsweetser: external node method? hmm... :)
[2008/06/13 09:34:43] <Volcane> but then i have tons of options for extending/custoising it
[2008/06/13 09:35:17] <raasdnil> Volcane: that's cool. Right now I have a list of all the websites in an array, and I set a node variable to that and then have the template iterate over it. Works.. but pretty coupled up and I want to uncouple it some
[2008/06/13 09:35:25] <Disconnect> still fighting that timeout. any ideas? nothing is showing up in --test except the timeout - http://pastie.org/214436 .. server side doesn't even see that - http://pastie.org/214441
[2008/06/13 09:35:54] <Volcane> yeah let me just sanitise mine and i'll paste a link
[2008/06/13 09:38:13] <raasdnil> Volcane: cool
[2008/06/13 09:40:31] <Volcane> raasdnil: http://nephilim.ml.org/~rip/puppet/apache/ and theres a sample use in http://nephilim.ml.org/~rip/puppet/apache/manifests/vhosts/test_com.pp
[2008/06/13 09:40:48] <raasdnil> ta! :D
[2008/06/13 09:40:57] <Volcane> you want to check out manifests/vhosts.pp for some reade on the various options and whats possible
[2008/06/13 09:41:45] <Volcane> readme, damn i have my 'm' key
[2008/06/13 09:42:08] <Volcane> s/have/hate
[2008/06/13 09:44:05] <raasdnil> there's an irony in mistyping a correction of a mistype :)
[2008/06/13 09:44:33] <Volcane> hehe
[2008/06/13 09:44:39] <raasdnil> thanks hey. Gives me some good ideas. I thought it was something like that, wasn't quite sure where to start. Been using puppet for a few days now and it is changing my sysadmin life :)
[2008/06/13 09:44:52] <raasdnil> I refuse to log in to a server now except to check stuff :)
[2008/06/13 09:45:08] <Volcane> hehe
[2008/06/13 09:46:25] <mdray> heh heh
[2008/06/13 09:46:48] <mdray> i'm terrified puppet's going to go nutso in the middle of the night and take down our web farm
[2008/06/13 09:46:59] <raasdnil> :)
[2008/06/13 09:47:00] <mdray> note to self: check schedule {}
[2008/06/13 09:47:12] * raasdnil just got pf and firewall configs all in puppet
[2008/06/13 09:47:32] <Volcane> yeah i put my iptables in puppet recently and also fairly scared of it atm :)
[2008/06/13 09:47:41] <raasdnil> has an interesting side benefit, if you screw up a config and lock yourself out, 15-30 minutes later, you can get in again :)
[2008/06/13 09:48:41] <Volcane> if you dont screw it up so that outgoing doesnt work :P
[2008/06/13 09:49:55] <Disconnect> ...anyone? :( this has me completely stumped (and i have to do a demo in a few hours for the weekly review board)
[2008/06/13 09:50:25] <Volcane> Disconnect: no bells :(
[2008/06/13 09:52:01] <raasdnil> anyway, off to bed soon. Thanks again Volcane
[2008/06/13 09:52:26] @ Quit: raasdnil:
[2008/06/13 09:59:10] @ shenson_not_here is now known as shenson
[2008/06/13 10:03:25] <Disconnect> rebuilding the vm, some slightly different opts, we'll see.
[2008/06/13 10:03:33] <mikepea> has anyone there introduced the concept of 'milestones' into their puppet manifest? - that is, actions that don't do anything, but are used in 'require' and 'before'?
[2008/06/13 10:03:48] <ashp> what would be the purpose of that?
[2008/06/13 10:04:24] <mikepea> to prevent the need for lots of references between modules, really, but to try to simplify my dependencies too.
[2008/06/13 10:04:55] <ashp> to handle my dependencies I ended up splitting into a bootstrap env and a regular environment
[2008/06/13 10:05:09] <ashp> because previously I had to put require => Yumrepo['hls'] on about half my packages
[2008/06/13 10:05:13] <ashp> and it got annoying very quickly
[2008/06/13 10:05:46] <Volcane> ashp: Beginning Ruby is a sweet book
[2008/06/13 10:05:57] @ Quit: johano: Remote closed the connection
[2008/06/13 10:05:58] <ashp> volcane: I ordered the ruby way yesterday
[2008/06/13 10:06:06] <ashp> but I'm really annoyed because I paid for next day delivery and they estimate the 17th so far
[2008/06/13 10:06:09] @ skr joined channel #puppet
[2008/06/13 10:06:13] <Volcane> lol
[2008/06/13 10:06:18] <Volcane> they'll refund you
[2008/06/13 10:06:23] <Volcane> if u ask
[2008/06/13 10:06:30] <ashp> I'm going to, because that's some terrible next day :)
[2008/06/13 10:06:32] <mikepea> ashp: that's exactly the type of thing I mean - though i don't think my way would work for yumrepo - unless repo's are reread during a single run now?
[2008/06/13 10:06:52] <ashp> mikepea: There's an open ticket to add 'global dependencies'
[2008/06/13 10:07:13] <ashp> mikepea: So you'd be able to use some new syntax to effectively say Package{ require => "blah" } and have that always apply to all package{} statements
[2008/06/13 10:07:17] <ashp> which will resolve a lot of this
[2008/06/13 10:07:27] <mikepea> ashp: yeah, i remember it coming up...
[2008/06/13 10:07:32] @ a-priori joined channel #puppet
[2008/06/13 10:07:42] <ashp> I'm really looking forward to that one being tackled as I can massively clean my environment
[2008/06/13 10:07:53] <ashp> I want to be able to split 'site' dependencies from 'logical' dependencies
[2008/06/13 10:08:07] <ashp> that way I can distribute modules to other people without my own site requirements interfering with the dependencies tree
[2008/06/13 10:08:23] <mikepea> ashp: understood.
[2008/06/13 10:09:13] <mikepea> I'm trying to tackle the fact that our postgresql module depends on our sysctl module, which i'd rather express as 'install postgres after kernel configuration has been done'
[2008/06/13 10:09:47] <mikepea> s/install/start//
[2008/06/13 10:09:50] <ashp> I get what you mean, but I don't think there's an easy way to do that. :/
[2008/06/13 10:10:00] <Volcane> mikepea: easiest then it so make a class called 'kernelconfig' and require => Class["kernelconfig"]
[2008/06/13 10:10:11] <Volcane> then you can dup all the stuff in the class that needs to happen first
[2008/06/13 10:10:13] <ashp> but yeah, all you can do is encapsulate all kernel stuff into a singel class
[2008/06/13 10:10:15] <ashp> then require on that
[2008/06/13 10:10:25] <ashp> Volcane: can you do require => Module[]?
[2008/06/13 10:10:29] <ashp> I never tried but I'm curious :)
[2008/06/13 10:10:48] <Volcane> from that point of view module and class are the same thing
[2008/06/13 10:10:55] <Volcane> Class["module::sub"] works
[2008/06/13 10:11:15] <ashp> I mean I suppose it doesn't make sense to require a 'module' as it's just a collection of classes
[2008/06/13 10:11:20] <ashp> I just wondered what puppet did if you did :)
[2008/06/13 10:11:21] @ lak joined channel #puppet
[2008/06/13 10:11:41] <Volcane> doubt it knows the difference :)
[2008/06/13 10:12:43] <Volcane> ashp: remember your graphs, there were no Module[] rousources, they're just classes
[2008/06/13 10:12:57] <ashp> ahh that's true
[2008/06/13 10:13:10] <ashp> I haven't actually used any require => Class's
[2008/06/13 10:13:19] <ashp> I've tried to make it specific in individual bits within the classes where possible
[2008/06/13 10:13:22] <Volcane> they help a lot to satisfy loads of stuff at once
[2008/06/13 10:13:30] <ashp> I could probably simplify some bits to just require an entire class
[2008/06/13 10:13:42] <mikepea> me either - does it definitely work? I assumed require was just for types.
[2008/06/13 10:13:44] <Volcane> also, they help in containing change
[2008/06/13 10:13:48] @ jvanzyl joined channel #puppet
[2008/06/13 10:13:53] <Volcane> i mean, you can require Class["apache::package"]
[2008/06/13 10:14:03] <Volcane> and in package you can install either apache2 on debian or httpd on redhat
[2008/06/13 10:14:08] <Volcane> but the rest of your manifests dont care
[2008/06/13 10:14:20] <ashp> I had some trickiness getting the dependencies right in iclassify
[2008/06/13 10:14:21] <Volcane> which goes a long way towards making it more portable
[2008/06/13 10:14:33] <Volcane> mikepea: it works
[2008/06/13 10:14:34] <ashp> and it occurs to me I could have just done require => Class['mysql']
[2008/06/13 10:15:29] <Volcane> taking the above example, ifyou then want to go to apache3, you wouldnt need to go change all your requires that require Package["apache2"] etc, you can just keep working as before cos you require the encapsulating class
[2008/06/13 10:15:35] <Volcane> maybe apache is a bad example, but you get the idea
[2008/06/13 10:18:18] <mikepea> the only problem i can see with this is that it doesn't account for classes where the whole class is deployed over a long period (eg, postgres package installation, postgres database initialisation, postgres startup) - if I had a module that depended on the postgresql package being installed, but not started, then require => Class["postgresql::service"] would be a bit of an overkill.
[2008/06/13 10:18:47] <mikepea> I suppose modules should be divided up into sub-classes for the relevant sections though... hmmm.
[2008/06/13 10:19:06] @ Quit: keithlard:
[2008/06/13 10:19:08] <ashp> In all my modules I have class modulename, and that does include modulename::sub1, sub2 etc etc
[2008/06/13 10:19:13] <Volcane> generally i split my apache for example apache::package apache::service apache::config etc, and then i encapsulate apache to include those 3
[2008/06/13 10:19:19] <Volcane> mostly just a convenience thing
[2008/06/13 10:19:34] <Volcane> but i know i can just require Class["apache"] and all the default stuff related to it will be done
[2008/06/13 10:20:25] <mikepea> volcane: that's what i mean - i suppose it enforces decent class design. Hadn't thought about doing require => Class at all - i'm liking the idea a lot.
[2008/06/13 10:23:06] @ brscott joined channel #puppet
[2008/06/13 10:23:53] @ Quit: mikepea:
[2008/06/13 10:25:00] @ mikepea joined channel #puppet
[2008/06/13 10:26:24] @ ski98033 joined channel #puppet
[2008/06/13 10:33:50] <Disconnect> damn. here's my 5 minute warning :(
[2008/06/13 10:34:09] <Disconnect> hopefully this vm will Just Work and I can pretend it never went wrong
[2008/06/13 10:34:52] @ brscott left channel #puppet ()
[2008/06/13 10:36:07] @ keithlard joined channel #puppet
[2008/06/13 10:38:29] @ flakrat joined channel #puppet
[2008/06/13 10:40:05] @ glaw joined channel #puppet
[2008/06/13 10:47:50] <ashp> mikepea: One benefit about doing it just on the class
[2008/06/13 10:48:01] <ashp> is it shields other modules from the implementation details of each other
[2008/06/13 10:48:13] <mikepea> understood - me likey.
[2008/06/13 10:49:21] <mikepea> i'm going to tweak my modules to do this kind of require now, and see how I go. I wonder if i'll still need the milestone thing i was talking about - maybe, maybe not.
[2008/06/13 10:50:20] <mikepea> [hopefully not...]
[2008/06/13 10:51:02] <ashp> :) let us know how it works out!
[2008/06/13 10:51:26] <mikepea> is the multi-require syntax this btw:
[2008/06/13 10:51:29] <mikepea> require => [ File["/etc/init.d/postgresql-instance-${title}"], Class["sysctl::config"] ]
[2008/06/13 10:52:47] <Volcane> yes
[2008/06/13 10:53:04] <mikepea> sorry, rtfm n that - ta.
[2008/06/13 11:09:46] @ falsch joined channel #puppet
[2008/06/13 11:10:26] <mdray> gah. can i somehow get all of yum's output? yum install <foo>-<version> works fine but package { "<foo>": ensure => "<version>" } returns Could not update: Could not find package <foo>
[2008/06/13 11:15:40] <mdray> and.. huh. does class java {}; class java:foo inherits java {}; and class apache {}; class apache::foo inherits apache {}; cause problems? will foo be some kind of namespace clash?
[2008/06/13 11:15:53] <mdray> java::foo, of course
[2008/06/13 11:17:09] @ muerr joined channel #puppet
[2008/06/13 11:19:00] <Volcane> mdray: run puppetd with --trace --verbose --debug
[2008/06/13 11:19:24] <mdray> aha, lordy
[2008/06/13 11:19:30] <mdray> i thought --trace --test did it all!
[2008/06/13 11:19:43] <Volcane> its incredibly intuitive :P
[2008/06/13 11:20:00] <Volcane> not to mention unixy
[2008/06/13 11:20:17] <Volcane> i mean whatever happened of -vvvv
[2008/06/13 11:21:44] @ jbrothers1 joined channel #puppet
[2008/06/13 11:28:14] <Volcane> ashp: didnt you have a rpm for rubygems? /me is looking for one
[2008/06/13 11:29:14] <muerr> Hm. We mirror atrpms, rpmforge and centos 5.1 everything, and there's no rubygems rpm to be found :)
[2008/06/13 11:29:59] <Volcane> will get fro epel
[2008/06/13 11:30:12] <Volcane> muerr: Add EPEL to your list then :)
[2008/06/13 11:32:08] <ashp> I think I got my rubygems one off epel, yeah
[2008/06/13 11:32:33] <ashp> http://download.fedora.redhat.com/pub/epel/5/x86_64/rubygems-0.9.4-1.el5.noarch.rpm
[2008/06/13 11:32:47] <Volcane> yeah
[2008/06/13 11:32:50] <Volcane> doing sae
[2008/06/13 11:32:55] <Volcane> same
[2008/06/13 11:32:55] <muerr> well, we don't use any gems, so no need for that :)
[2008/06/13 11:33:05] <muerr> but i'll bring it up in my turnover meeting later this morning.
[2008/06/13 11:33:41] * Volcane wants to try capsistrano
[2008/06/13 11:37:18] @ jgonzalez left channel #puppet ()
[2008/06/13 11:39:14] <benp-> i started using it recently
[2008/06/13 11:39:23] <kambiz> if I want to turn a service off under some condition, or on under other conditions, what's the right way to do that? puppet only lets me have one service { "foo": ... for a given foo.
[2008/06/13 11:40:33] <muerr> I was intrigued that capistrano doesn't install in a normal system path by default on Ubuntu.
[2008/06/13 11:42:17] <ashp> kambiz: You could do an exec and use onlyif => or unless => to decide the conditions?
[2008/06/13 11:42:47] <kambiz> yeah ... that's what I thing I'll do ... use an exec for the on condition, and service for the off condition.
[2008/06/13 11:43:00] <kambiz> or exec for both cases.
[2008/06/13 11:45:46] @ jsgotangco joined channel #puppet
[2008/06/13 11:46:37] <Volcane> gem2rpm++
[2008/06/13 11:47:59] <mdray> damn you, yum
[2008/06/13 11:48:20] <mdray> can i somehow force the yum provider to go to our own repo before RHEL?
[2008/06/13 11:48:36] <ashp> increase the version of your local package to 99999999
[2008/06/13 11:48:41] <ashp> (i don't know if that would work!)
[2008/06/13 11:49:37] <benp-> thats actually not a bad idea, you could change the release (not version) of the rpm
[2008/06/13 11:49:56] <benp-> you'd have to rebuild it, but thats not a big deal if you have the spec or srpm
[2008/06/13 11:50:11] <sigmonsays> is there a way to manually fetch a file from the puppet master?
[2008/06/13 11:50:35] @ Quit: muerr: Read error: 104 (Connection reset by peer)
[2008/06/13 11:50:48] @ muerr joined channel #puppet
[2008/06/13 11:51:57] <mikepea> sigmonsays: you could use standalone puppet to do so I would expect.
[2008/06/13 11:52:05] <Volcane> you can use priorities
[2008/06/13 11:52:49] <Volcane> (for yum)
[2008/06/13 11:53:30] <sigmonsays> mikepea, interesting. i'll give it a shot
[2008/06/13 11:53:54] <mdray> benp-: yeah, they're all home-built rpms
[2008/06/13 11:54:01] <mdray> benp-: but it kind of sucks to have to do that!
[2008/06/13 11:55:03] @ Innocenti joined channel #puppet
[2008/06/13 11:56:06] @ shake-n-bake joined channel #puppet
[2008/06/13 11:56:30] @ trombik_ joined channel #puppet
[2008/06/13 11:56:33] @ Quit: strerror_work:
[2008/06/13 11:57:03] <benp-> i think the priorities thing needs a plugin
[2008/06/13 11:57:20] <benp-> but its probably a better solution
[2008/06/13 12:00:06] <muerr> yum-priorities.noarch - This plugin allows repositories to have different priorities.
[2008/06/13 12:00:16] <mdray> yeah, just saw that
[2008/06/13 12:00:19] <mdray> checking it out
[2008/06/13 12:01:31] <Volcane> you can also use exclude and includepkgs to fiddle around and force a certain rp from a certain repo
[2008/06/13 12:01:44] <Volcane> that doesnt require plugins
[2008/06/13 12:02:14] <mdray> but can i do that in a package {} ?
[2008/06/13 12:04:18] <Volcane> nope thats repo settings
[2008/06/13 12:04:25] <mdray> mm
[2008/06/13 12:05:09] @ Quit: trombik: Read error: 110 (Connection timed out)
[2008/06/13 12:09:37] * Volcane wants to use Capistrano to do yum updates etc
[2008/06/13 12:11:22] @ Quit: kolla: Remote closed the connection
[2008/06/13 12:11:35] <benp-> it can do that
[2008/06/13 12:11:55] @ Quit: glaw: Remote closed the connection
[2008/06/13 12:11:57] <benp-> its basically ssh-and-a-for-loop++
[2008/06/13 12:13:04] <Volcane> yeah i do that now
[2008/06/13 12:22:32] @ Quit: Innocenti: Read error: 110 (Connection timed out)
[2008/06/13 12:25:34] <Volcane> need to see some more involved samples of cap i guess
[2008/06/13 12:25:50] <Volcane> yet to see the big benefit over for loops :P
[2008/06/13 12:27:30] <benp-> one of them is it already comes with a very nice deploy recipe for rails apps. which you can override if you're not using rails
[2008/06/13 12:29:12] @ Quit: jsgotangco: "Ciao"
[2008/06/13 12:29:40] <benp-> you don't need to type out a list of servers every time, you just assign servers to roles.. role :db, "dbhost1", "dbhost2"
[2008/06/13 12:29:56] <benp-> tasks are executed in parallel
[2008/06/13 12:30:04] <benp-> and you can make reusable tasks
[2008/06/13 12:30:19] <mdray> thanks everybody
[2008/06/13 12:30:21] @ Quit: mdray: "Leaving"
[2008/06/13 12:30:40] <Volcane> benp-: well, all my redhat machines sends daily 'yum check-update's to a central box
[2008/06/13 12:30:57] <Volcane> on that web page i just click a machine and at the bottom it builds a forloop to update all thats clicked
[2008/06/13 12:31:25] <Volcane> so little advantage over that for me at this point
[2008/06/13 12:34:18] @ Quit: DerekW: "Leaving"
[2008/06/13 12:34:27] <benp-> yep, if you're using puppet and you have that, you likely don't have a need for cap
[2008/06/13 12:36:29] <Volcane> nods, atm i manually add my hosts to the thing that builds it (it needs a list of achines to accept mail from) soon I'll use exported resources for that and it will be sweet++
[2008/06/13 12:37:15] <muerr> Volcane: there's a pretty good example of capistrano on the puppet wiki for regenerating debian ssl certificates :)
[2008/06/13 12:38:03] <Volcane> lol
[2008/06/13 12:38:46] <Volcane> the code deploys my clients do are way too fragile to try and automate atm
[2008/06/13 12:38:56] <Volcane> but its kewl, worthwhile tool to be aware of
[2008/06/13 12:39:18] <Volcane> I do have some fancy scripts to do stuff on my mail clusters, will replace with cap
[2008/06/13 12:39:45] <Volcane> though still, i have ones that goes off, loads mailq output into local files, then cat the lot into exiqsumm to get a global view of stuck mail
[2008/06/13 12:39:52] <Volcane> sees above cap's abilities
[2008/06/13 12:42:04] <Volcane> muerr: oh thanks for that example, its pretty neat yeah
[2008/06/13 12:42:26] @ Quit: z00dax: Nick collision from services.
[2008/06/13 12:42:43] <muerr> Volcane: :)
[2008/06/13 12:43:15] @ Quit: jvanzyl:
[2008/06/13 12:46:39] @ Quit: mikepea: Read error: 104 (Connection reset by peer)
[2008/06/13 12:47:06] @ mikepea joined channel #puppet
[2008/06/13 12:50:59] @ lak_ joined channel #puppet
[2008/06/13 12:52:11] @ Quit: mikepea:
[2008/06/13 12:54:11] * lak gives limechat a try instead of colloquy
[2008/06/13 12:54:11] <sigmonsays> can someone explain to me why I can't read docs on facter with "ri facter"
[2008/06/13 12:54:43] @ Quit: lak:
[2008/06/13 12:54:56] @ lak_ is now known as lak
[2008/06/13 12:57:07] <ashp> afternoon lak :)
[2008/06/13 12:57:19] <lak> hi
[2008/06/13 12:57:45] <muerr> sigmonsays: maybe lak can :-)
[2008/06/13 12:57:50] <muerr> howdy ashp, lak
[2008/06/13 12:58:05] <lak> i know almost nothing about ri
[2008/06/13 12:58:23] <lak> partially because it basically never works as i expect, so i gave up on it a long time ago
[2008/06/13 12:58:24] <sigmonsays> apparantly nobody uses "man-like" documentation anymore
[2008/06/13 12:58:39] <sigmonsays> they all say to go to the web
[2008/06/13 12:58:40] <muerr> sigmonsays: i blame RMS and info.
[2008/06/13 12:58:59] <muerr> "complete documentation can be found in the info page!"
[2008/06/13 12:59:00] <lak> i've tried to provide man pages, but it's been a lot more work than i would have liked :/
[2008/06/13 12:59:20] <muerr> info <command>: info, command not found. yum install info; install emacs too? please? argh!
[2008/06/13 12:59:23] <sigmonsays> well python has pydoc that works pretty well. ri is supposed to be the same thing
[2008/06/13 12:59:33] <sigmonsays> but it doesn't work :)
[2008/06/13 12:59:50] <sigmonsays> muerr, info is lame. man is kick ass though
[2008/06/13 13:01:35] <jbrothers1> only when the man page has examples
[2008/06/13 13:02:29] <muerr> sigmonsays: concur.
[2008/06/13 13:02:35] <muerr> jbrothers1: concur.
[2008/06/13 13:02:44] <muerr> the nmap man page, for example, is A+ awesome ^2
[2008/06/13 13:05:48] * Volcane also wish for a better doc method than the wiki, mostly cos its slow
[2008/06/13 13:05:57] <Volcane> not navigating it, its just slow
[2008/06/13 13:06:16] <Volcane> I'd host a mirror for free if there was some static export or some such
[2008/06/13 13:08:19] <jbrothers1> muerr: my favorite is xsublim(6)
[2008/06/13 13:10:14] @ z00dax_ joined channel #puppet
[2008/06/13 13:10:53] <muerr> not familiar with it.
[2008/06/13 13:16:27] @ Quit: kambiz: "Leaving"
[2008/06/13 13:16:48] @ exodos joined channel #puppet
[2008/06/13 13:17:47] @ Quit: falsch: Read error: 104 (Connection reset by peer)
[2008/06/13 13:18:47] @ plathrop joined channel #puppet
[2008/06/13 13:29:05] <bda> hm.
[2008/06/13 13:29:13] <bda> Anyone have a suggestion for how this might me made to work:
[2008/06/13 13:29:25] <bda> exec { "/usr/sbin/svcs svc:/icg/system/syslog-ng:default || /usr/sbin/svccfg import /sw/icg-admin/etc/smf/manifest/site/syslog-ng.xml" : }
[2008/06/13 13:30:12] <bda> Just reimporting it seems to cause it to HUP, which is not so preferable.
[2008/06/13 13:31:07] <Volcane> might be able to do soething with the onlyif option to exec
[2008/06/13 13:31:35] <bda> oh ho.
[2008/06/13 13:31:41] * bda looks at the wiki.
[2008/06/13 13:31:56] @ jvanzyl joined channel #puppet
[2008/06/13 13:32:10] @ Quit: exodos: "Lost terminal"
[2008/06/13 13:33:26] @ notbrien joined channel #puppet
[2008/06/13 13:34:16] @ exodos joined channel #puppet
[2008/06/13 13:34:43] <ashp> yeah, you need onlyif or creates => possibly
[2008/06/13 13:34:53] <ashp> i find creates => is more useful as I can just check for a file I know SHOULD exist :)
[2008/06/13 13:35:12] <muerr> ashp: yum
[2008/06/13 13:35:27] @ notbrien left channel #puppet ()
[2008/06/13 13:35:30] <bda> Volcane: onlyif is totally what I wanted. Thanks!
[2008/06/13 13:35:52] <bda> ashp: I was checking rv. If the svc doesn't exist, svcs returns 1. :)
[2008/06/13 13:36:13] @ dysinger joined channel #puppet
[2008/06/13 13:37:48] @ Quit: jvanzyl:
[2008/06/13 13:38:15] @ ski98033 left channel #puppet ()
[2008/06/13 13:38:18] <Volcane> bda: no prob
[2008/06/13 13:39:53] <bda> (Well, I really wanted unless, but it's on the same page, so. :)
[2008/06/13 13:41:38] <Volcane> hehe
[2008/06/13 13:52:06] @ falsch joined channel #puppet
[2008/06/13 13:52:53] @ jvanzyl joined channel #puppet
[2008/06/13 13:54:11] <plathrop> Oh man, RSpec has me so spoiled.
[2008/06/13 13:54:19] <plathrop> I blame lak.
[2008/06/13 13:54:38] <plathrop> Now I have to program in Python and I don't want to because the test frameworks all suck.
[2008/06/13 14:06:06] <ashp> i am so tired, i wonder if I could just craw under my desk
[2008/06/13 14:15:28] <holoway> plathrop: yeah, rspec is great
[2008/06/13 14:15:31] <holoway> have you odne user stories yet?
[2008/06/13 14:15:39] <holoway> best integration testing framework ever
[2008/06/13 14:15:45] <holoway> at least, that I've ever used
[2008/06/13 14:17:43] <muerr> i haven't written anything meaningful in python in almost 10 years, and it looks like my future programming endeavours will be in Ruby, primarily.
[2008/06/13 14:18:10] <holoway> lots of people with the big python love
[2008/06/13 14:18:24] <muerr> Yeah
[2008/06/13 14:18:43] <muerr> python was okay.
[2008/06/13 14:18:57] <muerr> i like ruby better :)
[2008/06/13 14:19:25] <ashp> i really liked python, it made more sense to me than ruby
[2008/06/13 14:19:30] <ashp> I know that sounds insane, but there it is
[2008/06/13 14:19:50] <holoway> ashp: doesn't sound insane
[2008/06/13 14:19:55] <holoway> different strokes for different folks
[2008/06/13 14:20:00] * holoway sings the theme song
[2008/06/13 14:20:03] <ashp> i'd do better at ruby if I could get amazon to ship my book already
[2008/06/13 14:20:08] <ashp> instead of making me read a bad quality pdf
[2008/06/13 14:21:10] @ shake-n-bake_ joined channel #puppet
[2008/06/13 14:21:47] @ Quit: plathrop: Read error: 110 (Connection timed out)
[2008/06/13 14:29:40] @ Quit: shake-n-bake: Read error: 110 (Connection timed out)
[2008/06/13 14:29:51] <muerr> I'd be better at ruby if i spent more time writing code in it :)
[2008/06/13 14:30:02] <muerr> which is more likely to happen in two weeks than any sooner.
[2008/06/13 14:30:09] @ Ol_ joined channel #puppet
[2008/06/13 14:30:32] @ Wicky656 joined channel #puppet
[2008/06/13 14:30:51] <ashp> if i could keep my eyes open i'd tackle ruby harder, but my wife is slowly killing me
[2008/06/13 14:35:24] <gileswork> umm we've got a working module for debian network interfaces
[2008/06/13 14:35:55] <gileswork> but i'm wanting to do an exec to ifup and ifdown the interfaces
[2008/06/13 14:36:12] <gileswork> has someone already done a decent module?
[2008/06/13 14:40:17] @ mikearr joined channel #puppet
[2008/06/13 14:58:17] <gileswork> I suppose has anyone done a fact that reads all the network interfaces
[2008/06/13 15:00:01] <muerr> there's a set of facts that return the known IP addresses on the systme.
[2008/06/13 15:00:14] <muerr> facter | grep ipaddress
[2008/06/13 15:00:25] <ashp> lak: hey when you write that GUI, write a puppet module that installs it too :D
[2008/06/13 15:00:48] <muerr> ashp: pfft. package { "puppet-gui": ensure => latest }
[2008/06/13 15:01:15] <ashp> Well, I HOPE it's that simple!
[2008/06/13 15:01:20] <ashp> But I assume it'll be a little more complex ;)
[2008/06/13 15:05:43] <lak> ashp: that's the plan, certainly
[2008/06/13 15:05:52] * lak lunch &
[2008/06/13 15:06:03] <ashp> Always nice to not have to make my own module :)
[2008/06/13 15:08:52] @ sazma joined channel #puppet
[2008/06/13 15:09:24] @ Quit: flakrat: "Leaving"
[2008/06/13 15:10:25] <sazma> we created a custom fact and put it in the module under modules/name/plugins/facter and now with factsync we get an error "failed to generate additional resources during transaction: Fileserver module 'facts' not mounted"
[2008/06/13 15:10:42] <sazma> tips? hints? cluebats?
[2008/06/13 15:10:55] <ashp> well, there is a good hint in that error :)
[2008/06/13 15:11:00] <ashp> what do you have for [facts] in fileserver.conf?
[2008/06/13 15:11:09] <sazma> nothing
[2008/06/13 15:11:17] <ashp> [facts] allow * path /etc/puppet/facts
[2008/06/13 15:11:21] <ashp> Whoops, that messed up
[2008/06/13 15:11:23] <benp-> i thought you needed to use pluginsync instead of factsync if your facts live in modules?
[2008/06/13 15:11:32] <ashp> that was [facts] \n allow * \n path /etc/etc
[2008/06/13 15:11:40] <benp-> or does factsync work with modules?
[2008/06/13 15:11:49] <sazma> we had a [facts] section and it just gave a different error
[2008/06/13 15:11:49] <ashp> I.. I don't have any custom facts yet, so I don't know. :)
[2008/06/13 15:12:31] <sazma> so try with pluginsync?
[2008/06/13 15:12:50] <sazma> hmm we have pluginsync turned on already
[2008/06/13 15:12:51] <benp-> to be honest, im not really sure. i just use factsync and i dont use it with modules
[2008/06/13 15:13:30] @ kolla joined channel #puppet
[2008/06/13 15:14:53] <sazma> I was hoping lak was around and would pop his cute little head up
[2008/06/13 15:15:12] <muerr> he's lunching.
[2008/06/13 15:15:22] <muerr> about 3 minutes before you got here :)
[2008/06/13 15:15:51] <sazma> figures :)
[2008/06/13 15:17:48] @ kambiz joined channel #puppet
[2008/06/13 15:18:33] <Disconnect> so is there a simple way to apply iclassify tags to puppet nodes? (eg "for every node with tag tomcat6, do these things")
[2008/06/13 15:27:35] * muerr pokes holoway
[2008/06/13 15:29:44] @ Quit: sigmonsays: Read error: 110 (Connection timed out)
[2008/06/13 15:30:25] @ sigmonsays joined channel #puppet
[2008/06/13 15:41:16] @ Quit: a-priori:
[2008/06/13 15:41:54] @ jvanzyl_ joined channel #puppet
[2008/06/13 15:42:19] @ Innocenti joined channel #puppet
[2008/06/13 15:44:53] @ Quit: keithlard:
[2008/06/13 15:49:52] @ Quit: jvanzyl: Read error: 110 (Connection timed out)
[2008/06/13 15:50:51] <holoway> Disconnect: yeah - do a search for the nodes that match, and you can replace all the tags easily
[2008/06/13 15:53:02] <Disconnect> maybe i'm having a stupid day (ok, i've had a stupid week.. culminating with my jeep getting broken into yesterday) but i'm looking to add something to the manifest to do stuff like "if the server has tag 'apache' then 'include apache::foo' .."
[2008/06/13 15:53:18] <Disconnect> unclear on how they tie together
[2008/06/13 15:53:39] <holoway> tags map 1:1 to classes
[2008/06/13 15:53:55] <holoway> so you would have "application-server" that includes apache, apache::foo, etc
[2008/06/13 15:54:00] <Disconnect> how does it find the iclassify install?
[2008/06/13 15:54:28] <holoway> you configure puppet to use icpuppet or icpuppet-thin as an external node exec terminus
[2008/06/13 15:54:38] <Disconnect> ah ok. thats the piece i was missing i think.
[2008/06/13 15:55:07] <holoway> (the difference being, icpuppet-thin goes directly to Solr, which is way less overhead, and works fine as long as iclassify and your puppetmaster are the same host)
[2008/06/13 15:55:13] <Disconnect> wow. "did you mean mcpuppet?" .. i'm almost afraid (visions of rapping muppets..make them stop..)
[2008/06/13 15:55:22] <Disconnect> works fine even if not in the old default config :)
[2008/06/13 15:55:28] <holoway> no, but I should have called it mcpuppet
[2008/06/13 15:55:32] <Disconnect> heh
[2008/06/13 15:58:54] @ flakrat joined channel #puppet
[2008/06/13 15:59:36] <muerr> if i have a facts dir in my puppet manifests, the facts contained therein are available on clients, no?
[2008/06/13 16:00:03] <Volcane> you need factsync and a facts mount
[2008/06/13 16:00:19] <muerr> k
[2008/06/13 16:02:05] <muerr> and i just enable that on the puppetmaster's puppet.conf, not on the clients right
[2008/06/13 16:02:12] * Volcane tried to get wildcards or cidr's into autosign.conf but it seems to barf on it, anyone know if its supported?
[2008/06/13 16:02:29] <Volcane> muerr: u wish :)
[2008/06/13 16:02:33] <muerr> boo
[2008/06/13 16:02:55] <Volcane> theres a wiki page iwth samples, i think its AddingFacts not sure
[2008/06/13 16:04:12] <muerr> yeah i was already there ><
[2008/06/13 16:04:34] <muerr> that seems to work :)
[2008/06/13 16:04:38] <Volcane> kewl
[2008/06/13 16:04:53] <muerr> now to create a puppet recipe that makes puppet update its config file everywhere
[2008/06/13 16:05:07] <muerr> we don't actually have client puppet.conf changes (until now).
[2008/06/13 16:06:30] <Volcane> i add graph, report, factsync, factpath and factdest
[2008/06/13 16:07:04] <sazma> lak bak yet?
[2008/06/13 16:07:05] <Volcane> last 2 soi canjust add puppet libdir to RUBYLIB to test facts
[2008/06/13 16:07:14] <lak> i'm around
[2008/06/13 16:07:26] <Volcane> god knows why the puppet defaults isnt more sane in the choise for those
[2008/06/13 16:08:02] <sazma> we created a custom fact and put it in the module under modules/name/plugins/facter and now with factsync we get an error "failed to generate additional resources during transaction: Fileserver module 'facts' not mounted"
[2008/06/13 16:08:04] @ mikepea joined channel #puppet
[2008/06/13 16:08:21] <lak> sazma: you need to create a facts section in your fileserver.conf
[2008/06/13 16:08:52] <sazma> lak, we had one and it just gave a different error
[2008/06/13 16:09:03] <sazma> about not being mounted or something... I can re-create if necessary
[2008/06/13 16:09:21] <lak> please
[2008/06/13 16:10:04] <bda> Has anyone been working on a ZFS provider?
[2008/06/13 16:10:39] <sazma> "Failed to generate additional resources during transaction: fileserver module "facts" not mounted" and "could not describe /facts: fileserver modules facts not mounted"
[2008/06/13 16:11:00] * sazma shivers at the thought of zfs
[2008/06/13 16:11:12] <Volcane> sazma: and u gave it a path too?
[2008/06/13 16:11:13] <lak> that's the same error, isn't it?
[2008/06/13 16:11:26] <sazma> sec sorry
[2008/06/13 16:12:23] @ Quit: Innocenti: Read error: 110 (Connection timed out)
[2008/06/13 16:12:24] <muerr> ok, good grief. i had added the facts section to fileserver.conf long ago, but never added a puppet.conf for clients with factsync=true
[2008/06/13 16:12:33] <muerr> that explains why those facts weren't getting used and forgotten about :-x
[2008/06/13 16:12:39] <sazma> working on re-creating it
[2008/06/13 16:13:52] <Volcane> lak: is there any compelling reason why fileserverconfig cant be added to environemtn sections? to override which fileserver config to use per env?
[2008/06/13 16:13:59] <sazma> fileserver.conf on the nodes or on the puppetmaster or both?
[2008/06/13 16:14:23] <muerr> puppetmaster. you don't need a fileserver.conf on nodes.
[2008/06/13 16:14:25] <lak> sazma: server
[2008/06/13 16:14:25] <lak> Volcane: well, it would be a signfiicant refactoring internally
[2008/06/13 16:14:26] <lak> there's only one fileserver
[2008/06/13 16:14:28] <lak> with one config
[2008/06/13 16:14:54] <sazma> ok so assuming my custom fact is in a module, what path do I use?
[2008/06/13 16:15:05] <sazma> or is it irrelevant?
[2008/06/13 16:15:07] <Volcane> lak: hmm, pity cos it would mean all old manifests etc can immediately benefit even if it's strewn with non own-defined file overrides etc
[2008/06/13 16:15:21] <muerr> we eschewed module-specific facts. that much i remember when i was working on this before.
[2008/06/13 16:15:29] @ Quit: Ol_:
[2008/06/13 16:15:37] <lak> muerr: that's because they don't work yet :)
[2008/06/13 16:15:50] <sazma> so the wiki is wrong again?
[2008/06/13 16:15:57] <sazma> what's a guy to do?
[2008/06/13 16:16:10] <Volcane> use the old way :P
[2008/06/13 16:16:27] <lak> what?
[2008/06/13 16:16:33] <sazma> I mean about the greater problem of documentation always being wrong
[2008/06/13 16:16:33] <lak> sazma: what's wrong about the wiki?
[2008/06/13 16:16:45] <sazma> http://reductivelabs.com/trac/puppet/wiki/PluginsInModules
[2008/06/13 16:17:04] <sazma> and the custom facts page points to that page saying it's the rec method
[2008/06/13 16:17:50] <Volcane> i use the customfacts way it works great
[2008/06/13 16:18:04] <sazma> cool tell me what I'm doing wrong
[2008/06/13 16:18:06] <muerr> lak: oh yeah!
[2008/06/13 16:18:08] <Volcane> also had hassles with facts in modules
[2008/06/13 16:18:24] <muerr> lak: well hell that explains a lot then. is there a ticket? because i might have seen that and said oh.. nm.
[2008/06/13 16:18:44] <lak> dunno; there are lots of tickets, and i can't keep the whole list in my mind at once :/
[2008/06/13 16:18:53] <muerr> lak: oh cmon.
[2008/06/13 16:18:56] <muerr> lak: :)
[2008/06/13 16:18:56] <kambiz> if I want multiple "onlyif"s to be true, do I just do : onlyif => ["shell_command_1", "shell_command_2"],
[2008/06/13 16:19:00] <kambiz> and assume &&
[2008/06/13 16:19:01] <kambiz> ?
[2008/06/13 16:19:06] <sazma> so use the old way is the answer?
[2008/06/13 16:19:13] <lak> use the old way for now, yes
[2008/06/13 16:19:18] <sazma> ok
[2008/06/13 16:19:24] @ Ol_ joined channel #puppet
[2008/06/13 16:19:53] <muerr> sazma: i have a [facts] section in fileserver.conf, a 'facts' directory in /var/lib/puppet (where my manifests dirs, master and modules live as well), and the facts are in there.
[2008/06/13 16:19:59] <lak> we're working on it, and the facter 1.5 release will fix a lot of it
[2008/06/13 16:20:18] <muerr> lak: i saw some of the devel discussion on facter, looking forward to 1.5
[2008/06/13 16:20:55] <lak> i think it's actually ready, not sure if we're waiting for anything else
[2008/06/13 16:21:08] <muerr> actually, speaking of, and the reason i revisited custom facts today - the operatingsystemrelease fact on centos doesn't pull the version out of /etc/redhat-release, it gets it from the next block of code which grabs the kernel version.
[2008/06/13 16:21:36] <Volcane> muerr: use the lsb facts?
[2008/06/13 16:21:38] <muerr> operatingsystemrelease has a confine for :redhat, but not one for :centos. so on centos (even though its release is in /etc/redhat-release).
[2008/06/13 16:21:50] <muerr> Volcane: lsb != installed on our centos boxes :)
[2008/06/13 16:21:59] <Volcane> thought so :P
[2008/06/13 16:22:07] <muerr> and won't be on our rhel boxes when we tune the package list.
[2008/06/13 16:22:16] <muerr> there's only two rhel boxes, so thats lower priority.
[2008/06/13 16:22:16] <Volcane> but yes, i also got annoyed by that
[2008/06/13 16:22:21] @ Quit: flakrat: "Leaving"
[2008/06/13 16:22:29] * benp- senses an hour long conversation about lsb about to start
[2008/06/13 16:22:34] <Volcane> lol
[2008/06/13 16:22:37] <muerr> lol
[2008/06/13 16:22:43] <muerr> i think we've already had that actually.
[2008/06/13 16:22:46] <benp-> we have
[2008/06/13 16:23:10] <muerr> lsb wants cups, and we don't want cups (because web servers don't have printers goddamnit), and thats all i should need to say about that.
[2008/06/13 16:23:18] @ plathrop joined channel #puppet
[2008/06/13 16:24:44] @ Quit: markl_: "Lost terminal"
[2008/06/13 16:26:14] @ markl_ joined channel #puppet
[2008/06/13 16:30:25] <e^ipi> doesn't hurt to have stuff installed.
[2008/06/13 16:30:29] <Disconnect> lsb full has cups - afaik lsbrelease (or however its packaged) doesn't..
[2008/06/13 16:30:36] <e^ipi> disk is /cheap/
[2008/06/13 16:31:12] <Disconnect> e^ipi: sure it does, from a security standpoint. elevation of privs from commandline is BAD "so he broke in as www-data, and that was fine, but then he used {random installed crap} to become really evil"
[2008/06/13 16:31:26] <Volcane> Disconnect: not on redhat
[2008/06/13 16:32:00] <Disconnect> no, not on redhat. cuz in general you can't make redhat any more evil than it was before you got there.
[2008/06/13 16:32:03] <muerr> its not a disk space issue, its a potential security issue. it is impossible for someone to exploit a program that doesn't exist.
[2008/06/13 16:32:24] <muerr> yeah what Disconnect said :-x.
[2008/06/13 16:32:32] <muerr> yes you can disable the cups service.
[2008/06/13 16:32:54] <muerr> but part of what we teach is "Defense in Depth" and disabling the service isn't enough when you can remove the package entirely.
[2008/06/13 16:33:08] <e^ipi> i guess i'm just used to solaris, where RBAC can prevent you from exploiting anything other than what you're explicitly allowed to
[2008/06/13 16:33:24] <e^ipi> "so he broke in as www-data, but since that role doesn't have exec(2) priveleges, he just kinna sat there"
[2008/06/13 16:33:27] <muerr> Trusted Solaris?
[2008/06/13 16:33:33] <e^ipi> no, solaris proper
[2008/06/13 16:33:33] <Disconnect> i hate the redhat method of "install everything under the sun so that its there, JUST IN CASE"..
[2008/06/13 16:33:43] <muerr> how long has solaris had rbac?
[2008/06/13 16:33:48] <Disconnect> and how do you run cgi without exec()?
[2008/06/13 16:33:49] <muerr> Disconnect: amen.
[2008/06/13 16:33:52] <e^ipi> trusted solaris is something you use when you have different "classified" levels
[2008/06/13 16:34:01] <muerr> i haven't touched solaris since version 8
[2008/06/13 16:34:07] <e^ipi> Disconnect: you remove exec for anything except what you explicitly need to
[2008/06/13 16:34:10] <Volcane> you could easily create a selinux policy to disable all the cups bins
[2008/06/13 16:34:16] <Disconnect> ya
[2008/06/13 16:34:17] @ marcoecc joined channel #puppet
[2008/06/13 16:34:26] <muerr> and even then, most of my experience with solaris is on 2.6, not many clients were on 8 when i worked on it
[2008/06/13 16:34:28] <Disconnect> Volcane: or just not install them to begin with
[2008/06/13 16:34:36] <Volcane> anyway, i prefer when redhat says its lsb compliant its lsb compliant, unlike the debian way
[2008/06/13 16:34:40] <e^ipi> default-deny is the /only/ security policy that works
[2008/06/13 16:35:10] <Disconnect> (fwiw in ubuntu fsb-release depends on basic-python and nothing else)
[2008/06/13 16:36:01] <Volcane> the whole points of lsb is to provide a standard bunch of stuff that all distros complies too
[2008/06/13 16:36:05] <Volcane> thats its purpose
[2008/06/13 16:36:07] <Disconnect> not sure if its different now, but it used to be that LSB was a "standard" the same way IE html parsing is "standard" .. redhat and co said "this is how we do it so its standard" and there ya go..
[2008/06/13 16:36:18] <e^ipi> Volcane: like POSIX, or SUS
[2008/06/13 16:36:36] <Volcane> only creating version files and distro info in lsb configs but non of the enabling provides doesnt add any value at all
[2008/06/13 16:36:51] <Disconnect> oh and fyi from the lsb page,s entence one: "The Linux Standard Base delivers interoperability between applications and the Linux operating system. Currently all major distributions comply with the LSB" .. so I dunno where you get "the debian way" above
[2008/06/13 16:37:31] <Volcane> Disconnect: debian lets you install only a very very small portion of lsb compliance that doesnt actually add any value from a lsb point of view
[2008/06/13 16:37:48] <Volcane> Disconnect: where with redhat once you install lsb compliance, you're compliant period
[2008/06/13 16:38:08] <muerr> heh.
[2008/06/13 16:38:22] <Volcane> lsb says you need /usr/bin/lpr so redhat gives it that
[2008/06/13 16:38:47] * e^ipi giggles at linux's attempt at "standards"
[2008/06/13 16:39:11] <Disconnect> http://packages.debian.org/etch/lsb .. "This package provides an implementation of all modules of version 3.1 of the Linux Standard Base for Debian on the Intel x86, Intel ia64 (Itanium), IBM S390, and PowerPC 32-bit architectures with the Linux kernel." .. lsb spec is broken into modules on PURPOSE.. and so are the packages. or are you saying that the lsb-core package (for example) doesn't support the Core spec of the LSB?
[2008/06/13 16:40:00] <Volcane> Disconnect: i am saying lsb-core provides version info and almost nothing else, it doesnt actually add any of the compatility bits that would mean anything to a program.
[2008/06/13 16:40:12] <muerr> I didn't mean to stir up the hornet's nest about LSB :-\
[2008/06/13 16:40:14] <Disconnect> even lsb says that lsb-release is really useful outside lsb. so being able to install that (without rsync, without a mail daemon, without ...etc..) is a big boon
[2008/06/13 16:40:30] <Disconnect> Volcane: such as?
[2008/06/13 16:40:50] <muerr> I do wish that Red Hat's "minimal" install were actually minimal. Last I knew it still installed a pile of GNOME crap.
[2008/06/13 16:40:51] <Disconnect> seriously.. point to the spec and say "that, right there, thats missing from core" and I'll file the bug right this second
[2008/06/13 16:40:52] * Volcane ignores Disconnect
[2008/06/13 16:41:09] <muerr> lol
[2008/06/13 16:41:31] <Volcane> muerr: you could create a little package that symlinks /bin/true to /usr/bin/lpr and install that instead of cups - and the few small other bins it needs
[2008/06/13 16:41:35] @ Quit: kenvandine: "Ex-Chat"
[2008/06/13 16:41:45] <Volcane> muerr: and lsb in redhat will be happy
[2008/06/13 16:42:04] <Volcane> muerr: but, from a adding lsb compliance in a usable way point of view, it would suck
[2008/06/13 16:42:52] * Disconnect forgot the cardinal rule - don't ask redhat fanboys to provide examples for any of their claims about redhat superiority..
[2008/06/13 16:43:02] <muerr> i really don't care about lsb compliance for our boxes.
[2008/06/13 16:43:20] <Volcane> Disconnect: my argument wasnt against or for redhat or debian, my argument was that adding just the core module doesnt add any real value to anyone
[2008/06/13 16:43:26] @ Quit: marcoecc: "Quitting"
[2008/06/13 16:43:45] <muerr> Disconnect: Red Hat superiority example: ifcfg-$interface scripts instead of munging everything up in /etc/network/interfaces.
[2008/06/13 16:43:57] <muerr> makes it a lot easier to manage network interfaces with puppet.
[2008/06/13 16:44:01] <muerr> :-)
[2008/06/13 16:44:30] <muerr> i also really like chkconfig / service.
[2008/06/13 16:44:37] <Disconnect> all i asked was for an actual example where it violated the spec.. if the spec is lacking ("core doesn't do anything useful") thats hardly debian's fault
[2008/06/13 16:44:59] <Volcane> muerr: so make a little stub package that provides lp and lpr and then you'll be happy :)
[2008/06/13 16:45:17] <Volcane> Disconnect: i didnt say it violated the spec, i said i *prefer the redhat way*
[2008/06/13 16:45:22] <muerr> Volcane: i'd rather not.
[2008/06/13 16:45:31] <Volcane> Disconnect: you insisted that my preference be somehow linked to a spec
[2008/06/13 16:45:38] <Volcane> muerr: nods :)
[2008/06/13 16:45:39] <Disconnect> interfaces has its ups and downs. tons easier to manage interactively but yah, not as much from ifcfg
[2008/06/13 16:45:41] <muerr> i'd rather operatingsystemrelease confine :centos to get the version out of redhat-release.
[2008/06/13 16:45:45] <Disconnect> er, from scripts that is :)
[2008/06/13 16:45:47] <muerr> like :redhat does :-)
[2008/06/13 16:46:47] <Volcane> muerr: so submit a simple patch in a ticket :)
[2008/06/13 16:46:53] <muerr> Volcane: ya
[2008/06/13 16:47:30] <Disconnect> but seriously.. if lsb-core (packed for deb/ub) doesn't do what you want, but only what lsb requires for lsb-core, either install lsb (everything) or hit up linux standards for a new 'core' that does in fact do stuff you want .. thats not "omg it sucks" thats "hey, its broken just like the spec suggests"
[2008/06/13 16:47:36] @ Quit: sazma: "Ex-Chat"
[2008/06/13 16:47:41] <muerr> is this legal syntax? confine :operatingsystem => :centos, :redhat
[2008/06/13 16:47:56] <Volcane> muerr: i think so
[2008/06/13 16:48:53] <muerr> uhh no
[2008/06/13 16:48:54] <muerr> hehe
[2008/06/13 16:49:05] <muerr> epic fail
[2008/06/13 16:49:10] <Volcane> hehe
[2008/06/13 16:49:10] <Disconnect> lol
[2008/06/13 16:49:23] * Volcane just started reading his ruby book today
[2008/06/13 16:49:29] @ lak_ joined channel #puppet
[2008/06/13 16:49:41] <Disconnect> epic fail is "wow, that reformatted my puppetmaster and now its rampaging around killing the office admin staff, even the cute new intern"
[2008/06/13 16:50:52] <Disconnect> so if i use an external classifier it stops reading node.pp right?
[2008/06/13 16:51:05] <Disconnect> scratch that, node.pp was local :)
[2008/06/13 16:53:16] <benp-> 20:20 * benp- senses an hour long conversation about lsb about to start
[2008/06/13 16:53:21] <benp-> 20:20 * benp- senses an hour long conversation about lsb about to start
[2008/06/13 16:53:26] <benp-> oops
[2008/06/13 16:53:33] <benp-> didnt mean to paste that twice
[2008/06/13 16:53:39] <benp-> anyway, i went away.. and it happened
[2008/06/13 16:54:08] <Volcane> you obviously caused it :P
[2008/06/13 16:54:20] <Disconnect> wacky timezone :)
[2008/06/13 16:54:23] <muerr> lol
[2008/06/13 16:54:29] <muerr> glad i checked the facter 1.5 source.
[2008/06/13 16:54:40] <benp-> pacific time
[2008/06/13 16:54:41] <Disconnect> and random side note, anyone using icpuppet-tiny as a classifier?
[2008/06/13 16:54:52] <Volcane> muerr: hehe is it fixed already?
[2008/06/13 16:54:57] <Disconnect> cuz adding "munin::host" (straight from the manual config) breaks iclassify..
[2008/06/13 16:54:57] <muerr> Volcane: yup
[2008/06/13 16:55:00] <Disconnect> badness
[2008/06/13 16:55:36] <holoway> Disconnect: I'm pretty sure that's not an iclassify bug, but an external nodes bug
[2008/06/13 16:55:58] <Disconnect> iclassify returns a blank page when i hit "edit node" and put that into the (existing) tags list
[2008/06/13 16:56:04] <holoway> no shit?
[2008/06/13 16:56:11] <Disconnect> and icpuppet-tiny (from cmdline) returns nothing until i hit 'back' and save the old list
[2008/06/13 16:56:25] <Disconnect> haven't tested icsearch
[2008/06/13 16:56:27] @ hacim joined channel #puppet
[2008/06/13 16:56:29] <Disconnect> cuz i'm afraid
[2008/06/13 16:56:44] <muerr> Mm... facter-1.5.0rc1.tgz23-May-2008 07:29 - over two weeks since a change has occurred - reasonable to assume 1.5 will be released soon?
[2008/06/13 16:56:53] <holoway> can you file a bug with the output of the production.log in the iclassify jira instance?
[2008/06/13 16:56:54] <hacim> what is a good null operation to do in an exec? I though just 'exit 0' would be good, but it returns 1
[2008/06/13 16:57:28] <benp-> /bin/true
[2008/06/13 16:57:41] <hacim> thanks
[2008/06/13 16:58:08] <Disconnect> sure
[2008/06/13 16:58:25] @ Quit: kambiz: "Leaving"
[2008/06/13 17:01:31] @ Quit: lak: Read error: 113 (No route to host)
[2008/06/13 17:04:06] @ shenson is now known as shenson_not_here
[2008/06/13 17:05:43] @ Innocenti joined channel #puppet
[2008/06/13 17:05:50] @ Quit: jshare: "Leaving."
[2008/06/13 17:06:16] @ jshare joined channel #puppet
[2008/06/13 17:06:37] @ lak_ is now known as lak
[2008/06/13 17:07:48] <muerr> lak: facter-1.5.0rc1.tgz23-May-2008 07:29 - over two weeks since a change has occurred - reasonable to assume 1.5 will be released soon?
[2008/06/13 17:07:56] <lak> heh
[2008/06/13 17:07:57] <lak> well
[2008/06/13 17:08:00] <muerr> uh oh
[2008/06/13 17:08:12] <lak> i can talk to jamesturnbull when he shows up and see if he thinks it's ready
[2008/06/13 17:08:23] <lak> i think he was waiting for freebsd fixes, but they haven't been forthcoming
[2008/06/13 17:08:31] <muerr> As in 'for immediate release'?
[2008/06/13 17:09:15] <lak> sure
[2008/06/13 17:09:41] <muerr> Cool!
[2008/06/13 17:10:37] <lak> i don't suppose you'd be willing to test that the 1.5 version actually fixes factsync, would you?
[2008/06/13 17:10:40] <Volcane> 1.5 adds an option to run facts that was synced via puppet doesnt it?
[2008/06/13 17:10:44] <lak> i tried it, but hey, i've made mistakes before
[2008/06/13 17:10:48] <lak> yeah, and a good bit more
[2008/06/13 17:10:51] <muerr> i'd love to but we don't have a 'reasonable' test environment.
[2008/06/13 17:10:58] <lak> :/
[2008/06/13 17:11:01] <Volcane> nice, i am not on the devel list but i recall that mentioned
[2008/06/13 17:11:12] <lak> well, actually; 1.5 is *mostly* a refactor to provide autoloading
[2008/06/13 17:11:31] <muerr> And I'm really busy trying to fix up a few loose ends and finish turning over all my current project work so I can start the new job.
[2008/06/13 17:11:47] @ cabbiepete joined channel #puppet
[2008/06/13 17:11:58] @ Quit: cabbiepete: Remote closed the connection
[2008/06/13 17:12:20] <Volcane> lak: autoloading?
[2008/06/13 17:12:20] <Disconnect> added #1363
[2008/06/13 17:12:22] <gepetto> Disconnect: #1363 is http://reductivelabs.com/redmine/issues/show/1363
[2008/06/13 17:12:32] <lak> and it will automatically look in puppet's lib directory
[2008/06/13 17:12:50] <Disconnect> so how -do- i specify an iclassify tag that puppet will see as munin::node?
[2008/06/13 17:12:50] <Volcane> lak: yeah that's a good addition, means i can ditch my little path overrides
[2008/06/13 17:13:02] <lak> Disconnect: that looks like an iclassify bug, not a puppet bug
[2008/06/13 17:13:29] <lak> Disconnect: am i wrong about that?
[2008/06/13 17:13:56] <lak> Volcane: what path overrides?
[2008/06/13 17:14:09] <Volcane> lak: so that RUBYLIB=/var/lib/puppet works for loading facts
[2008/06/13 17:14:17] <lak> ah
[2008/06/13 17:14:22] <lak> grr, limechat is driving me crazy
[2008/06/13 17:14:29] <lak> i can't get it to stop bouncing the stupid dock icon
[2008/06/13 17:14:30] <Volcane> means the facts need to be 'facter'
[2008/06/13 17:14:38] <lak> pretty sure i'm going back to colloquy
[2008/06/13 17:14:39] <Disconnect> ...why do 3 apps from the same company use different bugtrackers? yah it belongs in jira. but i'm frustrated enough today, someone else can move it. or i'll get to it on monday.
[2008/06/13 17:14:42] <Volcane> ....need to be in
[2008/06/13 17:14:47] <muerr> lak: irc clients pretty much universally suck
[2008/06/13 17:15:37] <lak> Disconnect: iclassify is not by reductive labs
[2008/06/13 17:15:40] <muerr> Yeah
[2008/06/13 17:15:43] <muerr> I was going to say.
[2008/06/13 17:15:43] @ rcoup joined channel #puppet
[2008/06/13 17:15:54] <muerr> holoway wrote iclassify.
[2008/06/13 17:16:32] <muerr> HJK solutions uses Jira. Reductivelabs uses Redmine now, and used to use Trac.
[2008/06/13 17:16:42] <Disconnect> thought he was with redactive. but yah, i missed the division.
[2008/06/13 17:16:51] <muerr> Ah. Nope, he's w/ HJK
[2008/06/13 17:18:16] <Disconnect> ya
[2008/06/13 17:18:29] <Disconnect> doesn't matter anyway, gotta request an account to file bugs (no wonder there are so few..)
[2008/06/13 17:19:01] <Disconnect> wow and the "administrator" link from that message just leads to names, not emails, contact pages or anything..
[2008/06/13 17:19:01] <muerr> heh heh
[2008/06/13 17:19:12] <Disconnect> (https://jira.hjksolutions.com/secure/Dashboard.jspa?os_destination=%2Fsecure%2FIssueNavigator.jspa if anyone wants to play along at home)
[2008/06/13 17:19:41] <muerr> holoway = Adam
[2008/06/13 17:20:34] <Disconnect> yah but unless you are hanging here, its most of impossible to file a bug. certainly its beyond reasonable - go to the bugtracker, discover you can't, look for an admin, then google to find an email address, then request an account, then (when he gets it and sets it up) log in and file a bug....
[2008/06/13 17:20:59] <Disconnect> i'd rather spend 5 hours filling in arbitrary bugzilla fields than go through that :)
[2008/06/13 17:21:05] <muerr> lol
[2008/06/13 17:21:33] <Disconnect> at least with bugzilla i go in knowing its out to get me
[2008/06/13 17:21:39] <muerr> lol
[2008/06/13 17:21:46] <muerr> We use mantis here.
[2008/06/13 17:21:49] <muerr> and otrs.
[2008/06/13 17:22:04] <lak> Disconnect: well, our bug trackers are hopefully easier than that -- no requesting accounts necessary
[2008/06/13 17:22:07] <muerr> otrs for customers to email in support queries. mantis for internal ticket tracking.
[2008/06/13 17:22:12] <lak> we're just mid-transition, so it can be a touch messy
[2008/06/13 17:22:16] <lak> anyway, gotta run
[2008/06/13 17:22:16] <lak> laters all
[2008/06/13 17:22:40] <Disconnect> this process is like being taken advantage of by a friend-turned-bum .. each request seems reasonable at the time but somehow you still end up with him living on your couch at the end of the day, eating all your food and possibly having taken your girlfriend.
[2008/06/13 17:22:53] <Disconnect> (taken away, not "taken" :) ..)
[2008/06/13 17:23:28] <e^ipi> i had a friend like that
[2008/06/13 17:23:34] <e^ipi> she didn't take my girlfriend though
[2008/06/13 17:23:42] <Disconnect> heh
[2008/06/13 17:31:26] <holoway> Disconnect: sorry, we're still getting things set up for the open source projects
[2008/06/13 17:31:59] <Disconnect> np.. but can i suggest allowing in-band registration or even captcha-protected anon bugfiling? :)
[2008/06/13 17:32:11] <holoway> yeah, that's what it should be
[2008/06/13 17:32:14] @ Quit: jvanzyl_:
[2008/06/13 17:32:50] * Disconnect is a big fan of allowing anon access to everything, moderated by captcha and such, but thats becoming more and more the minority
[2008/06/13 17:33:05] <holoway> Disconnect: we use it for internal-only stuff too
[2008/06/13 17:33:16] <holoway> and we just need to update the permission scheme for the open source stuff
[2008/06/13 17:33:20] <Disconnect> ya
[2008/06/13 17:33:24] <Disconnect> makes sense
[2008/06/13 17:39:03] <muerr> have a good weekend folks
[2008/06/13 17:39:10] @ Quit: muerr: "Leaving."
[2008/06/13 17:39:16] @ Quit: lak: Read error: 113 (No route to host)
[2008/06/13 17:39:20] @ patobrien joined channel #puppet
[2008/06/13 17:45:46] <holoway> Disconnect: you can sign up for an account now
[2008/06/13 17:45:49] <holoway> sorry aobut that
[2008/06/13 17:47:05] <Disconnect> cool
[2008/06/13 17:49:22] @ Quit: jfluhmann: "Open Source Symposium - http://www.texasoss.org/"
[2008/06/13 17:51:53] @ jvanzyl joined channel #puppet
[2008/06/13 17:56:40] @ mmcgrath left channel #puppet ()
[2008/06/13 17:58:16] @ Quit: Wakko666: "Leaving."
[2008/06/13 17:58:26] @ jbrothers1 left channel #puppet ()
[2008/06/13 18:02:08] @ Wakko666 joined channel #puppet
[2008/06/13 18:02:41] @ Quit: plathrop: Remote closed the connection
[2008/06/13 18:09:29] @ Quit: mikepea:
[2008/06/13 18:13:18] @ plathrop joined channel #puppet
[2008/06/13 18:26:27] <chadh> I am getting ready to go on a trip and will be offline a good deal of the time. What's the best way to take the puppet documentation with me?
[2008/06/13 18:26:51] <chadh> is puppetdoc sufficient? do I need the source?
[2008/06/13 18:26:59] <jamesturnbull> chadh: well there is the book but puppetdoc shold be fine
[2008/06/13 18:27:15] <barnum> source is cheap!
[2008/06/13 18:27:17] <holoway> jamesturnbull: come on, man, hard sell!
[2008/06/13 18:27:27] <chadh> jamesturnbull: I have the book too :)
[2008/06/13 18:27:30] <holoway> you couldn't possibly survive with the book, chadh
[2008/06/13 18:27:31] <holoway> :)
[2008/06/13 18:27:42] <jamesturnbull> holoway: I am not a hard sell kinda guy
[2008/06/13 18:27:54] <chadh> I'd really like the whole wiki :)
[2008/06/13 18:28:08] <chadh> but I have like 3 minutes left to grab last minute stuff
[2008/06/13 18:28:31] <chadh> jamesturnbull: so does puppetdoc read from puppet's ruby files?
[2008/06/13 18:28:51] * chadh is not looking forward to being without The Google
[2008/06/13 18:29:21] <chadh> if my blackberry works, I might be able to make due :) And I guess I could spend some time with the fam ;-)
[2008/06/13 18:29:49] <jamesturnbull> chadh: yes - draws from the code
[2008/06/13 18:29:52] <barnum> cool
[2008/06/13 18:38:06] @ Quit: Wakko666: Read error: 104 (Connection reset by peer)
[2008/06/13 18:39:47] @ Wakko666 joined channel #puppet
[2008/06/13 18:41:30] @ Quit: rcoup:
[2008/06/13 18:47:13] @ Quit: markl_: "Lost terminal"
[2008/06/13 19:00:55] @ lak joined channel #puppet
[2008/06/13 19:08:23] @ z00dax_ is now known as z00dax
[2008/06/13 19:13:46] <jamesturnbull> fujin: did you log that ticket re ipaddress_eth0 not working now?
[2008/06/13 19:14:52] @ yure_ joined channel #puppet
[2008/06/13 19:24:07] <jamesturnbull> lak: I don't think I can write to the reductivelabs.com/puppet repo
[2008/06/13 19:24:35] <jamesturnbull> lak: I just tried to push to the 0.24.x branch and it dropped the connection
[2008/06/13 19:24:55] @ Quit: lak: Read error: 113 (No route to host)
[2008/06/13 19:30:29] @ Quit: yure: Read error: 110 (Connection timed out)
[2008/06/13 19:37:58] @ Quit: falsch:
[2008/06/13 19:38:27] @ mikepea joined channel #puppet
[2008/06/13 19:43:48] <ashp> Oh no, I'm turning into a puppet advocate
[2008/06/13 19:43:58] <ashp> I'm writing long replies about how great it is on forums :O
[2008/06/13 19:46:06] @ a-priori_ joined channel #puppet
[2008/06/13 19:46:12] @ Quit: mikepea:
[2008/06/13 19:46:35] @ Quit: Wakko666: Read error: 104 (Connection reset by peer)
[2008/06/13 19:48:18] @ Wakko666 joined channel #puppet
[2008/06/13 19:50:32] <Disconnect> uhoh
[2008/06/13 19:50:40] <Disconnect> quick, pick up a ruby book
[2008/06/13 19:50:50] <Disconnect> if hitting yourself with it doesn't work then its time for drastic measures
[2008/06/13 19:50:51] <Disconnect> read it
[2008/06/13 19:50:52] @ Quit: patobrien: "Leaving"
[2008/06/13 19:50:55] <Disconnect> that'll cure you
[2008/06/13 19:58:29] @ Quit: Wicky656:
[2008/06/13 20:14:00] @ lak joined channel #puppet
[2008/06/13 20:18:48] @ Quit: Innocenti: Read error: 110 (Connection timed out)
[2008/06/13 20:22:10] @ Quit: plathrop: "ERC Version 5.2 (IRC client for Emacs)"
[2008/06/13 20:25:59] @ Quit: jvanzyl: Read error: 104 (Connection reset by peer)
[2008/06/13 20:27:29] @ jvanzyl joined channel #puppet
[2008/06/13 20:34:39] <gileswork> anyone about?
[2008/06/13 20:36:11] @ Quit: andrewcshafer:
[2008/06/13 20:37:02] <Disconnect> nope
[2008/06/13 20:37:58] @ Quit: lak: Read error: 113 (No route to host)
[2008/06/13 20:39:25] <gileswork> heh
[2008/06/13 20:39:31] @ andrewcshafer joined channel #puppet
[2008/06/13 20:40:06] <gileswork> ah s'okay was just a ruby warning non fatal
[2008/06/13 20:43:16] @ Quit: dsch04: Remote closed the connection
[2008/06/13 20:47:39] <gileswork> I want to use facter in a separate ruby file
[2008/06/13 20:47:47] <gileswork> well I want to use confine specifically
[2008/06/13 20:47:51] <gileswork> anyone got any tips?
[2008/06/13 20:59:23] <gileswork> warning: instance variable @have_which not initialized
[2008/06/13 20:59:35] <gileswork> thats the warning I get not sure why yet
[2008/06/13 20:59:56] * Disconnect doesn't grok ruby yet. newb. and it looks like you missed the timezone shuffle
[2008/06/13 21:00:10] <Disconnect> seems to mostly be a .eu crowd, so .eu business hours are best
[2008/06/13 21:05:36] @ tim|imac_ joined channel #puppet
[2008/06/13 21:05:41] @ Quit: tim|imac: Read error: 104 (Connection reset by peer)
[2008/06/13 21:06:20] @ Quit: shake-n-bake_: Read error: 110 (Connection timed out)
[2008/06/13 21:19:00] @ rcoup joined channel #puppet
[2008/06/13 21:25:31] @ keithlard joined channel #puppet
[2008/06/13 21:38:06] @ kenvandine joined channel #puppet
[2008/06/13 21:50:36] @ Quit: kenvandine: "Ex-Chat"
[2008/06/13 21:51:10] @ Quit: a-priori_:
[2008/06/13 22:01:23] @ Quit: lutter: "Leaving."
[2008/06/13 22:03:40] @ Quit: andrewcshafer:
[2008/06/13 22:11:22] <fujin> jamesturnbull: sorry haven't had a chance to yet.
[2008/06/13 22:11:32] <fujin> jamesturnbull: will do it right now before my wife starts putting my dreads in
[2008/06/13 22:12:09] <jamesturnbull> fujin: no worries
[2008/06/13 22:12:12] <jamesturnbull> fujin: dreads?
[2008/06/13 22:12:26] <jamesturnbull> fujin: don't you grow them?
[2008/06/13 22:12:54] * jamesturnbull has way too many hippy friends who don't wash their hair
[2008/06/13 22:13:04] <fujin> naw.
[2008/06/13 22:14:46] <fujin> some do, but they become unmanageable
[2008/06/13 22:14:48] <fujin> www.dreadheadhq.com
[2008/06/13 22:14:50] <fujin> #1365
[2008/06/13 22:14:52] <gepetto> fujin: #1365 is http://reductivelabs.com/redmine/issues/show/1365
[2008/06/13 22:14:56] <fujin> afk for a good 4 hours or so.
[2008/06/13 22:14:57] <fujin> toodles ;)
[2008/06/13 22:19:42] <jamesturnbull> fujin: later mate enjoy the hair additions
[2008/06/13 22:21:14] @ john3 joined channel #puppet
[2008/06/13 22:27:02] @ andrewcshafer joined channel #puppet
[2008/06/13 22:30:10] @ jsgotangco joined channel #puppet
[2008/06/13 22:41:04] @ Quit: keithlard:
[2008/06/13 22:42:41] <fujin> got laptop. wo