[07:45] <xenol> hello, I'd like to know if cloud-init supports configuring networking on CentOS 7 as it does on Debian-based systems
[09:07] <Odd_Bloke> xenol: It should do; if not, please report bugs. :)
[09:08] <xenol> Odd_Bloke: network-interfaces should also work on centos based machines?
[09:10] <Odd_Bloke> xenol: Are you seeing behaviour that suggests otherwise?
[09:11] <xenol> Odd_Bloke: nope, I haven't tried it out as all the examples are for Debian.
[09:11] <Odd_Bloke> xenol: Which examples are you looking at?
[09:49] <xenol> Odd_Bloke: well, the CI nic configuration resembles debian way too much and I am not sure if the same mechanism works for CentOS
[09:49] <xenol> Maybe I was just mislead by network-interfaces syntax
[09:56] <Odd_Bloke> xenol: Yeah, I think you should be fine using that on CentOS.
[15:28] <eckes> Hello, I have a cloud-init 0.7.4 on OEL6.6 which is started via openstack kilo (kvm) and it has the problem that on first boot it sets the hostname to "imagename" but writes "imagename.novalocal" into /etc/sysconfig/network (and resolv.conf). I actually want the domain be part of hostname, but it should set that on first init boot as well. Any ideas?
[15:38] <smoser> eckes, what do you mean domain ?
[15:39] <eckes> domain name, de ec2datasource reports "novalocal" as domain. It should be part of the FQDN (uts_name, not the resolved one).
[15:40] <eckes> (or leave the domain off the hostname, its fine too. I just want to avoid the machine has two different host names, depending if it is the first or second boot)
[15:50] <smoser> oh. i think maybe this bug is fixed in trunk
[15:50] <ByPasS> if the domain is in the metadata you can always force it within the runcmd: section aswell
[15:51] <smoser> https://bugs.launchpad.net/cloud-init/+bug/1246485
[15:51] <eckes> thanks for the reference
[15:51] <eckes> I guess I need to produce my own image :)
[15:51] <ByPasS> u can mount the image and edit it aswell
[15:51] <smoser> fix was : http://paste.ubuntu.com/11873037/
[15:52] <eckes> yes the bug pretty much looks like my problem, thanks
[15:58] <eckes> ok works :)
[22:06] <SpamapS> o/
[22:06] <SpamapS> Question regarding cc_resizefs.py ..
[22:07] <SpamapS> How would people feel about a governer put on it that prevents it from resizing a root partition to the point where grub cannot read it, if no /boot is detected?
[22:08] <SpamapS> smoser: ^
[22:08] <SpamapS> harlowja_: ^
[22:08] <harlowja_> uh oh
[22:08] <SpamapS> governor? anyway... :-P
[22:08]  * harlowja_ goes back into the box
[22:08] <harlowja_> *still in the box*
[22:08] <harlowja_> lol
[22:08] <SpamapS> harlowja_: who let you out?!?
[22:08]  * harlowja_ it was the weekend, i got bored :(
[22:09] <harlowja_> i escaped
[22:09] <harlowja_> lol
[22:10] <harlowja_> SpamapS a governer seems ok to me, seems to make sense to not kill the partition so that grub dies
[22:10] <harlowja_> as long as it doesn't go past 65MPH
[22:10] <SpamapS> harlowja_: we'd have to leave the current behavior alone.. because PXE booted things and externally booted kernel things like xen might be annoyed.
[22:11] <harlowja_> k, so a configurable governer that ensures it doesn't go past 65MPH
[22:11] <SpamapS> but I'm more inclined to just resize it to 1TiB and basically say "resize it again if you really need that"
[22:12] <SpamapS> it seems like a unique situation that I have.. baremetal servers with one gigantic root disk exposed to the OS
[22:12] <harlowja_> ya, at y! we've been trying to get out of resizing anything for anyone automatically, they can do it themselves
[22:12] <harlowja_> it just takes to long
[22:12] <SpamapS> harlowja_: thats my current workaround.. just disabling resie
[22:12] <SpamapS> resize
[22:12] <harlowja_> ya
[22:12] <SpamapS> though it has also made me think what we really need is LVM because zomg thats a big RAID
[22:13] <harlowja_> we've been moving to a static thing provided by openstack/ironic/vms and if users need more, they can ask cloud-init to do it, or not, but its not something they can then say 'openstack took to long'
[22:13] <harlowja_> cause people complain it takes to long, and we remind them its not provisiioning time thats doing this crap
[22:14] <harlowja_> its there desire to have super-big disk (that they probably don't need, lol)