=== ctracey is now known as ctracey|away === shardy_afk is now known as shardy [15:56] smoser: (or whoever maintains DataSourceAltCloud.py) [15:57] would this be a nice litle fix in terms of consistency? [15:57] http://paste.ubuntu.com/6530317/ [15:57] I noticed this beeing done in one of the other modules [15:57] (generally, i dont care for the 'which' business at all [15:58] ie, execve does exactly that for me. [15:58] which woudl check that the program is available, but theere is no reason to hard code paths [15:58] that make sense ? [15:58] totally [15:58] hard coding paths just breaks stuff. [15:58] agreed [15:58] the solution just needs to be consequently applied, thats all :) [16:00] yeah. [19:39] hi spandhe ! [19:39] https://code.launchpad.net/~shraddha-pandhe/cloud-init/debug-module/+merge/198119/comments/459151 (fyi) [19:39] spandhe ^ [19:39] thx smoser :) [19:41] thanks smoser ! Going through the comments. [19:42] harlowja, you think those make sense ? [19:42] just more and more i'm trying to namepsace module config [19:42] sure [19:42] i think thats fine with me [19:42] +2 [19:42] :) [19:46] spandhe if it doesn't make sense, feel free to ask :) [20:03] in fact, if you think i make sense more than 1/2 the time, something is probably wrong. [20:07] hahaha [20:24] smoser: It does make sense :) [20:28] smoser http://lists.openstack.org/pipermail/openstack-dev/2013-December/021482.html :) [20:29] i really think unified guest is just plain silly. [20:29] err... unified guest agent. [20:29] possibly, haha [20:30] there are hunderds of perfectly functional "agents" that run in a guest [20:30] * puppet [20:30] * chef [20:30] * sshd [20:30] * jujud [20:30] sure sure [20:30] guess it depends on the use-case, for that kind of stuff i agree [20:31] and one to rule them all, seems highly unlikely [20:32] *depends on what the agent is doing i guess* [20:32] *or what the agent wants to do [20:32] what does "VMs are not accessible from controller." mean? [20:33] my guess is not accesible easily from control network (which i guess is where savanna runs) [20:33] so isolated control network from vm network, means savanna can't actually do that ssh [20:34] although not enabling ssh access seems odd [20:35] or it maybe means that savanaa has no business being on the control network [20:35] :) [20:35] right [20:35] so that might be what it means, not 100% sure either, my interpretation, ha [20:35] no. [20:35] i think your interpretation is right [20:36] but what i'ms aying is "why was savanah not operating on the tenant network" [20:36] which clearly has access to the tenant machines [20:36] sure, its a good question [20:36] probably complicates savanna deployments i would think [20:37] or would require savanna to open some connection to tenant network before it would do some work [20:37] *which i guess is setup hadoop [20:38] said temporary access might not be possible (almost requires something like oauth, which allows limited application access to work on behalf of a user for a given time) [20:39] so i guess then there idea of we'll just bypass all that with a agent :-P [20:39] ewll, in this case i'm mainly just arguing that savanah has no business as part of openstack [20:39] rather than "juju deploy hadoop" [20:39] (which needs no such special access) [20:39] or "ansible make me hadoop" [20:39] u won't get any disagreement from me there :-P [20:40] or create instnaces with tiny hadoop setup userdata script [20:40] :-/ [20:40] trove somewhat seems more valid. [20:40] but really, i dont see any reason why you need a transport mechanism that is not ipv4 [20:40] it turns out lots of things use ipv4 [20:40] and its pretty reliable [20:40] i think i'm even chatting with you over it! [20:41] for real? [20:41] haha [20:41] ipv4 -> nsa -> irc [20:41] exactly. [20:42] ya, hehe, the amount of projects in openstack that something like juju already does is a little weird, i agree [20:42] juju, ansible, .... [500+ others] [20:46] anyways, its openstack, never know what it will do at anytime, haha [20:47] right [20:47] i dont think juju is special [20:47] i just use it because thats what i'm told to use [20:47] :) [20:47] of course [20:47] i don't think rolesdb, pogo (yahoo things) are special either [20:47] *which serve similar purposes as juju [20:48] machine gets a role [role being a list of packages that it will have, for ex: mysql_server, blah blah], pogo can be used to ensure role gets onto instance... [20:48] yada yada, ha [21:32] smoser: https://code.launchpad.net/~harm-o/cloud-init/freebsd [21:32] i've hacked up some code for fbsd [21:33] perhaps you can take a peak at it some time :) [21:39] harmw, i forget, have you signed the CLA ? [21:40] harlowja, ^ https://code.launchpad.net/~harm-o/cloud-init/freebsd/+merge/198130 [21:42] I'm quite new to bazaar/launchpad when it comes to adding code and such, so i'm a bit clueless on what that means :) [21:43] harmw, http://www.canonical.com/contributors [21:43] you have to read the appropriate "individual contributors" or "entities" likn, agree, and then click on "signed online" link and sign it [21:43] ah, the 'all your base are belong to use' stuff :) [21:44] you'll be asked to sign in with your launchpad id. [21:44] harmw, exactly. [21:44] :-( [21:44] harmw, did sed -n really not work ? [21:44] nope [21:44] is it just '-n' ? [21:45] because its not 1987 any more [21:45] let me check on that again [21:46] sed '/^[0-9]\+[.][0-9]\+[.][0-9]\+:/ {s/://; p; :a;n; ba; }' ChangeLog [21:46] sed: 1: "/^[0-9]\+[.][0-9]\+[.][ ...": unexpected EOF (pending }'s) [21:47] hm.. [21:47] bugs me when a utility from a "full blown unix" is less capable than the one from busybox [21:48] hehe [21:49] uptime is clever. [21:49] yeah well, thats on the list of improvements [21:50] since it's totally whack to include the library stuff like i did there [21:50] should go in a more decent function of some kind [21:50] yeah [21:51] harmw, there is a bug for "use ip rather than route" [21:51] I want to skip some tests while running nosetests. How do I do that? [21:51] is ip more portable ? [21:52] spandhe, to run only on one file just run like: [21:52] nosetests tests/unittests/test_datasource/test_altcloud.py [21:53] good one... its from iproute, afaik thats specificly linux [21:53] smoser: thanks! [21:54] hm, -1 for adding a blank line [21:55] harmw, hm.. [21:55] that looks really nice though over all [21:56] thanks :) [21:56] can you online resize disks on freebsd? [21:56] ie, like i an on 3.8+ kernels on linux [21:57] s/disks/partitions/ [21:57] it requires a reboot, eg. like centos 6 [21:57] woah, nicefreebsd [21:58] well, on centos there is cloud-initramfs-growpart [21:58] (ie, that works under drakut on centos in initramfs when the partition isnt mounted) [21:58] indeed [21:58] is there an analog to that in freebsd ? [21:58] something we could make for the initramfs ? [21:58] not that i know of [21:59] is thre an initramfs ? [21:59] smoser now this will bring up the battle of does y! want to keep on enabling the usage of freebsd :-P [21:59] if we allow vms with it, then freebsd never die, ha [21:59] *die in yahoo [21:59] * smoser used *bsd probably last in like 2001. that was before harlowja could ride a bike. [21:59] likely [22:00] no initramfs on fbsd smoser [22:00] reboots suck [22:00] true, though freebsd boots quite fast nonetheless :) [22:01] harmw, we could make it reboot magically if necessary [22:01] also possible [22:01] ie, like cloudinit/config/cc_package_update_upgrade_install.py does [22:01] if it grew the root, then just reboot, and things continue on where you last left off. [22:02] ah, instant reboot on the spot? [22:02] yeah. [22:02] would be good enough [22:03] that module there does that if a package upload marks 'reboot-required' [22:03] s/upload/upgrade/ [22:03] yea, im going over the source now [22:03] *reading [22:06] i'm heading out for th eweekend. [22:06] thanks again, harmw and spandhe . [22:06] surething [22:06] harmw, please read/sign the cla [22:06] already did that [22:06] ah. great. [22:06] have a nice weekend all [22:06] same to you [22:09] smoser: enjoy your weekend!