[00:27] anyone here that knows vm-builder? === sixstringsg|away is now known as sixstringsg === bladernr_ is now known as bladernr_afk [00:58] zul: when are you planning on doing the e4 upload? friday? [01:00] Daviey, roaksoax: bug 943000 is hitting the QA lab [01:00] Launchpad bug 943000 in cobbler "update to system fails with Exception value: 'NoneType' object has no attribute 'os_version'" [Undecided,New] https://launchpad.net/bugs/943000 [01:02] Do they have white coats and stuff, or is it all inside a server in a rack [01:03] and mice in little cages [01:04] adam_g: yeah [01:05] huge bottle glasses [01:08] @kklimonda : I updated to the new kernel but the device still isn't booting properly [01:12] zul: ok, im close to finishing porting the charms to the new branch. hopefully have CI on that tomorrow, and we should be good to update keystone [01:15] kk === sixstringsg is now known as sixstringsg|away [01:37] I want to set up a VPN server on my server, so I can have a secure tunnel into my network. [01:37] What package do I need? [03:22] smoser: would it be acceptable to you if by default you couldn't run lxc/libvirt inside a container, but some workaround (disabling or changing apparmor profile) would let you? [03:22] I'm trying to decide whether it's worth immediately doing separate profiles per container [03:25] i'd prefer to punt on per-container profiles to make sure we take the time to do it right... [03:25] anyway we can talk about it tomorrow. [03:33] lxc inside a container? You mean like nested ones? === sixstringsg|away is now known as sixstringsg [03:40] yeah [03:41] nothing stopping it right now, but i'd like to clamp down on cgroups inside containers by default [04:10] Why? [04:10] Just because it'd be confusing? [04:15] to stop changes to the devices cgroup setting [04:26] Oh, you can break out of it that way? [04:27] well you could give yourself the rights to /dev/sda and write to it [04:28] now it's not something that would accidentally happen, and i mainly for 12.04 want to prevent accidental abuse of the host by containers [04:28] but still, it's blatant enough... [04:31] Nod [04:31] That's why my containers lack sys_admin cap [04:42] Can someone here tell me how to set the PE size for LVMs in Ubuntu server 10.04? [04:43] Do you mean PV ? [04:43] No the physical extent size [04:43] so I can get LVMs above 256GB [04:44] By default its at 4MB [04:44] For some reason there is no option in the LVM set up to make it higher [05:14] I want to set up a VPN server on my server, so I can have a secure tunnel into my network. What package do I need? [05:23] Smozius: um, I have LVs >256G and I haven't touched the extent size [05:23] Oh.... well then maybe I don't have to worry about it lol [05:23] Smozius: and I thought the default extent size was *4kiB* [05:24] its 4MB on CentOS [05:24] The extent size really only matters if you're trying to avoid write amplification, which is a PITA to do [05:24] and 16MB I heard pushes it up to 1TB [05:24] ok [05:24] It doesn't hurt to crank it up though [05:24] I think you set it at PV creation time [05:24] Well, it would hurt if you set it to 1G and you then wanted a 128M LV [05:24] My LVM is 4mb extents and I have 3TB volume.... [05:25] You can obviously only create LVs in exact multiples of the extent size [05:25] Bigger extents might also mean higher data:metadata ratio, I don't know [06:20] How would I set up an IRC server? Preferably with NickServ and ChanServ [06:23] apt-get install ircd-irc2 [06:23] Dunno about nickserv/chanserv [06:28] twb: Alright [06:29] There are several ircds, irc2 is the simplest and easiest [06:30] I use it for my in-office IRC [06:33] twb: Okay, so without nickserv and chanserv, I'll get a basic irc that anyone can join and speak in? Can I configure channels and accounts and stuff without chan and nick serb? [06:33] **serv [06:33] NFI [06:33] basic irc you just join channels to create them [06:33] I never cared about anything fancier [06:34] #freenode can probably direct you to detailed irc server info [06:34] dunno about modern irc servers but even original ircd had users and passwords, if that's good enough [06:34] twb: Alright I'll ask there about nickserv. Thanks for your help :) [06:36] onre: All I need is a simple irc server, I just want to allow users to register their nicknames ( on here it's /msg NickServ etc etc, but idk how to do it without nickserv [06:38] Where is the binary file for apache2 installed at in ubuntu server? I cannot find it in /usr/sbin/.... [06:38] Also did a find / -name httpd* and didn't come with any binary files [06:38] It'll be called apache not httpd [06:39] Since when apache httpd existed, it was the only *apache* product but not the only *httpd* [06:39] Ah okay, found it [06:39] i assume its /etc/init.d/apache2? [06:39] uh, that's not a binary [06:40] I guess you aren't familiar with System V [06:40] Nope [06:41] So then its /usr/sbin/apache2 [06:41] Looks like it [06:42] Here is how you work it out: http://paste.debian.net/158008/ [06:45] New bug: #943088 in nagios3 (main) "package nagios3-cgi 3.2.3-1ubuntu1.2 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/943088 [07:04] twb: Is /usr/sbin/httpd a deprecated apache2 version or a fork of apache? [07:05] NFI [07:06] It doesn't exist on my system. [07:07] In fact, it isn't provided by and package in Ubuntu. === sixstringsg is now known as sixstringsg|away [08:06] jamespage: ping === smb` is now known as smb [08:24] a [08:25] b [08:26] c [08:27] harh === yann4 is now known as Yann2 [08:41] Daviey: ping [08:41] lynxman: pong [08:41] Daviey: early morning sir :) [08:41] Daviey: was wondering, I have puppet 2.7.11 packaged, would it be any help? [08:42] lynxman: i would think so! :) [08:43] Daviey: cool, let me make it pretty and presentable then and I'll submit a merge request this morning [09:13] davepigott, pong [09:13] morning all [09:13] jamespage: Hi. :) [09:13] jamespage: Just wondering. Do I need to enable vlan on the public network switch as well as the private? [09:13] davepigott, I don't think so [09:14] jamespage: OK. Well, as I said, I can now ping my instances (there's a phrase I never expected to say!) but when I try to ssh I get connection refused [09:14] jamespage: Unless it's on cloud01 [09:14] jamespage: I opened ssh on the switch as well (it was off as well) [09:15] davepigott, where are you ssh'ing from? cloud01? [09:15] jamespage: No. From a client machine [09:15] jamespage: morning! I'll be introducing you today to your new disciple ;) [09:15] lynxman, w00t [09:16] davepigott, hmm [09:17] jamespage: Frustrating, isn't it. [09:17] hm, I don't get it.. lxc really is acting weird and unpredictable :/ [09:17] jamespage: I've come this far, I'm 95% there. I really want to make the vlan solution work. I'm suspecting a switch config again, but I may be wrong [09:18] davepigott, you might be but if you are accessing the instance through its public IP address I don't think that should matter [09:18] jamespage: And it's obviously talking to the instance, because it's "connection refused" not "host unreachable" or something like that [09:19] davepigott, thats useful [09:19] davepigott: ssh is running and you're not blocked by a firewall config or denyhosts or something like that? [09:19] davepigott: can you grab the console output from one of the failing instances for me [09:20] kai - ssh is running, not sure about denyhosts, and we're all on the same lan, no firewall [09:20] jamespage: Will do. One moment [09:21] jamespage: First of all, here's an output from ssh -v -v -v: http://pastebin.ubuntu.com/861673/ [09:21] davepigott: if you didn;t install denyhosts, you should be fine [09:22] davepigott: is this a cloud instance you are trying to connect to? [09:22] * Daviey has missed context. [09:22] Daviey: Yep [09:22] davepigott: Well, it seems you are either: Using the wrong key, the instance has no key, or the key failed to get set [09:22] davepigott, it might be that the instance can grab stuff from nova-api (metadata) on anywhere other than cloud01 [09:22] console should tell us [09:23] davepigott: can you euca-get-console-output i- ? [09:23] Daviey: I can connect to an instance running on the "controller" node - i.e. one running all nova/glance/swift - but not one running on a nova-compte/network node. Just getting output…. [09:23] O_o [09:23] lets see the console log.. [09:24] Daviey: jamespage http://pastebin.ubuntu.com/861675/ [09:24] ahhh [09:25] Ahhh? Please tell me it's simple. :) [09:25] davepigott: so, you are SURE you can ssh to it from the controller node? [09:25] davepigott, that looks to be the issue [09:26] Daviey: No, I ssh to an instance on the controller node from a client, but if I ssh onto an instance that's on another node I can't [09:26] ih [09:26] oh [09:26] davepigott, I suspect that if you grab the console log from one running on cloud01 it should be OK [09:26] davepigott: Well i'm guessing it's SNAT'd traffic. [09:26] jamespage: Yes, it is. I've done it. [09:26] SNAT'd? [09:26] Daviey: ^ [09:26] davepigott, I suspect that your configuration for the nova-api server might be borked on the other nodes [09:26] davepigott, you are running nova-api on all nodes right? [09:27] jamespage: According to all the documentation I've read you only run nova-api on the controller, but yes, I am [09:27] davepigott: Can you comment what the --ec2_dmz_host on compute node is [09:27] ? [09:27] Daviey: Yeah, one moment [09:29] Daviey: In nova.conf? [09:30] Daviey: If so, it's not set [09:31] hmm [09:31] Hmmm. Nor is ec2_api [09:32] Daviey: My nova.conf file (same on all nodes) http://pastebin.ubuntu.com/861681/ [09:32] davepigott: is ec2_host? [09:33] Daviey: nope [09:34] davepigott: try adding: --ec2_host=192.168.1.14 [09:34] --ec2_dmz_host=192.168.1.14 [09:34] --ec2_url=http://192.168.1.14:8773/services/Cloud [09:45] Daviey: OK. And then restart compute and network? [09:47] davepigott, yes [09:49] jamespage: Daviey: Are you sure nova-api should be running on *all* nodes? All the documentation says not [09:50] davepigott, I think you can OR you can run it in one location - but for the life of me I can't remember the rationale [09:50] davepigott: You can do either. WE tend to recommend running it on each node because nova-api doesn't scale well. [09:50] I think that nova-api provides the metadata service - so having it on all nodes scales better [09:50] ah- like Daviey said :-) [09:51] But the nova.conf points at the one node as being nova-api [09:51] So are the others redundant? [09:52] davepigott: Hmm, lets just see if that setting works.. if it does, lets change it to point the the local ip address. [09:52] Daviey: ok [09:55] biab [10:04] right, too much magic... [10:04] * kai just figured out part of his lxc trouble :) [10:05] now let's wait through the bootstrap [10:05] Daviey: jamespage: "ssh: connect to host 192.168.1.59 port 22: Connection refused" :( [10:16] davepigott, please check the console log again [10:17] jamespage: http://pastebin.ubuntu.com/861711/ [10:20] davepigott, hmmm [10:20] * jamespage scratches his head [10:21] jamespage: Join the club. === conntrack is now known as eyeofra [10:25] davepigott: tried a sniffer to see if the packets get to the server? === eyeofra is now known as conntrack [10:29] kai: No. Actually, the packets are obviously getting to the instance, because pings are working, and if I do an ssh -v -v -v I can see that the instance is responding, it's just refusing the connection [10:30] kai: And when the boot up happens, the instance doesn't appear to have a connection to the outside world. [10:30] davepigott, I'm pretty sure the issue is on startup - the instance does not initialise properly so you won't ever be able to get to it [10:30] its the link between nova-compute/nova-api for ec2 metadata thats causing the issue [10:30] jamespage: So why is it starting up badly on the compute nodes? [10:30] let me take a look in the test lab charms to see if anything is missing [10:30] jamespage: OK. Thanks [10:33] davepigott: you might need to put the bridge in promiscuous mode.. [10:34] Daviey: Hmm. Uh, and how do I do that? [10:34] davepigott: sudo ip link set dev br100 promisc on [10:34] On all nodes? [10:34] davepigott: well, if it were me.. i'd be experimenting with one compute node to start with :) [10:34] Daviey: Ah. Good point. :) [10:36] davepigott, the only flag we set during testing is --ec2_dmz_host [10:36] jamespage: Which I have set [10:37] davepigott, whats in /var/log/nova/nova-api.log on the failing nodes? [10:45] jamespage: http://pastebin.ubuntu.com/861735/ [10:47] jamespage: But since nothing references nova-api on the other nodes, it won't make a huge amount of difference [10:55] davepigott, OK - so I'll paste the config from our functional lab - it might help [10:55] Compute nodes - http://paste.ubuntu.com/861749/ [10:56] jamespage: So compute nodes are different from the main one? [10:57] jamespage: In terms of nova.conf I mean [10:58] davepigott, and the cloud controller - http://paste.ubuntu.com/861752/ [10:58] davepigott, yes they are [10:59] jamespage: OK. Let me do a little comparison work. Thanks! [10:59] please note that this config is for essex on precise so there will be quite alot different I suspect [10:59] jamespage: Yep. Noted === davepigott_ is now known as davepigott [11:16] hi === davepigott_ is now known as davepigott [11:30] jamespage: ping [11:31] lynxman, pong [11:31] jamespage: would like to introduce you to bbcmicrocomputer [11:31] lynxman, sure [11:31] bbcmicrocomputer: jamespage is our java expert (author of http://javacruft.wordpress.com/) [11:32] jamespage: feel introduced :) [11:32] lynxman, ta [11:32] hey bbcmicrocomputer [11:32] jamespage, Hi, nice to meet you [11:32] * jamespage shakes hands [11:33] bbcmicrocomputer, so lynxman tells me that you have a strong Java background? [11:35] davepigott: so, on the other nodes.. did you set it to use ec2* from the compute ip address? [11:35] Daviey: No. ec2* looks to the main server [11:36] davepigott: make it = the compute node.. see how that works out [11:36] (providing you have nova-api running!) [11:36] Daviey: Yeah, I do. OK [11:40] jamespage, yeah, although I keep it silent [11:40] Daviey: Nope. Same deal. Can ping but not ssh [11:41] davepigott: it's the euca-console-log which is more useful tbh [11:42] Daviey: OK [11:42] Daviey: One moment === Leseb_ is now known as Leseb [11:46] Daviey: http://pastebin.ubuntu.com/861801/ [11:48] well that makes no sense [11:50] davepigott: on the compute node, can you $ echo 1 | sudo tee /proc/sys/net/ipv4/ip_forward [12:14] Daviey: Sorry - got a phone call. It gives me "1" [12:15] New bug: #943208 in sysstat (main) "package sysstat 10.0.3-1 failed to install/upgrade: ErrorMessage: subprocess installed post-installation script returned error exit status 2" [Undecided,New] https://launchpad.net/bugs/943208 === Leseb_ is now known as Leseb [12:33] i want to setup remote logging for my applications, would rsyslogd be up to the task? [12:54] eutheria: guess it depends what logging mechanism the applications use, but basically yes, rsyslogd can be used for remote logging [12:54] eutheria: there's a whitepaper on rsyslog at http://www.canonical.com/about-canonical/resources/white-papers [12:56] i was going to use python's logging module [12:59] what are you guys using to analyse your logs ? [13:06] davepigott: sorry, that enabled ip routing... see if that makes a difference. [13:07] Daviey: Oh! OK. One moment [13:44] morning [13:47] zul: morning! [13:47] hey lynxman [13:53] smoser, I'm researching bug #943119 which led me to your bug #903897. what is your use case for aufs ? [13:53] Launchpad bug 943119 in linux "aufs.ko missing from the Precise kernels" [Undecided,Confirmed] https://launchpad.net/bugs/943119 [13:53] Launchpad bug 903897 in linux "-virtual kernel missing modules" [High,Fix released] https://launchpad.net/bugs/903897 [13:55] tgardner, i have no specific use case for aufs. [13:55] smoser, you complained that it would break scripts that expect aufs. [13:56] i do think that Bogdan (opener of that bug) is probably not going to be alone [13:56] of course it will break things that use aufs. [13:56] it deleted a kernel function [13:56] do you why overlayfs is inadequate? [13:57] know why* [13:57] i completely understand you not wanting to support something that is not in mainline especially when there is a solution that is . [13:57] i suspect that its only inadequacy is that it has different interface to it. [13:57] smoser, overlayfs is not mainline either, though its more likely to get that way then aufs [13:57] and things have been built (like this NAS that the guy mentioned) to use aufs. [13:57] Daviey: No. Didn't make a difference. :( [13:58] tgardner, i suspect there will be more complains about it missing. [13:58] but i completely understand your position [13:58] and would probably make the same decision myself [13:58] smoser, well, it turns out overlayfs doesn't do inotify correctly. things like 'tail -f' don't work. which is why I'm reexamining aufs [13:59] ah. well inotify is pretty important. [13:59] indeed [14:01] tgardner, sorry i couldn't be more help. [14:01] smoser, however, I'd be better armed with information if I had some specific cases where overlayfs is insufficient, and for which aufs solves the problem. apw is likely to fix the inotify bug, but I'd like to know what else is missing. [14:01] my guess is if you wait, others will tell you :) [14:02] smoser, well, by then its a bit late [14:02] yeah. [14:02] maybe ask on ubuntu-devel ? [14:02] its at least a wider audience. [14:02] but nothing compared to the onslaught that we'll see soon. [14:02] yeah, I guess thats a good place. [14:03] tgardner, you could cross post to ubuntu-cloud [14:03] or server [14:03] cloud might expose some use cases specifically [14:03] smoser, I needed more email in my inbox === bladernr_afk is now known as bladernr_ [14:04] exactly. [14:27] tgardner: gary_poster: ^ if tail -f doesn't work in overlayfs, that's gonna suck for the ephemeral containers whose rootfs is based on overlayfs [14:27] (i'm surprised it hasn't shown up more, in fact, if it's that blatant) [14:28] hallyn, it seems to be.. Andy is working on it, but he says the patches are pretty ugly. [14:28] overlayfs was *so* clean before it actually needed to do things like LSM an inotify [14:29] yeah. [14:29] maybe it's time to take yet another look at valerie's mount stuff [14:30] hallyn, unionfs is a ginormous patch set, and I don't think its fully coked yet [14:30] cooked* [14:30] it's been years since i looked at it [14:30] but i did understand why overlayfs was so seductive [14:30] (maybe still is - i [14:30] i'll have to look at apw's patchset :) [14:30] its much simpler. it at least works for the live CD use case [14:31] stgraber: jjohansen: i think i'm going to add a 'lxc.lsmdomain' or somesuch config variable to do simplest possible per-container apparmor policies for now [14:31] if not specified, it'll choose a default, ncie and tight one [14:32] ubuntu templatse well generate per-container ones to customize cgroup controls [14:32] (cgroups are the main reason i feel i need to do this) [14:32] (would love for someone to say i shouldn't worry about that :) [14:35] hallyn, I don't have the scrollback; lemme see if I can find the online log, assuming there is one [14:37] http://irclogs.ubuntu.com/2012/02/29/%23ubuntu-server.txt doesn't show this yet [14:53] gary_poster: sorry, just that overlayfs apparently doesn't support inotify, and tail -f doesn't work [14:53] gary_poster: tgardner was looking for aufs use cases [14:54] gary_poster: and I would have thought if tail -f didn't work in ephemeral containers, that would've bothered you by now :) [14:54] hallyn, Leann is gonna send out an information request on ubuntu-devel. [14:54] ok [14:58] hi all [14:59] i've attempted to install a gui on my server, but messed it up and i'd rather remove everything now. so, which packages do i remove to remove both gnome and unity completely? [15:01] lynxman: How is that puppet branch progressing? [15:01] :) [15:02] Daviey: just messaged you in another channel ;) [15:02] Daviey: all ready for review! [15:02] Daviey: https://code.launchpad.net/~lynxman/ubuntu/precise/puppet/newupstream/+merge/95185 [15:04] lynxman: rocking [15:05] Daviey: tested it and its working (on my servers) [15:06] Daviey: did I tell you I'm running oneiric in prod? :) [15:11] lynxman: No.. :) [15:12] zul: hi, what's the status of keystone for precise? I've seen MIR bugs and keystone light mentioned but no recent comments [15:13] Adri2000: should be uploaded this week [15:14] keystone or keystone light? (actually I don't even know what's the status of this upstream). and it's going to main? [15:17] keystone light has already been merged into keystone so it really doesnt matter at this point :) [15:18] hallyn, sorry, been having Precise issues on my desktop. We only run our test command in overlayfs. It is surprising that that limitation has not bitten us; it may be because we've been using a single subset of the tests to get everything else set up. Maybe not wise. We should hopefully have a full test run this week to examine. [15:18] zul: ok, didn't know that, thanks :) [15:18] maybe it would not bite us even in the entire test suite, but that does seem unlikely, given how much launchpad does and tests [15:19] yeah [15:19] so, we know aufs works/worked [15:20] certainly it would be nice to have that as an option. Since the inotify issue is already slated to be addressed that's promising, but I'll plan to have full test run results to look at the overlayfs situation [15:20] this week [15:24] i think enabling aufs is a lot more work than justified by "it would be nice" :) [15:25] gary_poster: tgardner said an email will go to ubuntu-server asking for use cases [15:26] hallyn, actually, we've been keeping up with aufs updates, but haven't tested it much. [15:27] oh, ok, i figured it was a whole new porting effort [15:27] though i recall bad bugs even in maverick [15:28] hallyn, yep, bad bugs are my concern as well as soon as aufs sees any usage === bladernr_ is now known as bladernr_afk === bladernr_afk is now known as bladernr_ [15:36] gah. laptop froze AGAIN. [15:36] i'm starting to think vim is causing it [15:39] hallyn: vim is evil ;) [15:41] yeah, better use oowriter ! [15:41] * benji performs the sign of vim and kisses his vim necklace. [15:42] vim is probably jsut too heavyweight. back to ed [15:42] that's not a bad idea [15:43] rest of the day, i'm using ed! [15:46] unfortunately, unlike yesterday, the pic i took with my phone of my screen isn't as clear so it's hard to retype [15:47] hallyn: just use xxd like real men do [15:47] xxd? [15:48] oh [15:48] hallyn: :) [15:48] xxd = porn lite [15:48] lol [15:48] i'm just not sure i need hex for my emails... [15:48] or porn [15:48] my recipients might disagree i guess [15:48] hallyn: I'm sure they'll be glad to hex translate your emails [15:49] my content is worth it! [15:49] all right, back to it. [15:52] (also playing some metal in case the laptop just got groggy adn fell asleep) [15:55] hallyn: ooh what metal? [15:58] adam_g, i just updated bug https://bugs.launchpad.net/nova/+bug/942865 [15:58] Launchpad bug 942865 in nova "upgrade from diablo leaves existing images with kernel unbootable" [Undecided,New] [15:59] it seems like generically, you can't make an ec2 published image public if it has a kernel and ramdisk. [15:59] so, rbasak until we get that fixed, i can't make public lucid images. [16:03] zul: right now volbeat, for some lighter fare [16:03] hallyn: never heard of them [16:03] might have to switch tos omething heavier, laptop is dozing [16:04] lol [16:04] laptop or hallyn is dozing [16:04] zul: you don't care about non-libvirt lxc right? [16:05] hallyn: i dont [16:05] k [16:13] zul: if you're going to upload python-keystoneclient, can I give you bug #934064 ? [16:13] Launchpad bug 934064 in python-keystoneclient "Installing openstack-dashboard on Precise removes Keystone package" [High,Confirmed] https://launchpad.net/bugs/934064 [16:14] Adri2000: yes that will be fixed when the new keystone is uploaded [16:14] bug assigned :) === Ursinha` is now known as Ursinha === bladernr_ is now known as bladernr_afk === bladernr_afk is now known as bladernr_ [16:51] hi guys [16:52] what can i use for newsletter management in ubuntu server? [16:54] diogo_79: do you know about mailchimp et al? They're web services, not ubuntu server. But sending out email newsletters is a complicated fast moving game, due to kickback from spam filters and suchlike. It may be easiest to use such a service. For a simple mailing list manager, you could use mailman which is packaged on ubuntu server I believe [16:54] jamespage: scale of 1 to 10 how complete do you consider the hbase charm? [16:55] jcastro, hmm - about 7 ATM [16:55] working on it ATM [16:55] it currently broken :-( [16:57] don't need it now, just wondering if when people ask about our hbase story I can say "we're on it." [17:02] jcastro, quack quack oops - it would help if it was actually pointing at HDFS [17:02] * jamespage faceplants [17:02] jcastro, we are definately on it [17:17] Daviey: ping [17:26] jamespage, still around ? [17:26] https://jenkins.qa.ubuntu.com/view/ec2%20AMI%20Testing/view/Overview/job/precise-server-ec2/4/ [17:26] zul: did you already merge the debian stuff into the CI packaging branches? [17:26] the two red dots failed similarly. the instance failed to start. [17:27] adam_g: yeah did it this morning [17:27] zul: er [17:27] zul: [ERROR] Failed charm: nova-compute, state: install_error [17:27] adam_g:meaning....? [17:27] the one thing i don't undertsand though is https://jenkins.qa.ubuntu.com/view/ec2%20AMI%20Testing/view/Overview/job/precise-server-ec2/4/ARCH=i386,REGION=ap-northeast-1,STORAGE=ebs,TEST=all-types,label=ubuntu-server-ec2-testing/console [17:27] that has 'failed to copy' for cloud-init-output.log [17:30] Daviey: bug 943000 [17:30] Launchpad bug 943000 in cobbler "update to system fails with Exception value: 'NoneType' object has no attribute 'os_version'" [Critical,Confirmed] https://launchpad.net/bugs/943000 [17:30] hggdh: hola [17:30] nicceee [17:30] Daviey: I hope you are doing good, sir, and sorry to dump yet another one on you [17:30] zul: meaning nova-compute had some problem installing. looking at logs now [17:32] zul: http://paste.ubuntu.com/862208/ [17:32] zul: it looks like installing nova-compute-kvm didn't install nova-compute, and there was no upstart job [17:33] adam_g: k ill have another look [17:33] zul: installing nova-compute manually seems to have gotten it started, but i havent looked any closer [17:36] adam_g: http://paste.ubuntu.com/862214/ [17:36] does the juju charms just go apt-get install nova-compute-kvm? [17:40] drat, hard kernel crash while doing nested kvm [17:42] smoser: thats OK - its the collect_data script trying to grab everything - but for that test cloud-init is not used I think [17:43]