[00:15] erichammond, still around ? [00:15] you're more than welcome to start packaging there. in fact i think taht would be great. [00:16] which autoscaling where you wanting to start on? [00:17] the rdscli package is i think the best package i have put together. [02:56] smoser: Just checked in again, but now heading home. I was thinking of the "as-*" command line tools from AWS. http://aws.amazon.com/developertools/2535 [02:56] smoser: Will check out rdscli. Thanks. === mosno2 is now known as mosno [07:30] Morning all [08:20] hi all === daker_ is now known as daker === da is now known as davinci === davinci is now known as davinci11 [09:22] TeTeT, ping [09:23] have a ubuntu-cloud problem, [09:25] when I restart instance, installed apache will disappear, it seems like cant permanently store into VM, can anyone facing this problem ? [09:25] can = did [10:01] davinci11: you have this with UEC? Probably you want to have a persistent instance, that stores all data on an EBS volume? [10:01] koolhead11: pong, what's up? [10:03] TeTeT: yes, and I using image from store [10:04] TeTeT, so UEC on oneric will have openstack (diablo) ? [10:04] davinci11: please have a look at http://people.canonical.com/~tspindler/UEC/ebs-based-instance.pdf , maybe it helps [10:04] koolhead11: I don't know, please ask one of the developers [10:05] koolhead11: I have no oneiric system yet [10:07] Thanks TeTeT, i will try it tomorrow, see if that work, got to go now see u [12:39] koolhead11: supposed to be diablo, yeah [12:41] TREllis, and LTS on essex :P [12:43] eh? [13:25] TREllis, next after diablo in the 6 month release cycle of openstack :D [13:28] koolhead11: ah didn't know the next codename [13:47] TREllis, :d === dendrobates is now known as dendro-afk [15:21] hallyn: around? I'm having some issues getting lxc containers to start on oneiric [15:26] hallyn: ahh, looksl ike cgroup wasn't mounted correctly [15:43] SpamapS: hm, cgroups-bin should be getting auto isntalled? [15:54] hallyn: it was installed, but didn't do the right thing. :P [15:58] SpamapS: can you file a bug? [15:58] (if you can reproduce) [15:58] hallyn: sure, it may be a while.. too many plates spinning to stop and file a bug ;) [16:03] understood === dendro-afk is now known as dendrobates [16:14] hallyn: shouldn't lxc-ps --lxc aux show me all the processes running inside containers? [16:15] hrm.. my lxc containers are starting in recovery mode.. [16:16] recovery mode? [16:16] yeah [16:16] weird === uvirtbot` is now known as uvirtbot === koolhead11 is now known as kooolhead11|afk [17:13] hallyn: seems that net-device-up or filesystem is not being emitted.. [17:14] oh.. .. lxc-guest [17:16] did you not create with ubuntu template? [17:16] I'm still using my uec template [17:16] it should either install lxcguest, or do the right rimming [17:16] oh [17:16] ok [17:16] s/rimming/trimming btw === dendrobates is now known as dendro-afk [17:17] I may switch that though, now that I understand how these things work.. I can always just chroot in and install the cloud-init package after the container is created [17:19] hallyn: been thinking about how to get the benefits of LVM w/o having to re-partition the machine.. was wondering if you had tried LVM on top of loopback files. [17:23] hmm.. no that wasn't it [17:24] SpamapS: I/we should really spend a few hours getting aufs and/or overlay mounts integrated [17:25] there's even an open bug for it thanks to rcollins :) [17:25] is overlay mounts a better aufs? [17:25] it's a much much simpler mechanism [17:25] basically we'd mount an empty new tmpfs, and then mount that over the container's rootfs [17:26] (overlay mounts only let you mount *one* dir on top of *one* other) [17:26] (but of course you can do that again and again :) [17:29] hallyn: I would predict that our next meeting won't leave much time for optimizations. :-P [17:31] hallyn: so the only way to add packages to the lxc container is to chroot into it right? [17:33] or pivot_root into it or start the container :). (but yes) [17:33] oh, as for overlay mounts being a better aufs - in that it's actually upstream (is it now?) and therefore usable on all linux, yes. [17:34] its in 3.0 yes [17:34] yay! [17:34] then I should really whip that up this week [17:34] saw it fly by the other day and wondered what it was [17:34] oh, presumably mount binary needs an update [17:35] I still think many of the tools you're working on are 99% the same as schroot/sbuild [17:35] W/ a slightly different focus === daker is now known as daker_ [17:39] oops! forgot to write meta-data [18:13] hallyn: btw, the ubuntu template is really nice [18:19] hallyn: although --trim has just confused the heck out of me. :-P [18:19] actually no.. thats not it [18:22] hallyn: hmm.. I think your debootstrap may be missing a key package [18:23] hallyn: w/o ubuntu-keyring, apt doesn't work [18:24] might also be a bug in debootstrap [18:26] yeah, debootstrap should be picking it up as a direct rdepend of apt [18:40] weird.. so now cloud-init can do apt-get update, but installing packages fails [18:42] SpamapS: you're creating oneiric containers right? [18:42] yeah [18:42] had to patch the ubuntu template to add ubuntu-keyrung [18:42] keyring even [18:42] otherwise apt will not function [18:43] hm, this may be a bug elsewhere as of today. bc i had weired problems with an schroot i created today [18:43] and then I chroot in and instal cloud-init and rsyslog ... [18:43] and the thing boots.. sort of ok.. but fails when trying to install the packages I asked for in my cloud-config [18:43] hallyn: yeah that package sync may be borking things [18:44] SpamapS: can you reproduce iwth plain debootstrap, and then ask on -devel? [18:44] hallyn: I'll try again later.. mucking with this as a favor to lifeless but I have other things to attend to. :-P [18:45] ok [18:45] * hallyn is trying now [18:45] of course it also might be that you have stale /var/cache/lxc/oneiric [18:45] but lesee if i reproduce [18:45] no I cleared that a few times [18:56] reproduced [18:59] reproduced [18:59] uh [18:59] yeah gnome-terminal is also funky :) [20:58] hi, I'm having problems getting clould-init to run cloud-config on ami-1aad5273 [20:58] i see this in cloud-init log [20:58] Failed loading of cloud config cloud-init [21:00] any ideas anyone? [21:28] SpamapS: oh, feh - that's right, lxc-ps (which you mentioned before) isn't going to work right with cgroups right now [21:32] hallyn: bummer [21:35] hallyn: I'd do natty lxc's, but cloud-init is borked in them and I don't feel like dropping in my hacked up cloud-init-local.conf ;) [21:35] SpamapS: well you can just remove cgroups-bin and manually mount cgroups, [21:35] SpamapS: but are you actually depending on lxc-ps? [21:35] no no no [21:35] was just confused [21:35] i've been ignoring it bc it kinda sucks anyway [21:35] ok [21:36] i'm fixing the package shortcoming right now [21:36] hallyn: was referring back to the broken bootstrap [21:36] ok, long as someone is around to push my fix once i test it, it can be fixed tonight [21:41] your fix for the bootstrap? [21:42] I wonder if we need to even bother "booting" these containers [21:42] Just tell lxc to run init and sshd.. let cloud-init do the rest. [21:42] * SpamapS should stop thinking out loud [22:07] SpamapS: you've got your choice of a debdiff attached to the bug, or the bzr tree linked to it :) care to push? [22:07] ( https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/817233 ) [22:07] Launchpad bug 817233 in lxc "oneiric containers need ubuntu-keyring package" [High,In progress] [22:09] hallyn: shouldn't you have PPU on lxc? ;) [22:15] * SpamapS is running sponsor-patch now [22:22] hallyn: uploaded [22:46] SpamapS: yeah I should, but I always decide to wait one more day bc (1) I'm not sure if to add it to my PPU set I need to do a whoel new wiki page, and then I think I should just do ubuntu server set, and then I figure I need time to get all the info [22:47] so I never get around to it [22:48] Yeah, the DMB is too heavy a process to get enthusiastic about [23:00] SpamapS: i do still see a bug with containers on some hosts not shutting down. need to look into that tomorrow [23:01] hallyn: is the stop action not waiting for them long enough, or is it that the 10 second timeout on all shutdown tasks is too short? [23:01] I'm not exactly happy with that 10 second thing