=== harlowja is now known as harlowja_away === ctracey|away is now known as ctracey [13:07] would maintainers be interested to have varied linux/bsd distro compatible logic to configure system settings like network/pasword in an adapter fashion [13:31] abk, we can configure network to some extent, and you can add users and set their password that works in linux. i'm not sure if it would work in bsd. [13:35] smoser, I'll be working on a binary reaching via cloud-init into instances taking care of all these base configuration on VMs which get pushed by Hypervisor during build [13:36] smoser, I thought if cloud-init can use it up... will instead write as a component of it instead [13:37] abk, we can make cloud-init consume such data for sure. [13:37] smoser, distros supporting (RHEL,Fedora,CentOS,Debian,Ubuntu,Gentoo,Arch,OpenSuse,FreeBSD,Scientific) and all major version of these [13:37] i'm not sure what you meant by "binary reaching via cloud-init". [13:39] smoser, I'm not very well-versed in cloud-init yet... but what I notice is I can get my service doing these tasks via any bootstrap script using cloud-init's userdata [13:39] smoser, but if it can be a part of cloud-init (as in if you folks and community is interested)... I'll instead put it somewhere where it's much easily usable by anyone desiring such feature [13:40] ok. yeah, right. i would be interested in that, yeah. [13:41] passing network settings in is currently kind of tricky/broken [13:41] it should work except for the case of 'eth0' (or any other "auto" configured devices with information already in the instance) [13:42] ie, ubuntu cloud images ship with 'auto eth0' (for dhcp). [13:42] and changing that is not really functional [13:42] because we don't bring *down* interfaces when we configure networking passed in. [13:47] :w [16:06] Hey guys, noticed the docs have been bumped to 0.7.5, but the latest tagged release is still 0.7.4 [16:06] is a new release imminent? [16:09] smoser: That's a question directed at you I guess :) [16:10] not imminent, but development is open. [16:11] the rtd is keyed off of Changelog i guess ? === harlowja_away is now known as harlowja [18:35] smoser i think the rtd is keyed off the version.py in cloudinit [18:35] not changelog afaik [18:46] http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/doc/rtd/conf.py#L41 [18:54] smoser yt, ok with merging https://code.launchpad.net/~harm-o/cloud-init/freebsd ? [18:54] sean here is gonna start pushing the people in the release team for freebsd to start making openstack images i think [18:54] using that freebsd stuff (and any future changes) [18:57] i hadn't really looked at that. [18:57] i recently changed cc_growpart [18:57] so i would have thoguht the merge thing would not like line 70 in that diff [18:57] https://code.launchpad.net/~harm-o/cloud-init/freebsd/+merge/198130 [18:58] and i think that VERSION (line 701) is going to piss off someone who was here before with some other distro. [18:58] but ... ok. merge i guess. [18:59] oh wait. [18:59] one thing [18:59] harmw yt :-P [18:59] ^^ [18:59] never mind. harmo signed agreement. cool. [19:01] smoser, harlowja: has anyone confirmed if the disk paritioning code works with this mp? [19:01] i havent [19:01] but my guess is its not 100% complete [19:01] thats why seanwbruno just arrived [19:01] hahaha [19:01] * seanwbruno stops trolling from afar [19:01] harlowja: that is my guess [19:01] seanwbruno to fix it all! [19:01] pfft [19:02] haha [19:03] seanwbruno i think is gonna try to bug the freebsd people to make better partitioned images, then hook this in [19:03] but he can explain i guess [19:03] dealing with $work horse shit at the moment, but tomorrow I'll get with the fbsd re team to put something out that looks like our franken-images at the office. [19:03] lol [19:03] I may way to scrape the fbsd cloudinit stuff into a port for others if that's not in the works already. [19:04] i don't know anyone else doing it [19:04] hrm, I wonder if a fbsd ports has a hook into launchpad yet. [19:04] let me parse some make files [19:04] parse those makes files, lol [19:05] *make files [19:05] harmw yt [19:06] well, my cursory "grep -ri launcpad" doesn't seem to indicate any kind of helper stuff. [19:07] is there a shortcut URL that would engrabinate a tgz/tar/whatever of a LP repo? [19:08] smoser ? [19:08] ^ [19:08] not vital, just being lazy [19:10] probably is some way [19:10] i dont think there is. [19:10] well, I can just checkout the tree and tgz it myself. [19:10] like I said, just being lazy [19:11] you're saying wget http://some/url/lp:cloud-init > trunk.tar.gz [19:11] right ? [19:11] smoser: aye [19:15] seanwbruno one think i beleive will need adjustment (especially for y! use-case) will be the config-drive usage, which i believe that branch doesn't fixup [19:16] *especially around network settings being written [19:17] basically the config-drive in openstack contains a network settings file, which cloud-init then translates into whatever version it understands for the current operating system [19:17] looking at that review [19:17] 407 + def _write_network(self, settings): [19:17] 408 + return [19:17] :) [19:17] * harlowja not really sure how freebsd handles network configuration files and stuff [19:18] if you want static configs, we'll have to construct an entry for rc.conf or rc.conf.local (preferred) and splat it in there [19:18] I suspect, that most configs will want to be splatted in there. [19:20] k, thats probably not that bad [19:20] i didn't implement setting up networking in rc.conf just yet iirc [19:21] ah, k, harmw were u planning on, maybe that can be seanwbruno 'get wet' commit, lol [19:21] lol [19:21] *ok that probably wasn't a very good selection of words [19:21] whatever [19:21] ;) [19:21] I smell a "print" command coming on. [19:22] I wasn't planning on coding that tonight though [19:22] lol [19:22] tommorow night then? [19:22] :-p [19:23] friday's more like it :p openstack shows networking stuff per ec2 api, right? [19:23] hmmm [19:23] besides config-drive (which I dont use) [19:23] not exactly [19:23] stuff like bla/local-ipv4 [19:23] ah, i think for the ec2 usage, people mainly rely on dhcp [19:24] for config-drive its pretty different [19:24] indeed it is, pretty much straightforward in that case [19:24] perhaps i can parse the dhcp lease file and configure that to be static [19:25] to cope with scenarios where networking is not done through a metadata api [19:25] http://paste.ubuntu.com/6798907/ is whats in the config-drive [19:25] basically a ubuntu network file :-P [19:26] yea, I noticed the rRH scripts rebuilding that blob to something RH understands :p [19:26] :) [19:26] ya, josh's custom parser [19:26] haha [19:27] I have some hackery somewhere to parse that things somewhere. [19:27] ya, the _write_network(self, settings): should get the parsed object [19:28] not the raw string (i think) [19:28] ooo [19:28] I even did it in /bin/sh [19:28] wow ... such code [19:28] ah, maybe not, http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/distros/rhel_util.py [19:28] thats josh parser [19:28] oh [19:28] so [19:29] that's another thing I wanted to ask [19:29] why on earth isn't all the network data in the meta.js file? [19:29] I was confused why I had to parse meta.js for a couple of config elements but all the network data had to come from some janky redhat file. [19:30] lol [19:30] legacy [19:30] gross [19:30] thats really all [19:30] *janky ubuntu file ;) [19:30] lol [19:31] I was so happy (for a moment) when I could just do json parsing from single command line invocations of python and then got super confused by the then parsing of compeltely seperate network config file [19:31] :) [19:31] openstack started off being much more ubuntu specific, so that piece has stuck [19:32] something like this was what I was going to put in a rc script somewhere. [19:32] http://bpaste.net/show/171009/ [19:32] soemday https://fedorahosted.org/netcf/ might be used [19:33] I'd have just voted to cram it into the meta.js object and be done with it. [19:33] but [19:33] whatever [19:34] also, hooray for outputting text with python only to then turn around and pass it to grep/awk [19:34] * seanwbruno is a silly, silly man. [19:34] they picked that /etc/network/interfaces file [19:34] that was a bad decision [19:35] but unfortunately, bad decisions mean legacy dependencies. [19:35] and they cant change *that* file. [19:35] they could provide it in a more sane format elsewhere though. [19:35] (and probably should) [19:35] ya [19:35] http://paste.ubuntu.com/6798955/ [19:35] so thats why u see weird RHEL converter [19:35] that make sense ^ [19:36] i'm gonna commit it. python parsing of requirements.txt and ChangeLog [19:36] i tihnk we can assume core python [19:36] seems ok to me [19:37] seanwbruno steal the RH converter, it already translates ubuntu format to agnostic format [19:37] and then just write agnostic format out to freebsd format [19:37] it probab ly doesn't need tobe in a file called rhel_util [19:38] read-version is 4 times slower after my python rewrite :) [19:38] but oh well [19:40] goooo python! [19:43] python3 is only twice as slow as python2 in this case. [19:44] :( [19:45] smoser another thing i was thinking about [19:45] to make cloud-init work in both 2 and 3 [19:45] already done quite a bit of that in taskflow using the six module [19:45] probably not so bad to do it in cloud-init [19:45] then it will be 2 and 3 compat [19:45] i'd like that. [19:46] we dropped boto that is good [19:46] the python-oauth is another woone that has to be replaced [19:52] kk [19:52] should be simple to do six stuff [19:52] now that i know how to use it pretty well [19:52] then u won't be blocking ubuntu move to py3 [19:54] seanwbruno u didn't go through the CLA stuff yet right? [19:54] *ubuntu CLA [19:55] *not openstack CLA [20:03] harlowja: nah [20:03] k [20:07] seanwbruno smoser i think https://code.launchpad.net/~harlowja/cloud-init/net-distro-util/+merge/202743 should make it more obvious that u can use the same function [20:07] harmw [20:08] totally [20:09] i wont block ubuntu move to py3. but dont think we'll get that in in trusty. [20:09] that'd have to happen "really soon now". [20:09] i was kind of planning on it as 0.8.0 [20:09] harlowja: please extend the comments a little to describe what this agnostic format looks like [20:10] since now one needs to go over the source, and I'm way to lazy to do that :p [20:10] otherwise, looks like a nice change [20:10] and usable, ofc [20:16] harmw sure [20:16] will do [20:17] if i can remember, haha [20:17] :) [20:17] smoser i don't think six stuff will take that long really [20:17] its pretty straightforward [20:17] i think its a pita. [20:17] i liked my ignorant magic handling of unicode [20:18] or string. [20:18] and open() and write(string_or_unicode) just working [20:40] lol [20:40] ya [20:40] its dumb, i agree [20:40] https://code.launchpad.net/~harlowja/cloud-init/net-distro-util/+merge/202743 harmw [20:41] comments added [20:45] nice [20:46] (though I'm missing a newline on L85 :p) [20:47] hmmm [20:47] k [20:47] anyway, I think it's not that hard to have fbsd use this function to get the networking done - provided the image was booting with a config drive [20:47] agreed, it shouldn't be [20:47] and if it wasn't booted using config-drive, it should figure out what to configure per what it got through dhcp [20:48] yup [20:48] that'll have to wait 'till the weekend though :) [20:48] seanwbruno can do it to :-P [20:48] get his feet wet [20:48] either way [20:48] ah, well in that case I'll o it right now :p [20:49] lol [20:49] tomorrow, probably [20:51] all good