[00:00] harmw: ping [13:26] hi guys, i still did not made nocloud to work [13:26] do i need to specify somewhere to clod.cfg path to iso file [13:29] this is what i have in logs [13:29] http://pastebin.com/CGmDzHBJ [13:33] RaginBajin: pong [13:33] harmw: Hey, So I was running your build last night. [13:34] ah great, enjoy me with al the bugs you encountered :p [13:35] Well, I need to go back and create another image. We don't always use the Openstack Metadata service but use the configdrive. So I need to create an additional instance to test with Openstack metadata [13:35] did it work with the script from tools/? [13:35] ah ok, yea I havent looked at configdrive yet [13:36] So, i downloaded the code, ran the python setup.py install but forgot to add --init-system. We should change that error message to mention to use --init-system. [13:36] Everytime I need to go into the setup.py script and look at the stupid file to find out what the name of the argument is [13:36] no no, what you should do is read/use the script in tools/ :) [13:37] The build-on-freebsd? [13:37] http://bazaar.launchpad.net/~harmw/cloud-init/freebsd/view/head:/tools/build-on-freebsd [13:37] yep [13:37] ahh [13:37] I needed some way of easy testing, and keeping any neccessary hacks documented until there is a real freebsd pkg/port [13:38] a clean/fresh OZ image only needs bzr installed, from there on you can use bzr to grab my branch and this script will do the rest [13:39] ahh ok. [13:39] Well let me try it that way then. [13:39] you can see the required steps as well, regarding the --init-system [13:40] I'm working on building a real port btw, just need to make myself familiar with the freebsd tools for building such a thing first [13:40] Yeah. I think the --init-system change is a good over-all change. If it's missing actually tell you what the argument to use. not just what options. [13:42] gotcha. I was looking at all the changes I needed to make for the configdrive to work. They are not many. Just a little specific type of things need to be changed when it's fbsd. i.e mounting a cdrom [13:42] yea, it's probably a few details in terms of device naming and such [13:43] cloudinit already knows how to read from configdrives so that should be no problem [13:49] yeah. The changes I made were all around mounting and finding the fbsd drive. So, I should clean those up and get them submitted up as well. But it would be good after your changes go in. that way it can be tested with your changes as well [13:50] hm, perhaps you can branch me and commit to that :) I don't realy know if thats supported though, but it might just work [13:51] (Ill just asume that branch will be able to merge with both mine and trunk) === zz_gondoi is now known as gondoi [14:17] harmw: Just ran it. One error that I did get was in the growfs. It seems that if your disk is already at the max size, it errors out with rc = 1 and the stderr of 'growfs: requested size 20GB is not larger than the current filesystem size 20GB\n' (20gb being my size of disk) [14:17] yea, I noticed that as well [14:17] handling that would be nice, someday [14:19] yeah. It also looks like the ssh-key injection didn't work for that user beastie [14:19] yea indeed [14:19] I didn't look into that just yet [14:19] I noticed for root they were added just fine [14:20] but it worked before which is strange. I think it's becasue the user is created after the key injection [14:20] it didn't create it for me on root either. [14:21] hm hm, then it might've done something else since I've put a user-data upon creating an instance that also does atleast something with users [14:21] I'll have it fixed somewhere soon though :) [14:22] also we can fix the update_etc_hosts as well [14:22] just need to add a template called hosts.freedbsd.tmpl [14:22] oh ok, sounds nice (though I have no idea what that function/module does) [14:22] I just copied over one from hosts.debian.tmpl and renamed it and it works easily [14:37] anyone has some more info about nocloud option ? [14:38] smoser: someone needs your help :) [14:41] nooo [14:41] nvucinic, whats up? [14:41] i suggest using cloud-localds from cloud-utils [14:42] harmw: Figured it out. We commented out the "ssh" part in the cloud.cfg. Once I added that back, all the keys were injected. [14:42] you did that? or was it in my branch? [14:43] hm I did [14:43] lame [14:44] smoser: trying to get nocloud option to work but i belive that i missed some steps [14:44] i have datasource_list: ['nocloud'] [14:44] in cloud.cfg [14:44] datasource_list: ['nocloud'] [14:45] http://pastebin.com/CGmDzHBJ this are error that i get [14:45] harmw: Yeah your branch. No biggie. just feel better that it wasn't something super crazy [14:47] I'll remove that # when I can :) [14:47] i'd like to see your configdrive code btw, despite the fact that I have no use for it yet [14:47] (so can't verify in-depth) [14:50] nvucinic, NoCloud [14:50] case sensitive as it is used for python import [14:51] nvucinic, if you're just randomoly using pastebin's , please use openstack or ubuntu. [14:51] those adds suck! [14:52] and along those same lines, 'pastebinit' is awesome. [14:52] (program that lets you : pastebinit foo [14:52] or: foo | pastebinit [14:52] ) [14:53] smoser: wow, thx === gondoi is now known as zz_gondoi === zz_gondoi is now known as gondoi [17:41] smoser: hey, have a question, I got a https://gist.github.com/pquerna/59a8671502dffc9d8843 status.json with "unhashable type", which sounds like an exception, but I can't find anything in the actual logs about the exception [17:41] hm.. [17:41] where did you see that ? [17:42] oh. i see. [17:43] kind of sucks that i populated 'finished' and not 'end' [17:43] :-( [17:44] and that finished is not finish [17:44] heh [17:45] i'm gonna change it to populate 'end' rather than 'finished' [17:46] na. i'm just going to drop end. thats the more grown up thing to do. [17:49] pquerna, can you recreate by just letting the error raise ? [17:50] in /usr/bin/cloud-init, just [17:50] if i nuke /var/lib/cloud, run cloud-init init, i get the same status.json every time [17:50] except Exception as e: [17:50] raise [17:50] ok [17:50] thx [17:51] smoser: yup, got an exception now. thanks. [17:52] https://gist.github.com/pquerna/a8aed8f963663308af7a [17:52] hmm [17:54] smoser: so one thing i changed, heh, the version of openstack configdrive that read_configdrive was using, was before vendor_data.json was even added in openstack [17:55] smoser: so it was always empty [17:57] oh. hm.. [17:58] yeah, so this is just vendordata_raw is actually parsed json [17:58] this code is expecting a string blob of raw... [17:59] right [17:59] yeah. :-( [17:59] self.vendordata_raw = results.get('vendordata') [17:59] you want to open a bug and submit merge proposal ? [17:59] in DataSourceConfigDrive.py... [18:06] smoser: i guess my question is what is the intended operation, it seems the code wants to treat vendor_data.json like user_data and do all the various operations on it... so really its not the same thing. So seems fix is to keep vendordata_raw as a vendor_data file (that doesn't exist) [18:06] smoser: and then put the vendor_data.json, which is more enviromental/etc as a thing to parse as json and use if needed for info. [18:08] so yeah, we want to do all the same things to vendor_data as we do to user_data. [18:08] https://github.com/pquerna/cloud-init/commit/eec8a00a328b296711ca201f3facdc5910886094 [18:08] eg, something like that [18:09] hm. [18:20] ok. i undretsand more now. [18:21] you were expecting it to be a dict (loaded json). [18:25] and that does seem to make sense. [18:26] pquerna, bah. [18:26] cloudinit/sources/DataSourceOpenStack.py does it right [18:27] smoser: ah, i see [18:27] smoser: ok [18:28] it looks like there might be other improvements there too :-( [18:29] sitl somewhat broken. but closer. === praneshp_ is now known as praneshp === gondoi is now known as zz_gondoi [21:43] hey smoser you around? [21:44] what’s the minimum user-data file I’d need to use to add users to an openstack vm when booting it up?