Time Nick Message 00:26 pdurbin ah, i see what comtona meant about the man git repos. after i run that rake task two dozen modules get cloned. most are from the puppetlabs github account (i.e. https://github.com/puppetlabs/puppetlabs-nova ) but some from other accounts such as https://github.com/bodepd/openstack-quantum-puppet 00:29 pdurbin modules cloned by `rake modules:clone` at https://github.com/puppetlabs/puppetlabs-openstack Dancebin! - http://danceb.in/tvI5foIk4hGcLKXUl9DNYg 00:45 westmaas pdurbin: when you are up and running: http://www.sandywalsh.com/2012/10/debugging-openstack-with-stacktach-and.html 03:50 pdurbin so, i'm stuck :( with the openstack-vagrant-centos-puppet thing. i just updated the readme if any puppet masters want to take a look: https://github.com/pdurbin/openstack-vagrant-centos-puppet 03:51 pdurbin (the git submodule thing seems to work fine by the way. i just keep linking them up: https://github.com/pdurbin/openstack-vagrant-centos-puppet/tree/master/modules ) 12:16 pdurbin hmm! 12:17 pdurbin sjoeboo: you're so smaht. i betcha i can not break our production puppet by simply developing in my own branch in git... since you put this in: modulepath=/etc/puppet/environments/$environment/modules 12:17 pdurbin i've been meaning to play around with this anyway. i usually just commit to master 12:22 pdurbin of course, if i'm going to switch over from vagrant to our physical servers, i need to re-kickstart them 12:49 pdurbin hmm, someone has already been working on these 4 servers... they're up already - Ganglia:: openstack Cluster Report - http://software.rc.fas.harvard.edu/ganglia2/ganglia2_misc/?r=hour&cs=&ce=&s=by+name&c=openstack&tab=m&vn= 14:38 pdurbin the other week i was saying that i didn't like the docs i had written on re-sizing disk images. i re-wrote them 14:38 pdurbin resizing a KVM disk image on VM using LVM - http://danceb.in/rBLiick4hGcLKXUl9DNYg 14:38 pdurbin resizing a KVM disk image on VM that uses raw partitions rather than LVM - http://danceb.in/Mq6PPgk4hGaDuW7l9DNYg 14:39 JoeJulian I thought that wasn't how documentation was done. From what I've been reading, documentation is done once. About half the time something in the code changes, the docs are updated. If they read poorly, everyone's supposed to complain about them but under no circumstances is anyone supposed to actually do anything to fix them. 14:40 pdurbin my questions... for the LVM example, is it normal/expected that I have to eyeball the output of fdisk and pick the next up cylinder (20807)? it's annoying: http://danceb.in/rBLiick4hGcLKXUl9DNYg#L-59 14:42 pdurbin also... can i get away from using gparted, which is a GUI tool? (this is for the non-LVM example) it "just works" so i keep using it but I'd prefer something from the command line. see my use of gparted here: http://danceb.in/Mq6PPgk4hGaDuW7l9DNYg#L-114 14:43 pdurbin JoeJulian: you're right. that's the usual case :) 14:44 JoeJulian Really? I've done this live. 14:45 pdurbin JoeJulian: done what live? 14:49 larsks|away pdurbin: Haven't we already covered alternates to gparted :) 14:49 larsks|away *alternatives 14:49 pdurbin larsks|away: kinda. but it didn't work 14:50 larsks I promise it does. Those docs I wrote up are exactly what I use all the time. 14:50 larsks Note the caveat about only working if you're resizing the last partition on the disk... 14:51 pdurbin larsks: right, we both posted our examples. there was some difference i wanted to ask you about... maybe primary vs. extended partitions... anyway, the links are in this channel's log somewhere 14:52 larsks Yup. 14:53 pdurbin larsks: your way seemed just as fiddly as my way anyway 14:54 larsks My way doesn't require any GUI tools :) 14:54 larsks Besides, *any* solution is going to involve approximately the same steps. 14:54 pdurbin larsks++ 14:55 * pdurbin saw people plus plussing each other in #vagrant the other day 14:55 JoeJulian pdurbin: resized raw lv images with the vm running. 14:57 pdurbin JoeJulian: sure. my lvm procedure involves shutting the vm down to get a nice clean back up of the qcow2 file before doing anything. and a reboot because `partprobe` gives a scary warning: http://danceb.in/rBLiick4hGcLKXUl9DNYg#L-95 15:00 JoeJulian I wish I remembered how I got around that... Something I did before I started blogging... 15:00 pdurbin #crimsonfu *is* my blog :) 17:46 JoeJulian pdurbin: Ok, how do you do the git links? I need to do that with ufopilot. 17:55 pdurbin git submodule add [url] path/to/where/you/want/the/checkout 17:55 * pdurbin loves git submodules 17:56 pdurbin crimsonfubot: lucky ufopilot 17:56 crimsonfubot pdurbin: http://www.ufopilot.com/ 17:57 pdurbin ufo is some openstack or gluster thing, right? 17:59 semiosis ufo is gluster's swift implementation/fork 18:04 pdurbin JoeJulian: the trick with git submodules is that when you got to clone the repo somewhere else, you have to run `git submodule init && git submodule update` example at https://github.com/pdurbin/greptilian-vagrant 18:07 pdurbin actually, some output even here - https://github.com/pdurbin/openstack-vagrant-centos-puppet . still looking for ideas on vagrant openstack puppet stuff! :) 18:08 pdurbin as i put in the readme (and mentioned earlier) i think i need a puppet master within vagrant 18:08 pdurbin but that sounds like work 18:09 pdurbin so i'm switching gears and going to try this on physical hardware instead 18:09 pdurbin ( `koan --replace-self` FTW, by the way) 18:10 pdurbin of course, the openstack puppet code has two dozen modules. and some conflict with modules we've written (mysql, ssh) 18:11 pdurbin so i need to use comptona's trick, i guess, and set up an "modules_external" directory and add it to my modules path 18:11 comptona it works well for us 18:12 pdurbin sjoeboo: you'll kill me if i break puppet 18:12 comptona we actually have a fairly crazy system where our modules are implemented as branches of the same git repo where the host config is stored, using git subtree merges to tie it all together 18:13 comptona that gives us perfect knowledge about every piece of puppet code on a box just from the git hash in the log from when it was applied 18:13 pdurbin so i'll probably just work in a git branch, a puppet "environment", and blow away the conflicting modules (mysql and ssh). just for testing 18:13 pdurbin comptona: that sounds pretty cool 18:14 comptona it is, if I do say so myself :) 18:15 pdurbin if a puppet modules is using storeconfigs... is there a way to somehow take them out? convert it to some other way so you can run masterless? 18:16 comptona you could use puppetdb to stash storeconfigs and keep not using a master: http://puppetlabs.com/blog/introducing-puppetdb-put-your-data-to-work/ 18:18 pdurbin huh. we use puppetdb but it's tied in with our masterful (heh) puppet setup. you're saying you can use puppetdb without having a puppet master? 18:19 comptona that's my understanding, although I've never tried 18:20 comptona one of my someday projects is to set up puppetdb and start exporting configs, and then write a nagios config generator that dumps puppetdb 18:20 pdurbin comptona: fork my vagrant repo and try it! :) 18:21 comptona so a host can say "include apache-service", and the apache-service class exports a "monitor my port 80 with nagios plzkthx!"; then the generator script grabs those notes and writes nagios configs to their effect 18:21 comptona haha, like I said, someday project :) 18:21 comptona we're doing a huge buildout right now so I'm swamped with that 18:23 pdurbin comptona: right. we're doing that. the nagios storeconfigs thing. with puppetdb. it's awesome. autopopulates our nagios for us 18:23 pdurbin sjoeboo: you da man 18:53 semiosis dammit gitlab. 18:54 semiosis your install procedure sucks 18:54 semiosis </venting> 19:00 pdurbin no puppet module for it? 19:02 semiosis curl | sh -- can you imagine how insane it would be to write a puppet module to install that? 19:03 semiosis and you know how i feel about using config mgmt instead of packaging (i'm against it) 19:03 pdurbin sounds like a problem upstream :) . need a better install procedure 19:03 semiosis https://twitter.com/pragmaticism/status/245634704097824768 19:03 semiosis they like it that way!!! 19:04 semiosis ooh, lunchtime 19:04 semiosis bbl 19:17 pdurbin comptona: the git branch/puppet env thing is working fine, by the way. very handy 19:20 larsks westmaas (et al): got our Windows images working. Boots up, generates random Administrator password, writes it to serial port (which shows up in OpenStack GUI in the "Log" tab) and also pulls ssh keys from metadata server. 19:20 larsks So now everybody is happy. 19:20 larsks Except for those people who are going to complain that I used server core. 19:20 pdurbin server core? 19:34 pdurbin huh. Error: Could not install module 'puppetlabs-openstack' (latest: v0.2.0) No version of 'cprice404-inifile' will satisfy dependencies. 'puppetlabs-keystone' (v1.0.1) requires 'cprice404-inifile' (>= 0.9.0) . Use `puppet module install --ignore-dependencies` to install only this module 19:34 pdurbin comptona: did you have to use --ignore-dependencies 19:34 pdurbin (i'm running `puppet module install puppetlabs-openstack --target-dir modules`) 19:43 westmaas larsks: sweet! 19:43 westmaas nicely done 19:45 pdurbin oh i see.. only 0.3.0 is available at http://forge.puppetlabs.com/cprice404/inifile . huh. i guess that means i have to go track down 0.9.0 19:47 pdurbin JoeJulian: i see you're filing bugs against this too :) https://github.com/puppetlabs/puppetlabs-openstack/issues/102 19:50 pdurbin the weird thing is... withouth --target-dir a bunch of openstack stuff seemed to be installed (properly?) into ~/.puppet/modules ... including inifile 0.3.0 and keystone... 19:51 JoeJulian I do need to follow that up. The broken common/cfg.py in glance is fixed in the openstack git tree. 19:52 pdurbin JoeJulian: did you install this puppetlabs-openstack thing from puppetforge or git? 19:53 JoeJulian I filed that while Dan was at puppetconf. He's back now so I should see if I can put together a proper pull request. 19:53 JoeJulian forge 19:54 pdurbin buh? "Latest Release: 0.9.0 on November 2, 2012" dude must have just uploaded this? http://forge.puppetlabs.com/cprice404/inifile 19:54 JoeJulian Or at least I think I did... It's been a while. 19:54 JoeJulian I was actually reinstalling it now from scratch. 19:55 pdurbin usually sjoeboo does all the `puppet module install`. probably good that he's away so i have to do it :) 20:03 comptona pdurbin: I did not have to do so, but the module has changed since I installed it 20:03 comptona that specific module doesn't seem familiar 20:05 pdurbin comptona: so you're not running 0.2.0 of http://forge.puppetlabs.com/puppetlabs/openstack ? i guess you're running 0.1.0, unless some were deleted from the forge 20:06 comptona pdurbin: you're correct, I'm running 0.1.0 still 20:07 comptona I'm planning to upgrade my stack to folsom soon, so pulling in new modules will be part of that 20:07 pdurbin i'm surprised neither --debug nor --verbose show me download progress. i finally straced this puppet to see some life 20:07 pdurbin yeah, i'm trying folsom 20:24 pdurbin huh, "OpenStack High Availability Modules" at https://github.com/danehans/puppet-openstack via https://github.com/lorin/openstack-ansible/issues/2#issuecomment-10025861 20:25 comptona interesting, I'll have to check that out 20:25 larsks Ooo, Folsom is available for EPEL: https://lists.fedoraproject.org/pipermail/cloud/2012-November/001848.html 20:26 pdurbin larsks: ah. epel-testing. cool 20:27 pdurbin i'll just use that instead 20:27 pdurbin JoeJulian: ^^ 20:28 pdurbin could http://forge.puppetlabs.com simply be overloaded? i'm on a different client now and `puppet module install` still doesn't work 20:29 pdurbin maybe they're blocking my ip 20:30 pdurbin anyway, gotta run... pick up the kids. have a great weekend, everyone