Tuesday, 2008-04-29

« Monday, 2008-04-28 Index Wednesday, 2008-04-30 »
[2008/04/29 00:00:17] @ m1n3s6 joined channel #puppet
[2008/04/29 00:02:04] <mikepea> so which value coming from your array is the correct one?
[2008/04/29 00:02:38] <ashp> the array returns two things, one is either production/development, the second is internal/external
[2008/04/29 00:02:48] <ashp> And based on the external/internal returned value, I have to set $puppet
[2008/04/29 00:02:50] <ashp> I hope that makes sense
[2008/04/29 00:03:07] @ Quit: johnf: Read error: 110 (Connection timed out)
[2008/04/29 00:03:13] <mikepea> yup, though does that not break puppet's own $environment support?
[2008/04/29 00:03:59] <ashp> Well, you're supposed to use LDAP to set that environment
[2008/04/29 00:04:07] <ashp> I don't know, to be honest the ldap support seems a little shaky
[2008/04/29 00:04:17] <ashp> the documentation says 'use environment, it'll return an array'
[2008/04/29 00:04:20] <mikepea> i've not used it yet...
[2008/04/29 00:04:22] <ashp> but then there seems to be no way to use that array :/
[2008/04/29 00:04:40] <mikepea> can you point me at the page in the doco where it says that?
[2008/04/29 00:04:49] <huangmingyou> why not use a special char split this vaule, to only one value,like env="enver+external"
[2008/04/29 00:05:09] <huangmingyou> then use ruby get the two value
[2008/04/29 00:05:45] <ashp> I don't know if I can do that within puppet, huangmingyou
[2008/04/29 00:06:08] <mikepea> you can using generate() - see the FAQ on complex conditionals...
[2008/04/29 00:06:28] <ashp> mikepea: http://reductivelabs.com/trac/puppet/wiki/LdapNodes
[2008/04/29 00:06:29] <huangmingyou> in the puppet tempate can do that
[2008/04/29 00:06:49] <huangmingyou> erb support ruby langues,it's so easy
[2008/04/29 00:06:58] <ashp> I guess it just says, Attributes with multiple values will be created as arrays.
[2008/04/29 00:07:21] @ Zothar_Work joined channel #puppet
[2008/04/29 00:08:18] @ lak joined channel #puppet
[2008/04/29 00:08:27] @ yure_ joined channel #puppet
[2008/04/29 00:09:02] <mikepea> yeah. Do you have a second environment for each host in your LDAP then?
[2008/04/29 00:09:37] <mikepea> i suspect that's going to cause you trouble down the track...
[2008/04/29 00:10:23] <mikepea> I think your 'location' schema object sounds like a better plan.
[2008/04/29 00:13:46] @ Quit: andrewcshafer:
[2008/04/29 00:14:39] <duritong> anybody ever used a number in a case statement?
[2008/04/29 00:15:58] <duritong> ah the problem was the number in the class-name
[2008/04/29 00:20:37] @ Quit: yure: Read error: 110 (Connection timed out)
[2008/04/29 00:25:53] @ Quit: roald: Remote closed the connection
[2008/04/29 00:28:05] @ Quit: m1n3s6: "Leaving"
[2008/04/29 00:29:44] <ashp> The only problem with moving stuff into LDAP is losing the ability to define additional network interfaces right in the node def, that was useful..
[2008/04/29 00:38:28] <duritong> well factor it out in a special class
[2008/04/29 00:39:30] <ashp> yeah, that's what i'd probably do
[2008/04/29 00:39:36] <ashp> well i'd make a module of the hostname
[2008/04/29 00:39:48] <ashp> so i end up with hlslinutil02/ as a module name, and just include that and do any host specific stuff in there
[2008/04/29 00:39:57] <ashp> that's not even THAT ugly as it keeps all extra changes localised
[2008/04/29 00:40:17] <ashp> I'm tempted to do away with including any IP information in ldap and just including a module per server
[2008/04/29 00:40:26] <ashp> but then most servers don't need any custom stuff other than eth0 setting
[2008/04/29 00:41:47] <gepetto> ::puppet:: Ticket #284 (enhancement closed): Native xinetd type @ http://reductivelabs.com/trac/puppet/ticket/284#comment:4 (by james@lovedthanlost.net)
[2008/04/29 00:42:21] <tim|mb> ashp: you setup networking from puppet as well?
[2008/04/29 00:44:30] <ashp> oh yeah, I configure eth0 via it
[2008/04/29 00:44:49] <ashp> i actually write /etc/sysconfig/network-scripts/ifcfg-eth0 and sysconfig/network and then restart the interface via scripts
[2008/04/29 00:45:25] <ashp> i wrote my own little module to do it as the offical one is still broken
[2008/04/29 00:49:19] <duritong> if I change something in site.pp, do I have to restart puppetmaster
[2008/04/29 00:49:21] <duritong> ?
[2008/04/29 00:50:31] <ashp> no
[2008/04/29 00:51:52] @ freiheit joined channel #puppet
[2008/04/29 00:51:56] <duritong> hmm well I had the fealing that something changed only then
[2008/04/29 00:52:14] <duritong> but well this might be just a wrong feeling
[2008/04/29 00:52:47] @ felix_da_catz joined channel #puppet
[2008/04/29 00:54:19] @ Quit: kolla: Remote closed the connection
[2008/04/29 00:55:16] @ erikh joined channel #puppet
[2008/04/29 00:58:09] @ stahnma left channel #puppet ("Time for something else....")
[2008/04/29 01:09:09] @ nigelk joined channel #puppet
[2008/04/29 01:10:17] @ andrewcshafer joined channel #puppet
[2008/04/29 01:15:19] @ kolla joined channel #puppet
[2008/04/29 01:15:57] @ Innocenti joined channel #puppet
[2008/04/29 01:18:54] @ shenson_not_here is now known as shenson
[2008/04/29 01:20:45] @ jfluhmann joined channel #puppet
[2008/04/29 01:23:53] @ Quit: [newbie]: Read error: 110 (Connection timed out)
[2008/04/29 01:24:08] @ [newbie] joined channel #puppet
[2008/04/29 01:24:12] @ jfluhmann left channel #puppet ("Open Source Symposium - http://www.texasoss.org/")
[2008/04/29 01:24:26] @ _newbie4 joined channel #puppet
[2008/04/29 01:24:59] @ brscott joined channel #puppet
[2008/04/29 01:25:24] @ Quit: _newbie_: Read error: 110 (Connection timed out)
[2008/04/29 01:27:04] @ brscott left channel #puppet ()
[2008/04/29 01:28:18] @ Quit: kolla: Remote closed the connection
[2008/04/29 01:29:45] @ Quit: |Innocenti|: Read error: 110 (Connection timed out)
[2008/04/29 01:36:25] @ Quit: tim|mb: Read error: 110 (Connection timed out)
[2008/04/29 01:38:11] @ Quit: [newbie]: "KVIrc 3.2.5 Anomalies http://www.kvirc.net/"
[2008/04/29 01:50:26] @ lutter joined channel #puppet
[2008/04/29 01:51:01] @ shenson is now known as shenson_not_here
[2008/04/29 01:52:16] @ Quit: Innocenti: Client Quit
[2008/04/29 01:53:26] @ Quit: _newbie4: Client Quit
[2008/04/29 01:55:39] @ plathrop joined channel #puppet
[2008/04/29 01:57:32] @ fsweetser_ is now known as fsweetser
[2008/04/29 02:02:18] <duritong> can I overwrite by inheritance parameters of a type which is defined in a define?
[2008/04/29 02:13:01] <hacim> arrrrg these checksum errors on facter changes are driving me nuts
[2008/04/29 02:25:57] @ Quit: lak:
[2008/04/29 02:29:34] @ chillitom joined channel #puppet
[2008/04/29 02:36:36] <martha> anyone have a sendmail recipe?
[2008/04/29 02:37:09] * duritong would be happy about one as well
[2008/04/29 02:38:24] <plathrop> My only recipe for sendmail is pakage_manager_of_choice deinstall sendmail
[2008/04/29 02:38:29] <plathrop> :-P
[2008/04/29 02:39:13] @ Quit: jvanzyl:
[2008/04/29 02:40:51] @ shake-n-bake joined channel #puppet
[2008/04/29 02:42:07] <Zothar_Work> just managing the files myself; no provider for sendmail type stuff
[2008/04/29 02:43:23] <gepetto> ::puppet:: Ticket #1208 (enhancement created): Solaris user provider lacks support for roles @ http://reductivelabs.com/trac/puppet/ticket/1208 (by martin.englund@sun.com)
[2008/04/29 02:49:39] @ happymcplaksin joined channel #puppet
[2008/04/29 02:53:20] @ ianm joined channel #puppet
[2008/04/29 02:53:28] @ ezralini joined channel #puppet
[2008/04/29 02:54:15] @ adoom42 joined channel #puppet
[2008/04/29 02:55:51] @ Quit: DerekW: "Leaving"
[2008/04/29 02:57:44] <adoom42> hey guys, I'm trying to upgrade to 0.24.4 and ran into a user issue. I put the details on pastebin: http://pastebin.ca/1000555
[2008/04/29 02:58:14] <adoom42> looks like 0.24.4 doesn't let you declare multiple virtual resources with the same name, whereas 0.23.x did?
[2008/04/29 03:01:43] <hacim> martha: did you find a reliable work-around to the mtime/md5 checksum changes on every run?
[2008/04/29 03:02:01] <hacim> martha: I thought that I just needed to rm my state.yaml and restart puppet, but that doesn't seem to work
[2008/04/29 03:02:10] <martha> hacim: no
[2008/04/29 03:03:28] @ Innocenti joined channel #puppet
[2008/04/29 03:04:17] <hacim> martha: so do you just ignore the message then?
[2008/04/29 03:05:01] <martha> yeah
[2008/04/29 03:05:10] @ Quit: shake-n-bake:
[2008/04/29 03:05:36] <hacim> i did manage to make it go away for a while, but as soon as I changed another fact....
[2008/04/29 03:05:41] @ jvanzyl joined channel #puppet
[2008/04/29 03:06:05] <martha> you have to delete the yaml file again
[2008/04/29 03:06:14] <hacim> a second time?
[2008/04/29 03:06:16] @ Quit: mikepea: Read error: 104 (Connection reset by peer)
[2008/04/29 03:06:25] <hacim> oh, after every fact change, yes
[2008/04/29 03:06:26] <martha> when you change it
[2008/04/29 03:06:33] <hacim> however this time it didn't seem to make a different
[2008/04/29 03:06:36] <hacim> difference
[2008/04/29 03:06:43] <martha> weird
[2008/04/29 03:06:43] <hacim> i was wondering if I had to delete it on the puppetmaster too
[2008/04/29 03:06:53] @ mikepea joined channel #puppet
[2008/04/29 03:09:16] @ Quit: mikepea: Client Quit
[2008/04/29 03:09:22] @ Quit: jvanzyl: Client Quit
[2008/04/29 03:10:23] @ gh joined channel #puppet
[2008/04/29 03:12:25] @ dysinger joined channel #puppet
[2008/04/29 03:12:37] @ randybias joined channel #puppet
[2008/04/29 03:13:17] @ Quit: choffee: Read error: 110 (Connection timed out)
[2008/04/29 03:15:39] @ Quit: dysinger: Client Quit
[2008/04/29 03:22:02] @ dysinger joined channel #puppet
[2008/04/29 03:29:41] @ Taziar joined channel #puppet
[2008/04/29 03:31:57] <ashp> So, does anyone use puppet to manage nagios?
[2008/04/29 03:32:04] <ashp> I'm thinking about beginning to bring that under puppet's control
[2008/04/29 03:33:37] <hacim> i do, I'm using davidS nagios module to do so
[2008/04/29 03:33:51] <hacim> the newer puppet has a native type for nagios, but I think its only nagios1, I have not tried that
[2008/04/29 03:33:53] <ashp> I haven't started to look into it yet, just wanted to hear how well it works
[2008/04/29 03:34:19] <hacim> it works quite well for my purposes
[2008/04/29 03:35:59] <Taziar> I could use a bit of help.. After we kickstart a server we use puppet to create new mount points.. We have a File entry to create the directory with the correct user/permissions, and then the mount point gets mounted.. But of course the mount belongs to root:root and I can't figure out a clean way to get File to run again to update the permissions.. Anyone found a solution to this?
[2008/04/29 03:37:37] <ashp> hmm, can't you use owner/group => in file {} to set the permissions right at the start?
[2008/04/29 03:37:41] <ashp> Or do I misunderstand?
[2008/04/29 03:38:20] <Zothar_Work> either that or the problem you're talking about is not Puppet specific but rather the behavior of mount in general
[2008/04/29 03:38:45] @ Quit: slally: Remote closed the connection
[2008/04/29 03:39:22] <Taziar> Lemme give you an example.. You have /opt as a mount point, and want to also have /opt/test as a new mount point. I use puppet to create /opt/test as a directory, create an ext3 file system, and then mount it to /opt/test.
[2008/04/29 03:40:09] <Taziar> Now before the mount the permissions are correct, but after the mount the permissions are no longer correct, which is standard behaviour and to be expected.. If I could call File again to make sure the permissions are good, that would be great.. But I can't figure out a way to call File again.
[2008/04/29 03:40:31] @ _newbie_ joined channel #puppet
[2008/04/29 03:40:35] <Taziar> I tried subscribe and notify, but end up getting cycle dependcies.
[2008/04/29 03:40:58] <ashp> hmm, so the problem is that when mounting it sets the permissions incorrectly
[2008/04/29 03:41:56] <Taziar> Well the problem is the File stuff applies to /opt/test the directory, not /opt/test the mount point.. But /opt/test has to exist before you can mount something.. I don't technically care about the permissions before it is mounted, but I need to create the directory.
[2008/04/29 03:42:00] <ashp> this is hideous BUT you could do this
[2008/04/29 03:42:15] <ashp> write a tiny script that sets the permissions right, and then do an exec{} with a require => File
[2008/04/29 03:42:16] <Taziar> My hideous solutions is exec to make the dir, mount, and then File to update the permissions :).
[2008/04/29 03:44:39] <ashp> it's ugly but it works :/
[2008/04/29 03:44:45] <ashp> i'm afraid i can't think of a decent solution to it
[2008/04/29 03:45:03] <ashp> maybe just do the mount via the exec, then have file require=> exec?
[2008/04/29 03:45:13] <ashp> i guess that won't work actually, as it wouldn't create the dir, damn
[2008/04/29 03:45:18] <Taziar> Yeah.
[2008/04/29 03:45:50] @ erlingre joined channel #puppet
[2008/04/29 03:46:38] @ Quit: jshar1: "Leaving."
[2008/04/29 03:49:31] <adoom42> hi again, I'm trying to test out the new 0.24.4 version. I started puppetmaster on an alternate port, but when I exec this command from a client (puppetd --test --noop --masterport=8141), it gives this error "err: Could not retrieve catalog: Could not connect to puppet.mydomain.com on port 8140". is --masterport not recognized anymore?
[2008/04/29 03:49:38] <adoom42> it's still in the puppet wiki
[2008/04/29 03:51:35] @ Quit: _newbie_: "KVIrc 3.2.5 Anomalies http://www.kvirc.net/"
[2008/04/29 03:52:20] @ shake-n-bake joined channel #puppet
[2008/04/29 03:56:45] @ jshare joined channel #puppet
[2008/04/29 04:03:18] @ Quit: Taziar: "ircII EPIC4-2.4 -- Are we there yet?"
[2008/04/29 04:05:30] @ mikepea joined channel #puppet
[2008/04/29 04:08:53] @ Quit: felix_da_catz: Connection timed out
[2008/04/29 04:26:47] @ _newbie_ joined channel #puppet
[2008/04/29 04:33:19] @ lak joined channel #puppet
[2008/04/29 04:33:33] @ shadoi joined channel #puppet
[2008/04/29 04:45:34] @ nward joined channel #puppet
[2008/04/29 04:49:59] @ Quit: nward: Client Quit
[2008/04/29 04:50:43] @ Quit: ezralini: Read error: 104 (Connection reset by peer)
[2008/04/29 05:02:53] @ Quit: daork: Read error: 110 (Connection timed out)
[2008/04/29 05:31:30] @ Quit: _newbie_: Read error: 110 (Connection timed out)
[2008/04/29 05:39:23] @ pleemans joined channel #puppet
[2008/04/29 05:45:06] <stick> is there a way to have puppet (standalone) read in a manfiest (.pp) and have it spit back out a yaml representation or something?
[2008/04/29 05:45:18] <stick> lak: ^^
[2008/04/29 05:45:31] @ Quit: erlingre: "Leaving"
[2008/04/29 05:45:39] <lak> huh, no; never thought about providing that
[2008/04/29 05:45:50] <stick> I'm trying to solve a data drift problem
[2008/04/29 05:45:52] <lak> wouldn't be that hard to code it up, but it's not available right now
[2008/04/29 05:46:22] <stick> as we are migrating away from cfengine, our nameserver infrastucture (currently) has our master(s) in cfengine with a subset of slaves in puppet
[2008/04/29 05:47:15] <stick> and we keep running into people adding zones in one place but not the other, so I was trying to hack something quick together that would let me see what's missing between the two
[2008/04/29 05:48:42] <lak> i guess i'd call that non-trivial
[2008/04/29 05:49:13] <lak> can you do that with cfengine? get the info out, i mean?
[2008/04/29 05:49:15] <ashp> Make them use a wrapper that handles it?
[2008/04/29 05:49:19] <ashp> So they can't put them in one place
[2008/04/29 05:49:24] <ashp> or, a large mallet...
[2008/04/29 05:49:41] <bda> Application of violence sounds like the appropriate response.
[2008/04/29 05:49:54] <stick> well it's not directly cfengine it's our perl based template engine that lives on top
[2008/04/29 05:50:20] <stick> so the cfengine file is just straight perl so I can require or eval it and have access to the data structures
[2008/04/29 05:51:10] <stick> well we haven't told everyone to do it in puppet yet, so application of violence isn't warranted (yet)
[2008/04/29 05:51:27] * stick <3 migrations
[2008/04/29 05:55:29] <lak> stick: are you hoping to catch these at commit, or in random reviews, or what?
[2008/04/29 05:55:48] <stick> on demand
[2008/04/29 05:55:53] <lak> because you could probably set the catalog to cache to yaml (which would likely add a bit to your compile time)
[2008/04/29 05:56:03] <lak> then you could just investigate the compiled catalogs, which are already yaml
[2008/04/29 05:56:16] <lak> it'd require modifying puppetmasterd, since it's not currently configurable
[2008/04/29 06:00:44] <stick> if there was something existing I'd use it, but since there's not I'll probably do something hackish with perl, it's a stopgap measure till we move our ns master to puppet
[2008/04/29 06:00:59] <stick> then people can update cfengine all they want and their changes won't go out :)
[2008/04/29 06:01:20] <ashp> can't you just.. tell them to stop using cfengine for it?
[2008/04/29 06:01:22] <stick> then I can use the large mallet approach
[2008/04/29 06:01:32] <stick> not until we migrate the master
[2008/04/29 06:02:00] <stick> we have more zones in cfengine at this point than puppet
[2008/04/29 06:02:43] <stick> plus alot of these updates are done by helpdesk who is going to lose their commit access to the config tree when things move to puppet
[2008/04/29 06:03:09] <stick> they'll get a nicely regimented tool that keeps them from breaking dns and dhcp 3 or 4 times a week
[2008/04/29 06:04:57] <machpo> what's the excitement in that?
[2008/04/29 06:13:15] @ Quit: pleemans: "Ex-Chat"
[2008/04/29 06:21:38] @ Zothar_Work left channel #puppet ()
[2008/04/29 06:21:43] @ Quit: lak:
[2008/04/29 06:44:51] <riddley> trac down for anyone else?
[2008/04/29 06:46:32] @ nward joined channel #puppet
[2008/04/29 06:51:41] @ martha left channel #puppet ()
[2008/04/29 06:58:10] <hacim> yes
[2008/04/29 07:08:18] @ lak joined channel #puppet
[2008/04/29 07:12:37] @ Quit: nward: Read error: 104 (Connection reset by peer)
[2008/04/29 07:15:52] @ nward joined channel #puppet
[2008/04/29 07:17:19] @ Quit: jeckersb: "Leaving"
[2008/04/29 07:19:23] <fujin> Anyone remember what that google guys name is?
[2008/04/29 07:19:30] <fujin> he was talking about yaml vs. marshal a while back
[2008/04/29 07:19:35] <fujin> I wanted to see what the performance differences were
[2008/04/29 07:21:11] <nigelk> hey
[2008/04/29 07:21:11] <nigelk> :)
[2008/04/29 07:21:22] <nigelk> the differences are insane
[2008/04/29 07:21:57] <nigelk> this is the stuff someone here whipped up to show the differences
[2008/04/29 07:22:05] <nigelk> http://pastie.caboo.se/188321
[2008/04/29 07:22:31] <nigelk> 1000 by 1000 array on a desktop machine
[2008/04/29 07:22:34] <nigelk> Marshal: 1.3 sec
[2008/04/29 07:22:40] <nigelk> YAML: 135.6 sec
[2008/04/29 07:22:51] <nigelk> Python is no better
[2008/04/29 07:23:01] <nigelk> I spent a while looking at libyaml, a C implementation
[2008/04/29 07:23:07] <nigelk> but it doesn't look particularly mature
[2008/04/29 07:23:56] <lak> nigelk: did you guys ever try to hack things to go back to using marshal?
[2008/04/29 07:24:06] <nigelk> I did spend a little while on it
[2008/04/29 07:24:18] <nigelk> but then ran into a conceptual problem with how to phase it out to all our clients
[2008/04/29 07:24:33] <nigelk> would have to come up with a fact or something that said "my client is patched to cope with marshal"
[2008/04/29 07:24:35] <holoway> perl does similar things with Storable vs everything else
[2008/04/29 07:24:55] <nigelk> I then just got swamped with other things, and throwing hardware at the problem has hidden it for now
[2008/04/29 07:25:02] <lak> heh
[2008/04/29 07:25:03] <lak> ok
[2008/04/29 07:25:19] <holoway> I wonder what the JSON numbers are like
[2008/04/29 07:25:19] <nigelk> esp given that I got the impression it would only be a temporary solution?
[2008/04/29 07:26:36] <lak> pretty similar to marshal
[2008/04/29 07:27:18] <lak> nigelk: yeah, that's basically true, since 0.25 should at least enable you to add your own formats, maybe with some development
[2008/04/29 07:27:28] <lak> i don't know if i'll be able to get anything other than yaml done to start with
[2008/04/29 07:28:15] <nigelk> lak: that's fine. I think this is a spot where the community needs to step up
[2008/04/29 07:29:11] <lak> that would be nice, but it's infrastructure code, which seems to scare most people away
[2008/04/29 07:29:17] <lak> or at lest, it sits right on the infrastructure code
[2008/04/29 07:29:26] <nigelk> sure, but if it's easy enough to add formats
[2008/04/29 07:29:32] <nigelk> and the existing one covers most use cases
[2008/04/29 07:29:35] <nigelk> I think it's fair enough
[2008/04/29 07:30:27] <fujin> wow, some of those comparisms are insane
[2008/04/29 07:30:29] <fujin> why's it so much fasteR?
[2008/04/29 07:31:20] <nigelk> I don't think the YAML libraries are very well optimized
[2008/04/29 07:31:21] @ jvanzyl joined channel #puppet
[2008/04/29 07:31:43] <nigelk> http://significantbits.wordpress.com/2008/01/29/yaml-vs-marshal-performance/
[2008/04/29 07:31:47] <nigelk> JSON links there
[2008/04/29 07:32:02] <nigelk> JSON does quite well actually
[2008/04/29 07:35:23] <waawaamilk> so their implementations... are they both/all ruby native?
[2008/04/29 07:35:28] <waawaamilk> or are they in c?
[2008/04/29 07:37:28] <fujin> can't really use JSON though nigelk
[2008/04/29 07:37:50] <fujin> unless we're just doing simple serial/deserialisations
[2008/04/29 07:38:08] * fujin not looked at it
[2008/04/29 07:46:19] @ ezralini joined channel #puppet
[2008/04/29 07:46:33] <lak> well, we can use it, we just can't dump ruby objects
[2008/04/29 07:46:44] <lak> which is preferred anyway, so your interfaces are then language-agnostic
[2008/04/29 07:46:54] <lak> because right now all of puppet's interfaces are really only useful for ruby
[2008/04/29 07:48:49] <nigelk> lak: how set are you on the current install.rb script?
[2008/04/29 07:48:58] <lak> not a bit
[2008/04/29 07:48:59] <nigelk> i'm trying to get a fink and MacPorts submission for puppet going
[2008/04/29 07:49:09] <lak> it works, kind of, but it's pretty crappy
[2008/04/29 07:49:12] <nigelk> and it's a bit difficult with the current script. I was thinking of looking at setup.rb
[2008/04/29 07:49:14] <nigelk> ok
[2008/04/29 07:49:17] <nigelk> which seems a bit more flexible
[2008/04/29 07:49:25] <nigelk> http://i.loveruby.net/en/projects/setup/
[2008/04/29 07:49:34] <lak> sure, as long as it's not much work on my part :)
[2008/04/29 07:49:42] <lak> and fink/macports would be great
[2008/04/29 07:49:44] <nigelk> i wouldn't submit it if it involved work for you :)
[2008/04/29 07:50:03] <lak> apparently you googlers are making waves in the apple enterprise group
[2008/04/29 07:50:06] <nigelk> i've got macports going, but the paths all require patches to install.rb as DESTDIR doesn't cut it
[2008/04/29 07:50:07] <nigelk> heh
[2008/04/29 07:50:23] <nigelk> most of them were my buddies before they were swallowed into the mothership
[2008/04/29 07:50:30] <lak> ah
[2008/04/29 07:50:31] <nigelk> sometimes Steve lets them out to chat still
[2008/04/29 07:50:41] <lak> why doesn't DESTDIR work?
[2008/04/29 07:50:58] <nigelk> well you end up with things like /opt/local/usr/bin/puppetd
[2008/04/29 07:51:10] <nigelk> and /opt/local/usr/lib/ruby/site_ruby
[2008/04/29 07:51:11] <nigelk> etc
[2008/04/29 07:51:14] <lak> you want the binaries in one place and the libs in another?
[2008/04/29 07:51:33] <lak> ah, i see; because os x's ruby is in some weird-ass location, as your recent bug shows
[2008/04/29 07:51:47] <nigelk> bindir = "#{destdir}#{Config::CONFIG['bindir']}"
[2008/04/29 07:51:47] <nigelk> nope
[2008/04/29 07:52:00] <nigelk> because destdir only prefixes the bindir from CONFIG
[2008/04/29 07:52:10] <nigelk> and you really want to specify them completely
[2008/04/29 07:52:17] <lak> ah
[2008/04/29 07:52:33] <nigelk> which is why I was looking at setup.rb as it lets you specify with more granularity
[2008/04/29 07:53:19] <nigelk> OSX's ruby is actually in a good spot I think. I'm just not sure whether the bindir from config is meant to represent "where built in binaries are" or 'where you stick your binaries'
[2008/04/29 07:53:31] <nigelk> it's correct for the former, and incorrect for the latter
[2008/04/29 07:54:37] <nigelk> oh. so I have preliminary approval to write up our setup. now just have to write it and then get the actual article approved
[2008/04/29 07:54:45] <lak> awesome
[2008/04/29 07:54:48] <nigelk> but the PR/Open Source teams are into the idea
[2008/04/29 07:55:17] <nigelk> ok. enough procrastination. documentation time.
[2008/04/29 07:55:18] * nigelk shudders
[2008/04/29 07:56:15] <lak> heh
[2008/04/29 07:56:19] <fujin> heh
[2008/04/29 07:56:21] <fujin> I'm doin the same thing
[2008/04/29 07:56:29] <nigelk> I'm in that catch 22 situation
[2008/04/29 07:56:36] <nigelk> I'm making work for myself by not having documentation
[2008/04/29 07:56:45] <nigelk> that is taking up my time I should be spending documenting
[2008/04/29 07:57:55] @ |Innocenti| joined channel #puppet
[2008/04/29 07:58:10] <fujin> I have to flowchart out my recently finished phone system
[2008/04/29 07:58:14] <fujin> i hate flowcharting code
[2008/04/29 07:59:12] @ Quit: ezralini:
[2008/04/29 07:59:51] <fujin> man, the 1000x1000 YAML serializer is still running
[2008/04/29 07:59:55] <fujin> It's been going for like 5 minutes
[2008/04/29 08:00:33] <nigelk> yeah
[2008/04/29 08:00:43] <nigelk> i mean that's a pathological edge case
[2008/04/29 08:00:49] <nigelk> (as far as puppet is concerned)
[2008/04/29 08:01:12] <nigelk> but we still see client YAMLifying taking up the vast majority of our server workload
[2008/04/29 08:01:48] <fujin> mm
[2008/04/29 08:02:13] <fujin> nigelk: any estimate as to how much work converting everything to marshal would be?
[2008/04/29 08:02:50] <fujin> or lak ?
[2008/04/29 08:03:19] <lak> who's paying for the work? :)
[2008/04/29 08:05:39] <lak> as we were discussing earlier, if i can get the work done for 0.25, adding new formats (e.g., marshal) shouldn't be that hard
[2008/04/29 08:05:41] <lak> hours, not days
[2008/04/29 08:05:52] <lak> but that feels like a pretty big 'if' right now
[2008/04/29 08:07:19] <nigelk> we'd do the work if it was a longer term fix before 0.25, but as it stands, I think we'll hold off till then
[2008/04/29 08:08:18] <nigelk> need to upgrade my macs to 0.24 first... and 0.24.4 client not working with 0.24.1 server is holding that off till I upgrade the servers
[2008/04/29 08:09:25] <nigelk> here's a snippet of what we'll be describing on the wiki
[2008/04/29 08:09:28] <nigelk> ./integrate.py --promote gmac_laptop_leopard_unstable
[2008/04/29 08:09:29] <nigelk> Promoting environment gmac_laptop_leopard_unstable to gmac_laptop_leopard_testing..
[2008/04/29 08:09:41] @ _newbie_ joined channel #puppet
[2008/04/29 08:12:46] @ Quit: _newbie_: Client Quit
[2008/04/29 08:15:08] <fujin> lak: oh, no one is paying for it, i was just wondering at an estimate
[2008/04/29 08:15:29] <fujin> to gauge against skill level, something which would take you days would take me weeks, and somethign that'd take you hours would probably take me a weekish.
[2008/04/29 08:16:13] <lak> well, i guess what i mean is, if someone really wants it done, it won't take long; if someone just "wants it" but isn't willing to put time or money into it, it'll probably be slow going
[2008/04/29 08:16:28] <lak> but hopefully my answer above is sufficient
[2008/04/29 08:16:34] <fujin> man, I can't believe how slow to_yaml is on 1000x1000 compared to Marshal - that's insane
[2008/04/29 08:16:41] <fujin> yeah. I'm looking forward to 0.25 :>
[2008/04/29 08:16:54] <fujin> lak: are there any major tickets holding up 0.25's progress?
[2008/04/29 08:17:00] <lak> tickets? no
[2008/04/29 08:17:08] <lak> there's just tons of development
[2008/04/29 08:17:13] <fujin> hehe
[2008/04/29 08:17:45] <shadoi> lak: what do you think of having an external process that tracks the compiled yaml catalogs and stuffs them into the storeconfigs DB rather than doing it at compile time?
[2008/04/29 08:18:26] <fujin> a filewatcher thread would suffice? :>
[2008/04/29 08:18:49] <shadoi> export/collect is the only issue. If a client stores stuff and the tries to collect it in the same run... or any other time-lag scenario... But maybe it won't be a big deal.
[2008/04/29 08:18:52] <fujin> interesting idea though
[2008/04/29 08:19:27] <shadoi> It certainly removes all the overhead associated with storing. And then we can optimize for reads, etc.
[2008/04/29 08:20:22] <lak> yeah, i've been talking about doing something like that for ages
[2008/04/29 08:20:23] <jamesturnbull> fujin: the major ticket holding up 0.25 is #583
[2008/04/29 08:20:27] <gepetto> jamesturnbull: fujin: #583 is http://reductivelabs.com/trac/puppet/ticket/583 "Puppet should use RESTful http instead of XMLRPC"
[2008/04/29 08:20:32] <jamesturnbull> fujin: :)
[2008/04/29 08:20:33] <lak> it only really helps if you want to devote another machine to it, tho
[2008/04/29 08:20:44] <fujin> HA
[2008/04/29 08:20:49] <lak> because fundamentally, if you've got the cycles, great, and if you don't, your machine's hammered regardless
[2008/04/29 08:21:00] <lak> so it might help some, but... probably not atll that much
[2008/04/29 08:21:20] <shadoi> I think just decoupling it from the compile process would allieviate a lot of performance issues.
[2008/04/29 08:21:39] <shadoi> at least then you can take advantage of multi-core machines, etc.
[2008/04/29 08:22:02] <fujin> Doesn't it already take advantage of multi-core machines? It's always hammering all of my cores on my master VM
[2008/04/29 08:22:05] <fujin> ls
[2008/04/29 08:22:06] <shadoi> o
[2008/04/29 08:22:10] <fujin> wrong term
[2008/04/29 08:22:26] <shadoi> puppetmasterd isn't threaded
[2008/04/29 08:22:35] <fujin> mm, true
[2008/04/29 08:22:58] <fujin> fork()!
[2008/04/29 08:23:23] <shadoi> anyway, I can do it very easily if it's a good way to go.
[2008/04/29 08:24:34] <lak> shadoi: unless you want to try to shoehorn it into a 0.24.5 release (which does look somewhat likely), it's probably worth just switching the code over to the indirector stuff
[2008/04/29 08:24:42] <lak> in fact, if you would start on that, it'd be awesome
[2008/04/29 08:24:49] <lak> this week should see me finish all of the plumbing
[2008/04/29 08:24:55] <shadoi> lak: oh ok
[2008/04/29 08:25:01] <lak> so it'll then just be a question of migrating the rest of the networked classes
[2008/04/29 08:25:46] <shadoi> lak: are we going to keep going with the config_stores concept?
[2008/04/29 08:25:58] <lak> and really, you could do the work in the 0.24.x branch, use a yaml cache for the catalogs, and then people could do whatever the heck they wanted
[2008/04/29 08:25:59] <lak> nope
[2008/04/29 08:26:04] <lak> i generalized that idea into the indirector
[2008/04/29 08:26:08] <shadoi> ok
[2008/04/29 08:27:19] @ nward_ joined channel #puppet
[2008/04/29 08:28:24] <holoway> shadoi: external yaml parser++
[2008/04/29 08:29:01] <shadoi> lak: feel free to purge that config_stores directory from git
[2008/04/29 08:29:08] <lak> ok
[2008/04/29 08:29:22] <lak> lemme know if you think you'll have time to work on some of this, i can give you the run-down
[2008/04/29 08:29:27] <lak> over the phone or whatever
[2008/04/29 08:29:56] <shadoi> lak: ok, I'll see what my time is like in the next week. things are kind of crazy, leaving stanford and all.
[2008/04/29 08:30:08] <lak> so it's decided? wasn't sure
[2008/04/29 08:30:27] <shadoi> yeah, last day is the 15th
[2008/04/29 08:30:52] @ adoom42 left channel #puppet ()
[2008/04/29 08:31:06] @ Quit: nward: Read error: 104 (Connection reset by peer)
[2008/04/29 08:31:52] <lak> ok, i'll keep that in mind :)
[2008/04/29 08:32:03] <nigelk> as a higher ed escapee I congratulate you :)
[2008/04/29 08:32:09] <shadoi> nigelk: hehe
[2008/04/29 08:32:29] <shadoi> only took me 2 months to figure it out. :)
[2008/04/29 08:32:37] <shadoi> well 3 I guess.
[2008/04/29 08:32:38] <nigelk> took me 9+ years :)
[2008/04/29 08:32:58] <nigelk> oh. the leaving day.
[2008/04/29 08:33:02] <shadoi> yeah it's pretty crazy... some guys have been here 30+ years.
[2008/04/29 08:33:28] <nigelk> we had long service leave at 10 years, 3 months payout or holidays
[2008/04/29 08:33:35] <nigelk> I was stressing cause I left at 9.3 years
[2008/04/29 08:33:39] <jamesturnbull> lak: I presume the idea would be to make Marshall integration as easy and dropping in the replacement "require"
[2008/04/29 08:33:42] <jamesturnbull> ?
[2008/04/29 08:34:08] <lak> well, you'd need to define how marshal works somewhere
[2008/04/29 08:34:21] <jamesturnbull> nigelk: I worked in a research instititue - as close to academia as I wanted to get - and that scared me back into corporate pretty damn quick. :)
[2008/04/29 08:34:22] <shadoi> maybe a generic Serialize class
[2008/04/29 08:34:26] <lak> i'm expecting a 'format' class, where you code up how to do the serialize/deserialize
[2008/04/29 08:34:27] <shadoi> that can do either
[2008/04/29 08:34:43] <nigelk> I went from IBM Global Services to an Art/Design College
[2008/04/29 08:34:45] <lak> and then a 'format' setting that you use to pick your default format
[2008/04/29 08:34:45] <jamesturnbull> shadoi: happy to help - I had a bit of a go at this but failed dismally
[2008/04/29 08:34:47] <nigelk> culture shock from hell
[2008/04/29 08:34:58] <shadoi> jamesturnbull: cool
[2008/04/29 08:34:58] <jamesturnbull> shadoi: will write tests if that helps.. :)
[2008/04/29 08:35:03] <shadoi> hehehe
[2008/04/29 08:35:09] <lak> altho some things (e.g., files) would always choose their own formats
[2008/04/29 08:35:15] <shadoi> that would be extremely helpful actually.
[2008/04/29 08:35:18] <jamesturnbull> nigelk: GS to a uni? wow
[2008/04/29 08:35:26] <jamesturnbull> nigelk: talk about sloooowwwwww-doooowwwnnnnn
[2008/04/29 08:35:28] <nigelk> the uni was much more my kind of place
[2008/04/29 08:35:43] <nigelk> and I didn't get into trouble for taking my suit jacket off
[2008/04/29 08:35:44] <lak> at this point, one change i'm planning on making to the plumbing is to have the REST uri always be /$api_version/$environment/$format/
[2008/04/29 08:35:57] <lak> which makes the uri longer, but those are three critical pieces of metadata
[2008/04/29 08:36:04] <jamesturnbull> nigelk: that's one thing I like about my current employer - no dress code bullshit
[2008/04/29 08:36:07] <holoway> lak: ideally, you could leave $format off and let the Accept header do it
[2008/04/29 08:36:18] * waawaamilk has never worn a suit ;)
[2008/04/29 08:36:30] <nigelk> I'm trying to make my new guy stop wearing collared shirts. he's making me uncomfortable
[2008/04/29 08:36:32] @ Quit: sdodson: Read error: 104 (Connection reset by peer)
[2008/04/29 08:36:38] <lak> holoway: yeah, i've thought about that; just not sure if i'll be able to pull it off
[2008/04/29 08:36:48] <shadoi> lak: yeah that makes sense.. or even just have a header/metadata with that info.
[2008/04/29 08:36:50] <lak> nigelk: just start cutting them off
[2008/04/29 08:36:58] <lak> am i hearing volunteers? :)
[2008/04/29 08:37:07] <holoway> lak: I've got some perl code that would be easy to translate for parsing the Accept header
[2008/04/29 08:37:17] * nigelk tries to work out how to cut off collars over VC
[2008/04/29 08:37:19] <holoway> if it would come in handly
[2008/04/29 08:37:21] <shadoi> lak: I totally would love to, just don't know how much time I can focus on it.
[2008/04/29 08:37:27] <fujin> damn, broke me apt mirror
[2008/04/29 08:37:31] <holoway> but you want to be able to override it annyway
[2008/04/29 08:37:37] <holoway> with either /$format or .$format
[2008/04/29 08:37:47] <shadoi> .$format is rails standard
[2008/04/29 08:37:50] <shadoi> that would be cool to follow
[2008/04/29 08:37:52] <holoway> so it's a "nice to have" not a "must have"
[2008/04/29 08:38:10] <holoway> (if you're being a RESt weenie, .$format is right, and /$format is wrong)
[2008/04/29 08:38:23] <lak> holoway: well, you can see all of the related code at indirector/rest.rb, network/http/{webrick,mongrel}/rest.rb
[2008/04/29 08:38:29] <lak> can't do $format
[2008/04/29 08:38:39] <lak> what if someone is pulling down /my/file.pdf with puppet?
[2008/04/29 08:38:48] <lak> throw an exception because i don't support the pdf format?
[2008/04/29 08:38:49] <jamesturnbull> nigelk: you have secuirty right? "Here's $20 - go cut collar off - no no collar - the thing around ... no no back off good security..."
[2008/04/29 08:39:00] <nigelk> heh
[2008/04/29 08:39:08] @ _newbie_ joined channel #puppet
[2008/04/29 08:39:13] <holoway> lak: nah, it's a negotation
[2008/04/29 08:39:38] <lak> well, if you can point me to info, or start writing it up, i'm all ears
[2008/04/29 08:39:47] <lak> formats is my next step, once i finish the webrick ssl stuff
[2008/04/29 08:39:59] <holoway> ok -- reading material is easy
[2008/04/29 08:40:04] <holoway> one sec
[2008/04/29 08:40:21] <shadoi> lak: thinking along rails lines, the file controller just wouldn't support $formats.
[2008/04/29 08:40:30] <holoway> shadoi: exactly
[2008/04/29 08:41:04] <lak> well, i'm all for help
[2008/04/29 08:41:07] * jamesturnbull thankfully knows very little about rails
[2008/04/29 08:41:08] <lak> the code is there for everyone to see
[2008/04/29 08:41:13] <holoway> ya
[2008/04/29 08:41:19] <holoway> just giving unsolicited advice
[2008/04/29 08:41:20] <shadoi> lak: I'll definitely take a look and try to help.
[2008/04/29 08:41:37] <lak> i will say, tho, i'm not exactly thrilled at the idea of copying much in rails these days, given how much it considers itself its own little isolated world
[2008/04/29 08:41:48] <holoway> http://www.catalystframework.org/calendar/2006/9
[2008/04/29 08:41:59] <holoway> the beginning of that article has a good description of the REST triangle
[2008/04/29 08:42:14] <shadoi> lak: I'm just thinking more of doing it in a way that a REST client lib can consume the service by default.
[2008/04/29 08:42:25] <lak> yeah
[2008/04/29 08:42:29] <holoway> (and links to other places that describe it in more detail)
[2008/04/29 08:42:42] <holoway> if your'e looking for a place to copy, merb has excellent support for this stuff
[2008/04/29 08:42:45] <holoway> in a smarter way than rails
[2008/04/29 08:42:47] <holoway> fwiw
[2008/04/29 08:43:02] <lak> one of the problems with the current rest code is that it uses encoded exceptions instead of returning 40whatever
[2008/04/29 08:43:07] <lak> cool, i'll keep that in mind
[2008/04/29 08:43:11] <holoway> ah
[2008/04/29 08:43:24] <shadoi> merb routing is quite nice, yeah.
[2008/04/29 08:43:27] <holoway> yeah, using the status codes is a requirement for most REST consumers
[2008/04/29 08:43:37] <lak> yeah, i know
[2008/04/29 08:43:41] <lak> the routing is done
[2008/04/29 08:43:51] <lak> because it's one to one -- you want nodes, you do /node/
[2008/04/29 08:43:59] <lak> there's no other reasonable way to do it in this architecture
[2008/04/29 08:44:08] <holoway> yep
[2008/04/29 08:44:09] <lak> well, there's some plurality, too, i guess, so /nodes/ works
[2008/04/29 08:44:23] <lak> i'm not looking for some way for people to keep adding new services
[2008/04/29 08:44:38] <lak> you want to make a service visible, it has to have a name, and that name can't conflict with any other names, and done
[2008/04/29 08:44:40] <shadoi> /node/class/package/package_version/ensure? ;)
[2008/04/29 08:45:09] <lak> holoway: cool, thanks for the resource
[2008/04/29 08:45:21] <holoway> I've got lots of them
[2008/04/29 08:45:23] <lak> shadoi: no, that's actually a resource
[2008/04/29 08:45:25] <holoway> and I've written at least one REST layer
[2008/04/29 08:45:31] <holoway> from scratch
[2008/04/29 08:45:32] <fujin> hacim: init script is broken!
[2008/04/29 08:45:35] <holoway> so feel free to holler at me
[2008/04/29 08:45:36] <lak> qualified by node and stuff
[2008/04/29 08:45:37] <shadoi> lak: yeah I know, I want to allow puppetshow to do that sort of thing
[2008/04/29 08:45:38] <lak> ok
[2008/04/29 08:45:43] <lak> yeah, it would be nice
[2008/04/29 08:45:49] <fujin> hacim: /etc/init.d/puppetmaster: 98: let: not found
[2008/04/29 08:46:20] <lak> at this point, tho, it's basically going to be, for all cases, something like /$thing/$name and that's it, where thing == {catalog|node|resource|report|etc}
[2008/04/29 08:46:28] <shadoi> fujin: let is a bashism
[2008/04/29 08:46:35] <shadoi> for arithmetic
[2008/04/29 08:46:37] <shadoi> etc.
[2008/04/29 08:46:38] <fujin> ah
[2008/04/29 08:46:43] <fujin> It's cause of Ubuntu's dash again
[2008/04/29 08:46:45] <fujin> stupid frickin dash
[2008/04/29 08:46:57] <holoway> lak: that's good
[2008/04/29 08:47:02] <holoway> you can go crazy letting it get more specific
[2008/04/29 08:47:09] <fujin> that's better. removed dash.
[2008/04/29 08:47:11] <holoway> and most often, it's fine to parse the results as the client
[2008/04/29 08:47:30] <holoway> plus, if you expose the finer-grained resources, it's only a minute and a half before someone wants to update them
[2008/04/29 08:47:35] <holoway> (resource in the REST sense, not in the puppet sense)
[2008/04/29 08:47:46] <holoway> /user/bob is good
[2008/04/29 08:47:59] <holoway> /user/bob/fullname is a pita
[2008/04/29 08:48:07] <holoway> with not a ton of benefit, imnsho
[2008/04/29 08:48:58] <lak> well, you'll be able to update some of these resources, but it'll need to be a full put, not a post
[2008/04/29 08:49:02] <holoway> (since, if youu want to change bob's full name, you should be GETing bob, altering the data, and POSTing the update back)
[2008/04/29 08:49:17] <lak> e.g., /facts/$nodename will be put on the server from the client, as will /report/$nodename
[2008/04/29 08:49:25] <lak> right
[2008/04/29 08:49:35] <lak> that's what i'm doing
[2008/04/29 08:49:36] <shadoi> holoway: all the ActiveRecordish types do that that I've seen
[2008/04/29 08:49:53] <holoway> but doing GET /user/bob/fullname, then POST /usr/bob/fullname
[2008/04/29 08:50:07] <holoway> is neat, but not worth it in bang-for-the-buck land most of the time
[2008/04/29 08:50:18] <holoway> shadoi: yeah?
[2008/04/29 08:51:01] <lak> and really, it breaks all kinds of design rules
[2008/04/29 08:51:10] <holoway> yep
[2008/04/29 08:51:18] <shadoi> holoway: ActiveResource, and RestClient, but I haven't looked super closely.
[2008/04/29 08:52:14] <shadoi> holoway: I actually wrote some perl yesterday.
[2008/04/29 08:52:17] <shadoi> haha
[2008/04/29 08:52:19] <shadoi> it was horrifying
[2008/04/29 08:52:32] <holoway> shadoi: M00se will save you
[2008/04/29 08:52:44] <shadoi> this was just a dumb irssi script
[2008/04/29 08:52:46] <holoway> although I write much more ruby these days than perl
[2008/04/29 08:52:49] <shadoi> to use ruby-growl
[2008/04/29 08:59:49] @ Quit: nward_: Read error: 110 (Connection timed out)
[2008/04/29 09:00:00] <randybias> I started to write a Perl script once...
[2008/04/29 09:00:09] <randybias> but it was nasty by the 5th line
[2008/04/29 09:00:20] <randybias> so I scrapped it in favor of a ruby script
[2008/04/29 09:00:28] <randybias> and it was purdy
[2008/04/29 09:00:32] <waawaamilk> you go to five lines?
[2008/04/29 09:00:35] <waawaamilk> s/go/got/
[2008/04/29 09:00:48] <randybias> I don't know what I was thinking...
[2008/04/29 09:00:56] <waawaamilk> all perl can be done in one line ;)
[2008/04/29 09:01:04] * fujin beats randybias with a Perl book
[2008/04/29 09:01:06] <holoway> randybias: all ya'll are total weenies
[2008/04/29 09:01:10] <holoway> :)
[2008/04/29 09:01:14] <waawaamilk> have you heard of 'perl golf' ?
[2008/04/29 09:01:23] <randybias> Perl hurts my eyes
[2008/04/29 09:01:33] <randybias> makes them bleed...
[2008/04/29 09:01:40] <waawaamilk> you have a simple problem to solve, and you try solve it in the least number of characters possible
[2008/04/29 09:01:48] * holoway shows randybias some php 5
[2008/04/29 09:01:53] <fujin> perl obfuscation ftw!
[2008/04/29 09:01:54] <randybias> waawaamilk: naw man, that's awk
[2008/04/29 09:01:56] <randybias> ;)
[2008/04/29 09:02:08] * randybias explodes from PHP overload
[2008/04/29 09:02:23] <waawaamilk> damn php
[2008/04/29 09:02:27] <randybias> only thing worse is Java
[2008/04/29 09:02:29] * waawaamilk has to write it all day long
[2008/04/29 09:02:47] <shadoi> PHP is fine if you don't use for anything but simple dynamic web pages.
[2008/04/29 09:02:57] <nigelk> I notice no-one is talking Python around lak :)
[2008/04/29 09:03:07] <randybias> see, but when I see Perl it feels like the code is literally leaping out and assaulting me physically
[2008/04/29 09:03:07] <fujin> who talks python anyway?!
[2008/04/29 09:03:08] <lak> eh, hurt yourselves all you want, i don't care
[2008/04/29 09:03:13] <lak> and python does do some things correctly
[2008/04/29 09:03:16] * nigelk hearts Python
[2008/04/29 09:03:17] <lak> e.g., i mostly like docutils
[2008/04/29 09:03:27] <lak> and i wish there were a ruby rst2html converter
[2008/04/29 09:03:33] <randybias> python does one thing well in particular
[2008/04/29 09:03:34] <shadoi> I've seen some pretty perl before. It's as rare as the people who write it.
[2008/04/29 09:03:34] <lak> it would make jamesturnbull's life quite a bit easier
[2008/04/29 09:03:37] <lak> and mine, i guess
[2008/04/29 09:03:39] <randybias> which is trac
[2008/04/29 09:03:42] <randybias> and that's about it
[2008/04/29 09:03:43] <lak> heh
[2008/04/29 09:03:47] <lak> ask jamesturnbull about that :)
[2008/04/29 09:03:47] <waawaamilk> rst is teh nice
[2008/04/29 09:03:48] * randybias starts dodging the flames
[2008/04/29 09:03:54] <fujin> lak, jamesturnbull: how's redmine coming along?
[2008/04/29 09:04:06] <waawaamilk> but why would you write a converter in ruby?
[2008/04/29 09:04:14] <lak> well, all of our docs are either in rst or trac's custom "i hate my users" format
[2008/04/29 09:04:21] <fujin> haha
[2008/04/29 09:04:25] <fujin> <3 trac native markup
[2008/04/29 09:04:28] <lak> so if we change, then we need to convert everything or support rst in ruby
[2008/04/29 09:04:40] <waawaamilk> can't you just shell out to docutils? :)
[2008/04/29 09:04:44] <shadoi> trac's markup is much simpler than rst. heh
[2008/04/29 09:04:47] <lak> waawaamilk: unclear
[2008/04/29 09:04:47] <waawaamilk> would save a lot of effort..
[2008/04/29 09:04:49] <fujin> indeed shadoi
[2008/04/29 09:04:56] <fujin> that doesn't make it anymore portable, though
[2008/04/29 09:04:59] <lak> simpler? yes; portable? no
[2008/04/29 09:05:02] <shadoi> true.
[2008/04/29 09:05:06] <fujin> RST has a pretty steep curve though
[2008/04/29 09:05:09] <waawaamilk> define portable in this context?
[2008/04/29 09:05:10] <lak> really?
[2008/04/29 09:05:15] <lak> waawaamilk: rst is part of docutils
[2008/04/29 09:05:15] <waawaamilk> nah, rst is easy
[2008/04/29 09:05:23] <lak> so you can convert it to almost anything
[2008/04/29 09:05:35] <lak> tex, html being the big ones
[2008/04/29 09:05:37] <shadoi> let's just write everything in TeX
[2008/04/29 09:05:39] <waawaamilk> yep
[2008/04/29 09:05:56] <lak> what was someone saying about a learning curve?
[2008/04/29 09:05:58] <waawaamilk> err.. so are you arguing that it's good to just stick with docutils? ;)
[2008/04/29 09:05:59] <shadoi> haha
[2008/04/29 09:06:02] <fujin> restructured text, lak
[2008/04/29 09:06:08] <shadoi> that comment was oozing sarcasm.
[2008/04/29 09:06:25] <lak> indeed
[2008/04/29 09:06:31] <lak> fujin: not compared to tex :)
[2008/04/29 09:06:52] <lak> and i agree with waawaamilk -- for simple things it's simple, and for harder things... well, it's harder
[2008/04/29 09:06:54] <fujin> heh
[2008/04/29 09:07:07] @ Quit: |Innocenti|: Client Quit
[2008/04/29 09:07:14] @ _newbie_ is now known as f--z
[2008/04/29 09:07:23] <waawaamilk> my website is written in php, but I write all my blogposts in rst in vim, and compile them to html ;)
[2008/04/29 09:07:27] @ |Innocenti| joined channel #puppet
[2008/04/29 09:07:33] <lak> rst is the only format i've seen that at least approaches html's formatting capabilities in a far simpler and easier to read format
[2008/04/29 09:07:49] * waawaamilk can't stand wysiwyg editors or most markup languages, but rst is nice..
[2008/04/29 09:07:49] <nigelk> the only thing that annoys me about rst is that it hasn't won completely
[2008/04/29 09:07:52] <lak> i do the same, but my blog natively supports rst
[2008/04/29 09:07:58] <nigelk> and thus we have all these pseudo markup languages around
[2008/04/29 09:07:58] * lak == nigelk
[2008/04/29 09:08:02] <lak> yeah
[2008/04/29 09:08:11] <lak> like textile: "it's like html, but not as useful"
[2008/04/29 09:08:13] <nigelk> markdown, textile, bluecloth, redcloth, ARGH!
[2008/04/29 09:08:19] <nigelk> yeah
[2008/04/29 09:08:38] <fujin> jamesturnbull: ping?
[2008/04/29 09:09:06] <lak> btw everyone, i still have at least 3-4 slots open for training in june in portland
[2008/04/29 09:09:11] <fujin> Anyone here using puppet in a 'always noop' mode?
[2008/04/29 09:09:19] <fujin> & tag mailing?
[2008/04/29 09:09:24] <lak> whaymond does, but he doesn't seem to spend much time on irc
[2008/04/29 09:09:28] <lak> stupid banks :)
[2008/04/29 09:12:39] <shadoi> I like markdown, I write most of my text docs in that format without thinking about it.
[2008/04/29 09:13:08] <nigelk> yeah. that's the thing I liked about it. It already fit my text file format I use
[2008/04/29 09:15:23] @ pipegeek joined channel #puppet
[2008/04/29 09:31:05] @ johnf joined channel #puppet
[2008/04/29 09:38:41] <holoway> has anyone seen puppet leak shm sempahore's before?
[2008/04/29 09:39:18] <holoway> that seems so very unlikely
[2008/04/29 09:41:15] <holoway> huh - apache was leaking sempahores (as the puppet user, natch)
[2008/04/29 09:44:25] <fujin> heh
[2008/04/29 09:44:31] <fujin> How do I specify --noop on the cmd line?
[2008/04/29 09:44:34] <fujin> --noop false doesn't seem to be working
[2008/04/29 09:44:39] <fujin> oh, --no-noop.
[2008/04/29 09:46:07] <fujin> using tagmails..
[2008/04/29 09:46:10] <fujin> seeing this stuff Tue Apr 29 11:44:52 +1200 2008 //File[/var/lib/puppet/modules]/checksum (notice): is {mtime}Tue Apr 29 11:33:56 +1200 2008, should be time (noop)
[2008/04/29 09:46:17] @ Quit: lak:
[2008/04/29 09:46:20] <fujin> how can I make it ignore checksums?
[2008/04/29 09:46:37] <fujin> tag them with checksum, and then specify all, !checksum: in tagmail?
[2008/04/29 09:47:56] @ Quit: |Innocenti|: Read error: 110 (Connection timed out)
[2008/04/29 09:52:44] @ kolla joined channel #puppet
[2008/04/29 09:57:55] <fujin> so
[2008/04/29 09:57:56] <fujin> Many people are just using the store report and writing an external report processor that processes many reports at once and produces summary pages. It's easiest if these processors are written in Ruby, since you can just read the YAML files in and de-serialize them into Ruby objects. Then, you can just do whatever you need with the report objects.
[2008/04/29 09:58:01] <fujin> Who are these people??!
[2008/04/29 09:58:02] * fujin looks around
[2008/04/29 09:58:28] * randybias raises hand
[2008/04/29 09:58:36] <fujin> Can I take a look? :)
[2008/04/29 09:58:46] <randybias> At our report code?
[2008/04/29 09:58:54] <fujin> Yeah?
[2008/04/29 09:58:55] <randybias> Uh...
[2008/04/29 09:58:58] <randybias> I don't know.
[2008/04/29 09:59:01] <randybias> That's a good question.
[2008/04/29 09:59:05] <fujin> What does it do?
[2008/04/29 09:59:08] <randybias> I have to think about it.
[2008/04/29 09:59:10] @ Quit: johnf: Read error: 110 (Connection timed out)
[2008/04/29 09:59:14] <randybias> Makes the output 'pretty'
[2008/04/29 09:59:23] <fujin> I just wanna write somethign that deserializes it, amkes it pretty and emails it off
[2008/04/29 09:59:26] <fujin> heh
[2008/04/29 09:59:32] <fujin> want!
[2008/04/29 10:00:40] <shadoi> fujin: we have some that stuff reports into mysql that you can look at
[2008/04/29 10:00:53] <fujin> cool, probably a little superfluous, bit I'd appreciate it
[2008/04/29 10:01:00] @ nward joined channel #puppet
[2008/04/29 10:01:22] <shadoi> fujin: yeah the plan is to have it go into the storeconfigs db, but that's all a little up in the air right now.
[2008/04/29 10:01:40] <fujin> I see.
[2008/04/29 10:02:30] <randybias> fujin: I think I have to say no for right now
[2008/04/29 10:02:37] <fujin> That's cool mate, sensitive?
[2008/04/29 10:02:45] @ BarnacleBob joined channel #puppet
[2008/04/29 10:03:34] <fujin> I guess it'll be a good test of my Ruby-fu
[2008/04/29 10:03:39] <fujin> (which is slowly getting better)
[2008/04/29 10:03:45] <randybias> fujin: CloudScale hasn't really been surfaced publicly yet. Various considerations we have to be aware of before we start showing parts of it.
[2008/04/29 10:04:10] <BarnacleBob> hello, i was happily running puppet to manange all my fedora needs when the mgmt comes down with its time to move to centos. i've run into a problem where none of the remote files that should be copied over work. all i get is this: debug: Calling fileserver.describe
[2008/04/29 10:04:10] <BarnacleBob> err: /Main[top]/Node[cron03]/Node[cronrole]/Node[postfix]/Rfile[/etc/postfix/postfix-script]/File[/etc/postfix/postfix-script]/source: Could not describe /files/domain.com/cron03/etc/postfix/postfix-script: Cannot currently copy links
[2008/04/29 10:04:14] <BarnacleBob> anyone know whats up?
[2008/04/29 10:04:16] <fujin> err: //Node[homepages]/generic-systems/Cron[puppetd]: Failed to retrieve current state of resource: No command, somehow at /etc/puppet/manifests/classes/systems.pp:57
[2008/04/29 10:04:59] <fujin> BarnacleBob: version?
[2008/04/29 10:05:00] <shadoi> BarnacleBob: can we see systems.pp?
[2008/04/29 10:05:10] <shadoi> ah yeh that's a bug fixed in 0.24.4
[2008/04/29 10:05:13] <fujin> systems.pp is me :)
[2008/04/29 10:05:28] <shadoi> oh... lol
[2008/04/29 10:05:32] <BarnacleBob> hrm
[2008/04/29 10:05:34] * shadoi goes cross-eyed
[2008/04/29 10:05:50] <BarnacleBob> its 24.5-1.el5
[2008/04/29 10:05:55] <BarnacleBob> er
[2008/04/29 10:05:59] <BarnacleBob> 24.4-1.el5
[2008/04/29 10:06:09] <fujin> err: Could not prefetch cron provider 'crontab': No command, somehow at /etc/puppet/manifests/classes/systems.pp:57
[2008/04/29 10:06:25] <fujin> That's freakin weird. Cron's have been workign fine for ages.
[2008/04/29 10:06:55] <BarnacleBob> shadoi, not sure what you want to see in systems.pp i don't have one....
[2008/04/29 10:07:23] <shadoi> BarnacleBob: yeah that was fujin's, I merged to two together, sorry.
[2008/04/29 10:07:28] <BarnacleBob> ah :)
[2008/04/29 10:07:38] <BarnacleBob> well this "bug" how do i make it go away if i'm running 24.2
[2008/04/29 10:07:39] <BarnacleBob> er
[2008/04/29 10:07:40] <BarnacleBob> 24.4
[2008/04/29 10:07:42] <BarnacleBob> already
[2008/04/29 10:08:23] <shadoi> BarnacleBob: need to see the syntax you're using for that file resource
[2008/04/29 10:08:58] <BarnacleBob> http://rafb.net/p/lJdMAt55.html
[2008/04/29 10:10:54] <fujin> you need to use the links => setting
[2008/04/29 10:10:57] <fujin> manage, follow
[2008/04/29 10:11:20] <fujin> shadoi: so, can I take a look at your reports procesor? arjuna.christensen@maxnet.co.nz / aj@junglist.gen.nz
[2008/04/29 10:11:38] <BarnacleBob> ahhhh
[2008/04/29 10:11:47] <hacim> fujin: what version do you have?
[2008/04/29 10:11:54] <BarnacleBob> i want it to copy the file whether it be a link or not
[2008/04/29 10:11:59] <BarnacleBob> that would be manage then right?
[2008/04/29 10:12:01] <shadoi> fujin: yeah I'll just publish to my github in a new branch
[2008/04/29 10:12:09] <hacim> fujin: oh, you were using dash on ubuntu?
[2008/04/29 10:12:12] <shadoi> fujin: making sure I have the latest working code
[2008/04/29 10:12:34] <hacim> fujin: that is an error though, because I declare /bin/sh at the beginning of the script
[2008/04/29 10:12:42] <BarnacleBob> ok thanks everyone i'll be back tomorrow if i need more help
[2008/04/29 10:12:47] <BarnacleBob> its quiting time
[2008/04/29 10:13:07] <fujin> hacim: Ubuntu diverts /bin/sh -> /bin/dash
[2008/04/29 10:13:12] <shadoi> yeah
[2008/04/29 10:13:16] <nward> fujin: you're in NZ?
[2008/04/29 10:13:19] @ nward is now known as daork
[2008/04/29 10:13:23] <fujin> nward: aye, I am.
[2008/04/29 10:13:23] <daork> Hi :-)
[2008/04/29 10:13:26] <fujin> hi!
[2008/04/29 10:13:39] <daork> a dnb fan, even
[2008/04/29 10:14:42] @ Quit: BarnacleBob: "This computer has gone to sleep"
[2008/04/29 10:15:30] @ Quit: yure_: "Odhajam"
[2008/04/29 10:19:19] <nigelk> bloody kiwis
[2008/04/29 10:20:21] @ Quit: f--z: "KVIrc 3.2.5 Anomalies http://www.kvirc.net/"
[2008/04/29 10:21:02] <fujin> ha
[2008/04/29 10:22:56] @ johnf joined channel #puppet
[2008/04/29 10:24:12] @ Quit: randybias:
[2008/04/29 10:25:29] <fujin> shadoi: <3?
[2008/04/29 10:25:58] <shadoi> sorry dude, have to write this resignation letter by EOB. lol
[2008/04/29 10:26:01] <shadoi> hang on
[2008/04/29 10:26:21] <fujin> ah it's cool
[2008/04/29 10:29:36] @ pdt joined channel #puppet
[2008/04/29 10:33:23] @ Quit: Wakko666: "Leaving."
[2008/04/29 10:38:27] <jamesturnbull> fujin: not well
[2008/04/29 10:38:31] <jamesturnbull> fujin: stuck on how to handle the wiki
[2008/04/29 10:38:42] <jamesturnbull> fujin: when I have more time I'll look at it
[2008/04/29 10:39:01] @ Quit: shadoi: "leaving"
[2008/04/29 10:47:33] <jamesturnbull> fujin: I'd love someone else taking a look
[2008/04/29 10:47:45] <jamesturnbull> fujin: at the code to see how easy it'd be to integrate another wiki
[2008/04/29 10:49:42] @ gh left channel #puppet ()
[2008/04/29 10:55:50] @ Quit: plathrop:
[2008/04/29 10:57:37] <fujin> mm, probably a little out of my league
[2008/04/29 10:57:39] <fujin> what's redmine written in?
[2008/04/29 10:57:57] <holoway> fujin: ruby
[2008/04/29 10:58:40] @ twb joined channel #puppet
[2008/04/29 11:04:59] <jamesturnbull> fujin: I keep meaning too but I've another book on the boil, being my boss for eight weeks, sick cat, and absolutely no tuits left
[2008/04/29 11:06:57] <fujin> heh
[2008/04/29 11:06:59] <fujin> what are you writing?
[2008/04/29 11:07:07] <twb> Is there a way to express hard linking, without resorting to exec?
[2008/04/29 11:07:23] <fujin> symlinks can be done with file{}, dunno about hardlinks
[2008/04/29 11:08:22] <twb> Strictly speaking I want to only create a hard link if both source and target are on the same filesystem, otherwise silently create a copy.
[2008/04/29 11:10:54] @ randybias joined channel #puppet
[2008/04/29 11:13:44] <twb> I see I can use file/source to do a plain copy, which will suffice.
[2008/04/29 11:14:18] @ Quit: pdt:
[2008/04/29 11:15:01] @ Quit: nigelk:
[2008/04/29 11:16:06] @ Quit: glut: Remote closed the connection
[2008/04/29 11:21:30] @ Quit: shake-n-bake:
[2008/04/29 11:27:44] @ Quit: danstoner_: Read error: 110 (Connection timed out)
[2008/04/29 11:29:09] @ Quit: jvanzyl:
[2008/04/29 11:31:16] @ danstoner_ joined channel #puppet
[2008/04/29 11:37:16] @ Quit: tazz_: "Leaving"
[2008/04/29 11:37:49] <twb> Rather than clobbering /etc/exports, can I ask puppet to idempotently ensure that the file contains a set of desired entries (i.e. without deleting other entries that may be there).
[2008/04/29 11:41:28] @ pdt joined channel #puppet
[2008/04/29 11:42:08] <twb> Do the trac pages have a "print" version that doesn't include the menu at the top?
[2008/04/29 11:44:27] <jamesturnbull> twb: depends on the page you want
[2008/04/29 11:44:32] <jamesturnbull> twb: you can also use puppetdoc
[2008/04/29 11:44:47] <jamesturnbull> twb: to output most of the inbuilt doco
[2008/04/29 11:45:42] <twb> Ah, cool.
[2008/04/29 11:46:09] <twb> Is puppetdoc just for the type reference page?
[2008/04/29 11:47:52] <twb> Hmm, where is the list of arguments to puppetdoc -r?
[2008/04/29 11:48:56] <twb> Oops, --help says
[2008/04/29 11:49:27] <twb> --list isn't in the SYNOPSIS section of --help.
[2008/04/29 11:54:24] <jamesturnbull> twb: sorry - what's the issue?
[2008/04/29 11:55:47] <twb> Which issue?
[2008/04/29 11:57:47] @ jvanzyl joined channel #puppet
[2008/04/29 12:00:43] @ shadoi joined channel #puppet
[2008/04/29 12:07:39] <fujin> oh hey shadoi
[2008/04/29 12:07:45] <fujin> haven't had a chance to look at your stuff yet - do you just run them out of cron?
[2008/04/29 12:09:53] <shadoi> umm yeah I think so
[2008/04/29 12:10:09] <fujin> cool
[2008/04/29 12:10:11] <shadoi> I haven't really done much with them other than fix them :)
[2008/04/29 12:10:14] <fujin> ah
[2008/04/29 12:10:49] <fujin> oh
[2008/04/29 12:10:52] <fujin> you must not run them out of puppet
[2008/04/29 12:10:59] <fujin> those are actual plugin reporting parsers
[2008/04/29 12:11:03] <fujin> ## lib/puppet/reports/logcache.rb
[2008/04/29 12:12:32] @ Quit: huangmingyou: "Leaving."
[2008/04/29 12:12:47] <fujin> shadoi: don't suppose you'd be able to share the web frontend for it?
[2008/04/29 12:12:48] @ huangmingyou joined channel #puppet
[2008/04/29 12:12:54] <fujin> wanna see if it'll do pretty things if I just drop it in.
[2008/04/29 12:14:02] @ glut joined channel #puppet
[2008/04/29 12:17:02] <fujin> I guess I could just hack lib/puppet/reports/tagmail.rb to do what I want
[2008/04/29 12:21:15] @ Quit: pdt:
[2008/04/29 12:22:13] <jamesturnbull> twb: assumed from your comment that you had issues with puppetdoc
[2008/04/29 12:24:39] <shadoi> fujin: there isn't one :)
[2008/04/29 12:24:48] <fujin> oh
[2008/04/29 12:24:50] <shadoi> fujin: totally trivial to add to puppetshow though
[2008/04/29 12:24:51] <fujin> lol
[2008/04/29 12:25:07] <fujin> mm, I'm sure
[2008/04/29 12:25:31] <shadoi> the lastcheck info is already in there
[2008/04/29 12:25:50] <shadoi> logcache should probably be done a slightly more sane way.
[2008/04/29 12:25:55] <fujin> well, that's just the equivalent of my puppetlast script isn't it?
[2008/04/29 12:26:09] <fujin> I just want to write something which sends pretty emails out when it detects that changes should be made, that's all
[2008/04/29 12:26:09] <shadoi> essentially
[2008/04/29 12:26:21] <fujin> tagmail.rb seems to just blobl them together quite unprettily
[2008/04/29 12:26:49] <shadoi> yeah I'd recommend a rails app, it would be very easy.
[2008/04/29 12:26:51] <fujin> so all I really need to do is pretty up this: p.puts messages
[2008/04/29 12:27:05] <fujin> the messages is blobby
[2008/04/29 12:27:10] <shadoi> you'd just make an rhtml template
[2008/04/29 12:27:20] <shadoi> use a mail model
[2008/04/29 12:27:25] <shadoi> done
[2008/04/29 12:27:55] <fujin> hrm
[2008/04/29 12:28:24] <fujin> need to finish rebuilding this server, puppetizing it and then getting it live before I start playing with that stuff
[2008/04/29 12:45:07] @ felix_da_catz joined channel #puppet
[2008/04/29 12:59:11] <fujin> shadoi: so, are you familiar with the report processing at all?
[2008/04/29 12:59:17] <fujin> I really just want to pretty up the emails that tagmail sends
[2008/04/29 12:59:30] <fujin> might look at more indepth functionality later, but prettying up tagmail is my only requirement at the moment
[2008/04/29 12:59:36] <shadoi> fujin: lemme look
[2008/04/29 12:59:45] <fujin> pastie: come on fool
[2008/04/29 12:59:46] <pastie> fujin: get your own paste url, buddy
[2008/04/29 12:59:51] <fujin> what!
[2008/04/29 12:59:55] <fujin> pastie: url
[2008/04/29 13:00:04] <shadoi> hahaha
[2008/04/29 13:00:04] <fujin> I'm not your buddy guy!
[2008/04/29 13:00:33] <pastie> http://pastie.org/188483 by fujin.
[2008/04/29 13:00:45] <twb> Don't knock Buddy Guy.
[2008/04/29 13:01:00] <jamesturnbull> fujin: don't fuck with pastie
[2008/04/29 13:01:18] <shadoi> awesome
[2008/04/29 13:01:42] @ Quit: chadh_: Read error: 104 (Connection reset by peer)
[2008/04/29 13:01:45] <shadoi> fujin: I'd recommend using erb right there
[2008/04/29 13:01:50] @ chadh joined channel #puppet
[2008/04/29 13:01:53] <shadoi> then you can just make a template
[2008/04/29 13:02:24] <fujin> erb right there?
[2008/04/29 13:02:27] <fujin> not following ya
[2008/04/29 13:02:39] <fujin> as far as I can tell, messages is just a big ass string
[2008/04/29 13:03:01] @ lak joined channel #puppet
[2008/04/29 13:03:22] <shadoi> fujin: http://pastie.org/188485
[2008/04/29 13:03:25] <fujin> thanks
[2008/04/29 13:03:48] <fujin> hmm
[2008/04/29 13:03:51] * fujin cross eyed
[2008/04/29 13:03:53] <fujin> What does that do?
[2008/04/29 13:03:54] <shadoi> you'd just make an erb template that has the email format you want
[2008/04/29 13:04:03] <shadoi> use that snippet to load it and parse it
[2008/04/29 13:04:41] <shadoi> do whatever voodoo you want to do with the "messages" object
[2008/04/29 13:05:01] <fujin> ah
[2008/04/29 13:05:08] <fujin> interesting
[2008/04/29 13:05:19] <fujin> I don't suppose you know anything about said messages object, or more importantly what I could do with it?
[2008/04/29 13:05:26] <fujin> & can I specify the ERB in-code, instead of in a file?
[2008/04/29 13:05:33] <shadoi> nope, I'd probably use ruby-debug to take a gander at it
[2008/04/29 13:05:41] <shadoi> yeah you can
[2008/04/29 13:05:47] <shadoi> ugly, but possible
[2008/04/29 13:05:58] <fujin> yeah, well it just strips down the external requirements
[2008/04/29 13:06:03] <fujin> this probably won't be seen by anyone else bar me :>
[2008/04/29 13:06:08] <shadoi> using an <<EOF sorta thing
[2008/04/29 13:06:27] <shadoi> EOF<< ? I forget
[2008/04/29 13:06:48] <shadoi> But really you don't need erb in that case
[2008/04/29 13:06:54] <twb> When attempting to generate PDFs from puppetdoc(1):
[2008/04/29 13:06:57] <twb> /tmp/puppetdoc.txt:130: (ERROR/3) Unknown interpreted text role "trac".
[2008/04/29 13:07:10] <shadoi> But if you did enough with it I'm sure other people would love to be able to use a template for their tagmail too
[2008/04/29 13:07:26] <fujin> heh, that's a good idea shadoi
[2008/04/29 13:07:52] <fujin> tagmailtemplate=pathtowhatever.erb
[2008/04/29 13:07:57] <shadoi> sure
[2008/04/29 13:08:04] <fujin> That's a little more advanced than I'd like at the moment, but I'll get there soon
[2008/04/29 13:08:10] <fujin> want to get pretty emails with a dummy template coming out for a start
[2008/04/29 13:08:14] <fujin> and then i'll get it crackin more
[2008/04/29 13:08:16] <shadoi> *nod*
[2008/04/29 13:08:29] @ nigelk joined channel #puppet
[2008/04/29 13:09:12] @ Demosthenes joined channel #puppet
[2008/04/29 13:17:47] @ gileswork_ joined channel #puppet
[2008/04/29 13:18:24] @ Quit: gileswork: Read error: 104 (Connection reset by peer)
[2008/04/29 13:18:30] @ gileswork_ is now known as gileswork
[2008/04/29 13:26:00] @ Quit: andrewcshafer:
[2008/04/29 13:27:11] <fujin> lak: what tabspacing do you use for Puppet?
[2008/04/29 13:27:17] <fujin> (vim)
[2008/04/29 13:27:18] <lak> 4 spaces
[2008/04/29 13:28:08] <fujin> lak: I'm going to do some work on reports/tagmail.rb to make it support templating.
[2008/04/29 13:28:21] <fujin> well, a template;
[2008/04/29 13:28:35] <lak> you mean for the email?
[2008/04/29 13:28:39] <fujin> Yup.
[2008/04/29 13:28:57] <fujin> I mean, if that sounds sane and everything
[2008/04/29 13:29:03] <fujin> or would it be better to make a new reports provider?
[2008/04/29 13:29:09] <fujin> prettytagmail, or something.
[2008/04/29 13:31:42] <lak> it's the same info, just using an external template for the email itself?
[2008/04/29 13:31:53] <lak> how will you configure where the template goes?
[2008/04/29 13:32:45] <fujin> another setting, like tagmap
[2008/04/29 13:32:51] <fujin> tagmailtemplate=
[2008/04/29 13:33:11] <fujin> but yes, the exact same info, just an external ERB template for the reports inside the email
[2008/04/29 13:33:18] @ M- joined channel #puppet
[2008/04/29 13:33:21] <fujin> bad?
[2008/04/29 13:33:26] <lak> seems reasonable
[2008/04/29 13:33:40] <fujin> oh, cool
[2008/04/29 13:33:45] <fujin> I thought you were about to shoot me down!
[2008/04/29 13:34:08] <lak> too far away
[2008/04/29 13:34:15] <fujin> hehe
[2008/04/29 13:34:41] <fujin> so to add a new setting, I should simply add it to lib/puppet/defaults.rb, near the other tagmail stuff?
[2008/04/29 13:35:50] <fujin> can't see any other reference to tagmap in the repo
[2008/04/29 13:35:52] <fujin> must be the place!
[2008/04/29 13:35:56] <fujin> Gee, I'm good at this ;<
[2008/04/29 13:38:44] <jamesturnbull> fujin: you crack me up
[2008/04/29 13:40:56] <jamesturnbull> fujin: if you're playing in tagmail there are some tickets too - #1054, #1035, #1089 and #1107 that need love...
[2008/04/29 13:41:00] <gepetto> jamesturnbull: fujin: #1054 is http://reductivelabs.com/trac/puppet/ticket/1054 "tagmail should be sent on failed runs"
[2008/04/29 13:41:04] <gepetto> jamesturnbull: fujin: #1035 is http://reductivelabs.com/trac/puppet/ticket/1035 ""all" in tagmail.conf should allow exclusions"
[2008/04/29 13:41:07] <gepetto> jamesturnbull: fujin: #1089 is http://reductivelabs.com/trac/puppet/ticket/1089 "Log messages should get tagged with their log level"
[2008/04/29 13:41:10] <gepetto> jamesturnbull: fujin: #1107 is http://reductivelabs.com/trac/puppet/ticket/1107 "Allow exclusion in tags specified on the command line"
[2008/04/29 13:42:03] <fujin> oh god
[2008/04/29 13:42:12] <waawaamilk> lol
[2008/04/29 13:42:15] <fujin> LOL
[2008/04/29 13:42:19] <fujin> well, let me get my shit working first
[2008/04/29 13:42:23] <fujin> and then I'll piss around with those
[2008/04/29 13:42:26] <waawaamilk> 1089 looks easy...
[2008/04/29 13:42:28] <shadoi> Go JT.
[2008/04/29 13:42:30] <shadoi> haha
[2008/04/29 13:42:34] * waawaamilk says with no knowledge at all
[2008/04/29 13:42:42] <fujin> I've gotta work out what's actually in that messages object
[2008/04/29 13:42:47] <fujin> waawaamilk: go! fix!
[2008/04/29 13:42:50] <fujin> or tell me how to fix, I'll submit
[2008/04/29 13:42:55] <waawaamilk> heh
[2008/04/29 13:42:58] * waawaamilk can has work to do :(
[2008/04/29 13:43:07] <waawaamilk> setting up a shared git server
[2008/04/29 13:43:15] <fujin> That's easy!
[2008/04/29 13:43:15] <waawaamilk> which is kinda beside the point of git
[2008/04/29 13:43:17] <waawaamilk> but oh well
[2008/04/29 13:43:23] <fujin> git --bare init --shared
[2008/04/29 13:43:29] <waawaamilk> heh
[2008/04/29 13:43:31] <waawaamilk> not quite like that
[2008/04/29 13:43:32] <fujin> ;P
[2008/04/29 13:43:33] <waawaamilk> I mean like a server
[2008/04/29 13:43:38] <waawaamilk> on which many gitrepos can be hosted
[2008/04/29 13:43:49] <fujin> that's how I do it? :>
[2008/04/29 13:43:56] <waawaamilk> I have that all set up, need to control how users get access to them now...
[2008/04/29 13:43:58] <shadoi> waawaamilk: github.com :)
[2008/04/29 13:44:00] <fujin> ssh+git-daemon
[2008/04/29 13:44:05] <fujin> err,
[2008/04/29 13:44:19] <fujin> heh, github well
[2008/04/29 13:44:36] <fujin> I thought of putting my repo's into github
[2008/04/29 13:44:40] <fujin> costs money though doesn't it?
[2008/04/29 13:45:05] <shadoi> fujin: not for a small personal one
[2008/04/29 13:45:11] <shadoi> starts costing when you host a large project
[2008/04/29 13:45:14] <fujin> ah
[2008/04/29 13:45:31] <fujin> so any tricky Ruby guy able to tell me how I can inspect what's in an object?
[2008/04/29 13:45:37] <fujin> p.puts messages
[2008/04/29 13:45:43] <waawaamilk> isn't the whole point of git to be distributed?
[2008/04/29 13:45:51] <waawaamilk> why are the internets putting their gitrepos on one server??
[2008/04/29 13:45:52] <waawaamilk> :)
[2008/04/29 13:45:54] * waawaamilk trolls
[2008/04/29 13:46:09] <fujin> distributed = you can use it when the tubes are down
[2008/04/29 13:46:49] <shadoi> fujin: messages.inspect / pretty_inspect
[2008/04/29 13:47:38] <fujin> thanks
[2008/04/29 13:50:24] <jamesturnbull> nigelk: can i ask broadly what you had in mind to publish about Google's puppet environment?
[2008/04/29 13:50:56] <jamesturnbull> nigelk: my interest is that I am looking to consolidate the scalability, mongrel.nginx/etc sections of the wiki and structure them better
[2008/04/29 13:51:08] <jamesturnbull> nigelk: the release of some material would seem to be a good excuse
[2008/04/29 13:51:11] <nigelk> so I was going to basically discuss how you can use dns views and separate the puppet server/ca roles
[2008/04/29 13:51:15] <nigelk> to scale a lot better
[2008/04/29 13:51:15] <nigelk> and
[2008/04/29 13:51:32] <nigelk> what I think is the more interesting work... how we're using environments
[2008/04/29 13:51:41] <jamesturnbull> nigelk: oh yes - that'd be excellent
[2008/04/29 13:51:45] <nigelk> to basically have a structured release process with decent levels of control
[2008/04/29 13:51:56] <nigelk> so we have things set up now so we can go
[2008/04/29 13:52:01] @ andrewcshafer joined channel #puppet
[2008/04/29 13:52:06] <nigelk> ./integrate.py --promote foo_bar_unstable
[2008/04/29 13:52:13] <nigelk> and it promotes that environment changes to foo_bar_testing
[2008/04/29 13:52:17] <nigelk> and on to foo_bar_stable
[2008/04/29 13:52:59] <nigelk> so we'll be pulling our specific bits out of it, but talking about how you can use version control at the server end to automatically pull all this stuff out of version control
[2008/04/29 13:53:21] <nigelk> as we don't use puppet to manage our puppet servers
[2008/04/29 13:54:01] <nigelk> we really needed environments. Having several teams trying to integrate modules with each other but with different needs and up to a dozen different puppet admins....
[2008/04/29 13:54:13] <jamesturnbull> nigelk: yeah understand
[2008/04/29 13:54:23] <nigelk> it's working really seamlessly actually
[2008/04/29 13:54:35] <jamesturnbull> nigelk: frankly I was surprised Google used it all - given the Python-centric nature of the shop
[2008/04/29 13:54:42] <nigelk> heh :)
[2008/04/29 13:54:56] <nigelk> I have to say I wish it was written in Python
[2008/04/29 13:55:20] <jamesturnbull> nigelk: I regularly hang shit on Anthony Baxter since he started at Google - it's like going back to the Python womb... :)
[2008/04/29 13:55:21] <nigelk> I don't have anything against Ruby though
[2008/04/29 13:55:42] <nigelk> I hadn't done much before I came here. Just hacking on Mailman really
[2008/04/29 13:55:57] <nigelk> i spent so much time doing php web dev that it was my scripting language of choice
[2008/04/29 13:56:08] <nigelk> the monstrosities I wrote in PHP....
[2008/04/29 13:56:09] <jamesturnbull> nigelk: are you able to share tools as well as concepts?
[2008/04/29 13:56:27] <nigelk> I hope so yes, like our integration scripts
[2008/04/29 13:56:47] @ Quit: felix_da_catz: No route to host
[2008/04/29 13:56:51] <nigelk> but even if we don't, I can explain things in broad enough strokes that it should help people get started
[2008/04/29 13:57:01] <nigelk> like using XML to describe your environment layout, then all your tools read from that
[2008/04/29 13:57:18] @ Quit: andrewcshafer:
[2008/04/29 13:57:20] <jamesturnbull> nigelk: okay that's interesting - god I wish Trac's wiki's wasn't so damn crapola - I could properly structure somethign great with Drupal's book structure
[2008/04/29 13:57:34] <nigelk> heh. We're moving http://www.afp548.com from geeklog to Drupal now
[2008/04/29 13:57:37] * fujin blows up XML
[2008/04/29 13:57:48] <nigelk> XML is great for some things
[2008/04/29 13:58:29] <jamesturnbull> nigelk: yeah I agree - just not everything like some shops use it
[2008/04/29 13:59:00] <nigelk> yep. I'm on a bit of a structured documentation kick at the moment
[2008/04/29 13:59:03] <nigelk> docbook/xslt etc
[2008/04/29 14:00:02] * nigelk reads back. <3 github
[2008/04/29 14:00:35] <fujin> shadoi: you can use %q{ template template template }.gsub(/^ /, '')
[2008/04/29 14:00:41] <jamesturnbull> nigelk: yeah I just moved a blog site to Drupal so I could use their "book" concept - which includes automatic output to docbook
[2008/04/29 14:01:25] <nigelk> I did a big OpenCMS deployment at my last job, and if Drupal had been just a little bit better.... we would have gone with it
[2008/04/29 14:01:37] <nigelk> it just didn't do rollback,audit trail stuff properly at the time
[2008/04/29 14:02:12] <shadoi> fujin: ah, cool. still kinda ugly. :)
[2008/04/29 14:02:20] <fujin> yeah, guess so.
[2008/04/29 14:02:35] * fujin tries to trigger some action for tagmail
[2008/04/29 14:02:51] <jamesturnbull> nigelk: yeah Drupal 6 is a big step forward - even made me forgive my passionate hate of PHP
[2008/04/29 14:03:07] <shadoi> jamesturnbull: I ran a drupal community for a while, it's kinda crappy (but that was 4.7), hopelly the newer releases are better.
[2008/04/29 14:03:27] <nigelk> I'm really surprised the other web languages didn't learn from PHP
[2008/04/29 14:03:27] <jamesturnbull> shadoi: they are
[2008/04/29 14:03:29] <shadoi> The book thing was brand new in 4.x
[2008/04/29 14:03:34] <nigelk> it took off because of how easy it is to deploy
[2008/04/29 14:03:48] <nigelk> and it's still such a pain in the arse to deploy Ruby/Rails stuff with decent performance
[2008/04/29 14:03:49] <jamesturnbull> nigelk: yeah but security... dear dog
[2008/04/29 14:03:57] <nigelk> well yeah :) global everything!
[2008/04/29 14:04:01] <shadoi> nigelk: yeah, if there was a mod_rails or mod_rack it would be kicking ass.
[2008/04/29 14:04:07] <fujin> heh, want.
[2008/04/29 14:04:11] <nigelk> for sure.
[2008/04/29 14:04:19] <nigelk> so my biggest gripe with PHP is the API
[2008/04/29 14:04:20] <fujin> fcgi kinda does it, reasonalby well
[2008/04/29 14:04:30] <shadoi> nigelk: what API? :)
[2008/04/29 14:04:35] <fujin> lol.
[2008/04/29 14:04:36] <fujin> well played
[2008/04/29 14:04:45] <nigelk> http://docs.php.net/manual/en/ref.strings.php
[2008/04/29 14:04:48] <nigelk> that clusterfuck
[2008/04/29 14:05:06] <shadoi> str2pdf_cnvrt_txtretro
[2008/04/29 14:05:35] <nigelk> strtr str_replace trim parse_str
[2008/04/29 14:05:37] <shadoi> indeed
[2008/04/29 14:05:45] <nigelk> where's the consistency!
[2008/04/29 14:06:17] <shadoi> ruby has some goofy stuff too, it's fairly hidden most of the time thankfully.
[2008/04/29 14:06:52] <fujin> ugh, install.rb fails. gzip: /usr/share/man/man8/puppetmasterd.8.gz already exists; do you wish to overwrite (y or n)? y
[2008/04/29 14:07:01] <nigelk> fujin: that's been annoying me
[2008/04/29 14:07:08] <nigelk> I'm about to submit a change to suggest using setup.rb
[2008/04/29 14:07:12] <fujin> heh
[2008/04/29 14:07:17] <fujin> %x{#{gzip} #{omf}}
[2008/04/29 14:07:24] <nigelk> as fink/macports packages are almost impossible to do with that installer script
[2008/04/29 14:07:24] <fujin> %x{#{gzip} -y #{omf}}