[12:45] harmw, you there? [12:45] nah, not realy [12:46] daytime = worktime [12:46] smoser: just give it a try and hope I'll reply :p [12:48] so on freebsd you said that blkid depends on e2fsproggs ? [12:49] yea, I had to install that first (from either pkg or ports) [12:49] it then drops a binary in /usr/local/sbin [12:50] hiren_: you can probably confirm this :) I'm not aware of something in BASE that does the same though, so perhaps there is a reasonable alternative we could/should use instead [12:55] well, blkid is pretty much a dependency at the moment. [12:56] if it depnds on e2fsprogrs (possibly for identifying / getting info on extX partitions) then it seems like a bug if freebsd doesn't identify that dependency [12:57] in order to get that blkid binary, I must install that e2fsprogs pkg [12:57] but its no problem [12:57] just another dependency [12:58] did my comments make sense btw? [12:59] mostly, yeah. i have one fix there. and thank you for your work on that. [12:59] so freebsd doens't resolve dependencies in any way? [12:59] the user has to explicitly know that ? [12:59] (that blkid depends on e2fsprogs?) [12:59] ah no [13:00] perhaps i wasn't clear on it :) blkid isn't a package on its own [13:00] the binary comes with the e2fsprogs package [13:03] so i think we need to add /usr/local/sbin to PATH in the sysvinit/freebsd ? [13:04] harmw, ok. that clears that up. thanks. [13:05] that sounds quite reasonable, yes [13:05] (that's where all non-BASE binaries end up) [13:10] alright. well, take another look when you ahve a chance. [13:11] the PATH stuff i just simply declared. and you're welcome to improve the static definition. another route is to do: PATH="$PATH:/usr/local/sbin:/usr/local/bin" [13:12] Ill have a look tonight [13:12] thanks! === zz_gondoi is now known as gondoi === Guest73730 is now known as mgagne === gondoi is now known as zz_gondoi === zz_gondoi is now known as gondoi === alexpilotti_ is now known as alexpilotti [17:28] smoser: I no longer have to hardcode any datasource in the config now, so thats another change you may apply [17:38] harmw: hey! [17:38] noes! [17:39] didn't get your que about blkid [17:39] oh it's nothing :) [17:39] thats just some binary that c-i needs [17:39] ah. to understand blkid? [17:39] something like that [17:40] hiren_ so we all safe from the bash stuff now :-P [17:40] for a few days?? [17:40] lol [17:40] until the next crazy issue thats been existing for 20+ years, lol [17:40] lol [17:40] blkid is great stuff [17:40] proofs once again opentsouce hasn't got anything to do with security :P [17:41] hah [17:41] yeah, bash: the gift that keeps on giving :-) [17:41] :> [17:42] harmw, explain that ? [17:42] in the freebsd config file there is a line hardcoding the DS to use, currently thats openstack [17:43] I just ran it without that hackery, and it found the attached configdrive without issues [17:43] so whatever caused it not to find any DS earlier on is probably fixed [17:43] ok. so i'll just drop that. [17:43] yea, ill run a test with DHCP enabled on this instance in a sec [17:43] hm.. [17:44] well, lets just add configdrive [17:44] to the list. [17:44] for now. [17:44] could go that route as well [17:44] rather than letting it go willy nilly [17:44] willy nilly... [17:44] wtf [17:44] i'm sure there are other sources that have some issue. that are probably ignorable [17:44] but will pointlessly put 'WARN' in your log [17:44] ah yeas [17:45] yeah, for now lets do that. [17:45] how can I test/read which DS has been used (in freebsd.py)? [17:46] whihc has bee nused ? [17:46] yea, I want to set static networking when it's used configdrive [17:46] seemed legit [17:46] well, config drive doesnt necessarily imply static networking [17:47] cloud-init has a bunchy of networkign cleanups to do. really. an dfreebsd will have to come along. [17:48] if info.get('bootproto') == 'static': [17:48] I'm not realy getting that :P where is bootproto set/specified? [17:49] i'm not sure. there is a lot of growing that needs to occur here. [18:22] aw crap, wasn't it some annoying feature in openstack that it would specify network setup in a configdrive? [18:23] <.< >.> [18:23] uhm, whre it would specifically NOT do that [18:24] harmw: right now, Openstack transmits network information to cloud-init by way of a debian-style /etc/network/interfaces file [18:24] and thats in a configdrive? [18:24] harmw: there's a spec up for K (that Rackspace implements /today/ in vendor_data.json) that represents the network information from neutron for parsing as json [18:24] harmw: yeah [18:24] ok [18:25] so that should be something in /var/lib/cloud representing the ntwork settings [18:25] in my setup [18:25] it's a pointer to something in openstack/content/ [18:25] obj.pkl has the netsetup [18:26] or atleast the address [18:26] There's basically a template that nova uses to write network configuration out to the configdrive [18:26] I guess it's possible for that to not be enabled [18:26] oh, but thats local-ipv4 I'm seeing - no gateway [18:27] ok JayF [18:31] http://docs.openstack.org/user-guide/content/enable_config_drive.html [18:31] that shows how to inject the interfaces file with a networksetup inside, but thats just hackerish [18:33] that's not the most awesome document I've ever seen [18:33] nova /does/ 99% of that for you iirc [18:33] well I was hoping that it would do just that :P do you have a link to some more helpfull/updated docs? [18:34] I don't know much about generically operating Openstack [18:34] I *do* know about how we operate openstack [18:34] so I can show you what we do [18:34] sure thing :) [18:35] harmw: injected_network_template= in nova.conf [DEFAULT] [18:35] aha [18:42] so JayF , that configitem is specific to configdrive? [18:42] I think so? [18:43] ask me questions about Ironic if you want more certain responses :P [18:43] haha [18:43] :P [18:44] /etc/nova/nova.conf:injected_network_template = /usr/share/nova/interfaces.template [18:44] so, I'm already using that in my config [18:50] does that file exist and seem reasonably formatted? [18:50] we had to make a custom one :) [18:50] oh that file exists, and looks reasonable [18:51] harlowja: how would c-i drop that file or the content in /var/lib/cloud? [18:53] hmmm, maybe from http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/cloudinit/sources/DataSourceConfigDrive.py#L197 ? [18:56] I can see that's where it wants to apply it, yes [18:57] but I'm to lazy to find out how openstack hands the network config (through configdrive) to c-i :p [19:00] should just be a file present on the config-drive [19:01] http://paste.ubuntu.com/8460729/ [19:03] ah so, well, I'm missing that content folder [19:15] harmw that would do it :-P [19:16] could you give it back, please [19:18] nope, ha [19:19] u might want to ensure the following [19:19] https://github.com/openstack/nova/blob/stable/havana/etc/nova/nova.conf.sample#L1862 [19:19] i think it needs to be 'always' [19:19] force_config_drive=always [19:20] but I don't want configdrive, only on specific instances... which is why I've got this 1 instance booted with --configdrive (or something) [19:20] hmmm, i didn't think u could target it to specific instances [19:21] nova boot --configdrive [19:21] hmmm, guess that must be new [19:21] --config-drive=true [19:21] since havana, I guess [19:21] python-novaclient-2.13.0-1.fc19.noarch [19:22] (damn old, but hey) [19:22] You absolutely can use --config-drive=true [19:22] in newer novaclient [19:22] you should probably upgrade fwiw [19:22] * JayF wishes everyone would use pip versions of openstack clients [19:23] uhm, I was just saying I *am* using --config-drive=true :) [19:23] ya, for some reason u aren't getting into the code @ https://github.com/openstack/nova/blob/master/nova/virt/netutils.py#L69 [19:23] or @ https://github.com/openstack/nova/blob/master/nova/api/metadata/base.py#L170 [19:23] https://github.com/openstack/nova/blob/master/nova/virt/netutils.py#L107 is a common cause afaik [19:24] probably line 107 there thats stopping this [19:24] could be, yea [19:25] i belieeve if u can comment that crap out, a file will appear, ha [19:25] lol [19:26] or somehow set your network to 'injected' [19:26] and that would be on the controller's end, right? [19:26] ya [19:26] is this your own openstack, or someone elses :-P [19:26] so how is injected set in the first place :p [19:26] my own [19:27] somewhere when a network is setup i think [19:27] something like that, ha [19:27] don't you just love openstack at these moments [19:27] I know I do :> [19:28] that just scratches the surface ;) [19:28] i've seen things man [19:28] :P [19:28] lol [19:28] seen horrible things [19:28] 3 trillion config options... [19:28] and a multitude of possible fuckups [19:28] openstack is like a crossword puzzle with about 10k pieces [19:29] all sorta the same color to [19:29] yup [19:29] luckily, I'm colorblind [19:29] ha [19:29] bb [19:50] hiren_: ping [20:03] ok, so it's definately failing because there is no 'injected' meta [20:04] ya [20:04] thought so [20:09] compute is logging 'meta': {'injected': False, [20:09] god this is tiring [20:09] damn you, openstack! [20:16] :-/ [20:16] how did u setup your networks that nova is using? [20:16] with neutron? [20:19] hmmm, ya, got me then, ha [20:26] harlowja: Ive hacked out L108 and L107 and now I finally have content :> but it contains a litteral $bla on about every line [20:26] L108 and L117, oops [20:45] je [20:49] hmmm, literal blah, must not be replacing stuff it needs from the thing u commented out, haha [20:50] my guess is https://github.com/openstack/nova/blob/master/nova/virt/netutils.py#L146 is empty [20:50] or missing === gondoi is now known as zz_gondoi === harlowja is now known as harlowja_away