[11:28] Hi guys. Is there a way to suppress output of "apt_update" and "package_upgrade"? [13:10] aps, as in you dont want them going to /dev/console ? [13:10] http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/doc/examples/cloud-config.txt#L599 === Toger_ is now known as Toger === Toger_ is now known as Toger [18:48] mgagne if u get anytime reviews on https://code.launchpad.net/~harlowja/cloud-init/cloud-init-net-sysconfig/+merge/297115 would be cool [18:48] that should be the thing that lets u get rid of https://github.com/mgagne/cloud-init-fedora-pkg/blob/epel7/cloud-init-0.7.5-network-info-support.patch [18:48] if rharper then does a networkd (?) renderer, then the world will all be merry [19:00] harlowja: ok, will I need to rebuild a package? [19:03] likely, especially since that hasn't merged yet :-P [19:07] hello, i'm making a cidata volume for local attach to a vm (i.e. not cloud). i understand that network config has to be in metadata, not userdata. however it looks like straight debian format. does this format work for a centos image? will cloud-config "know" how to translate it? [19:36] smemsh,yes, cloud-init will convert the debian interfaces style format into appropriate distro config. [19:37] this is very m uch in progress thoguh, and recently improved. [19:37] smoser: it won't require me to be using systemd will it? [19:37] and harlwoja is working on the renderer for centos [19:37] is cloud-init/freebsd-configdisk included in 0.7.6? I installed py26-cloud-init on FreeBSD 10.3-RELEASE. and I'm having issues with it seeing a mounted ISO. [19:38] smoser: i'm using the centos6-generic cloud image, which uses afaik the "old" /etc/sysconfig/network-scripts way [19:38] harmw_, ^ ? drq i have very little experience there. if you wanted to let me in i could maybe quick poke around.. [19:38] smemsh, that should be fine. [19:39] i see that google cloud is listed as the providers that use cloud-init, but it doesn't seem to be in their stock images [19:40] On our deploy scripts for linux on VMware, we mount an ISO with the config data on it. We haven't started using FreeBSD yet, but I'm trying to get it ready. I have an ISO mounted (not real data), but it appears that it doesn't look at /dev/cd0 for anything. [19:41] datasource_list is [19:41] datasource_list: ['ConfigDrive'] [19:41] drq, do you have blkid available ? it uses that to find things. [19:42] i found it easier to use a vfat with a label. it's just a mkfs -t vfat -n cidata of any old thing [19:42] is that installed as a dependency? blkid isn't found on the host. [19:42] is blkid on non-linux? [19:43] pkg e2fsprogs-libblkid-1.42.13 might have it. Let me install that. [19:43] libblkid-1 [19:44] i assumed that just parsed stuff in /sys, surprised it works on freebsd [19:44] drq, see tools/build-on-freebsd [19:44] in trunk [19:45] OK. let me try to find that. [19:48] there are so many cloud-init's in github. which is it. [19:48] none :) [19:49] https://launchpad.net/ubuntu/+source/cloud-init [19:49] we shoudl be moving to git on launchpad soon. [19:50] which i might well maintain a mirror in github to help clear that confusion [19:52] i had same question. openstack has one i'm assuming would be reasonable fresh [19:53] they don't say in the repo comment that it's a mirror though, they should [19:53] got launchpad up in browser. I'll peruse that and see what I find and chew on it overnight. Thanks. Be back in the morning. [19:53] latest commit october 2015 hmm [19:54] smemsh, well, the stuff under openstack was intended to be for cloud-init 2.0 [19:54] which kind of fizzled as we 've focused more on the 0.7 branch [19:54] smoser: yeah howcome latest is 0.77, but next is 2.0 ? what happened to 1.0? [19:55] well, 2.0 was picked as next just because it was intende to include windows support, which woudl be heavy rewrite. but that has been put on hold at least in the near future [19:57] looks like there is https://git.launchpad.net/~usd-import-team/ubuntu/+source/cloud-init [19:57] smemsh, that is actually ubuntu packaging. [19:58] oh the debcheckout [19:58] right. [19:58] the usd-import-team branches are for all packages. they're what "ubuntu distributed development" used to be for bzr [19:58] they're just getting started really, but they will represent what was actually in the archive [19:59] smoser -> installing e2fsprogs installed blkid and it sees the ISO and does complain about the ISO I have mounted. Now we're cooking. Thanks. [20:05] debian's is based on bzr1215 [20:05] well, the jessie backport is bzr1156 and jessie is 976 [20:08] i used lp:~smoser/ubuntu/yakkety/cloud-init/pkg guess that is 0.7 tip [20:10] oh nice there is lxd module [20:10] smemsh, the network configuration rendering is *heavily* updated in the last yakkety upload (ie, really recent) [20:10] and will get more for centos soon. [20:10] with harlowja's work. [20:10] eventualy [20:11] hrm, well i'm hoping the latest upstream centos generic cloud images have whatever they need for it to work correctly... i will find out shortly. working out the syntax atm (docs are a little sparse for some things) [20:13] wow cool you can provide cc data via /proc/cmdline??? that's really neat [20:15] what is the instance-id used for? i can set this to gibberish right? [20:16] i'm just booting in local vm [20:24] smemsh, right. [20:24] it shoudl be unique to this "instance". [20:24] if you change it that image will then think it needs new. [20:26] to support reprovision without a data wipe / reset ? [20:26] right. [20:27] cloud-init supports shutdown and snapshot and new instance from taht snapshot [20:27] and i think that is a pretty common work flow on amazon.. [20:27] a.) start a ubuntu image [20:27] ah so like an iterative cloning [20:27] b.) ssh in or automated do some stuff [20:27] c.) shutdown [20:27] d.) snapshot [20:27] e.) new gold master .... profit! [20:30] yeah that makes sense. however i would advocate for reset, the changes should be done in the automation layer (in my case ansible) and easy to reproduce the initial parts. but it could be time consuming, i can see where it's useful to support the shortcut (and not everyone needs a layer after cloud-init) [20:33] i was thinking that after reading the list of all the modules, it's like, duplicating a lot of the modules in the config/orch tool. was thinking hm if i can do all this in cloud-init, do i even need ansible? but it's like you can't go backwards or be declarative with cloud-init so i dunno [20:34] i can sort of see that as happening though in some future. like cloud-init evolving into a more general-purpose orchestration tool that knows the full state of the machine [20:35] the truth store / metadata can be updated and cloud-init can poll it or subscribe or something and then determine what needs to change [22:06] with the NoConfig i.e. using a cidata-labeled filesystem, do i have to insert ssh key data into the config file or can i use #include and file:// urls just like for real metadata sources? idea being, to copy the key data from outside onto the vfat filesystem rather than inserting in /user-data [22:07] oh, it would still have to be the same format anyways, i couldn't just insert contents of a file. so that won't work. but the question is still valid, can i split up the files that way [22:11] i'm guessing /user-data can contain anything that it could from whatever source. so i guess the only question is whether file:// urls will look up from other files on the config filesystem or if it's only http scheme supported [22:55] are 'hostname' and 'local-hostname' just aliases? [23:41] doc/examples/cloud-config-resolv-conf.txt has "manage-resolv-conf: true" but should be underscores