=== dendrobates is now known as dendro-afk [04:41] I am getting http://pastebin.com/hC1tzxBP on every boot, what is this? [04:43] * flaccid goes to look [04:43] well thats cloud-init [04:43] mountall: Event failed is a bit of a concern [04:43] the cloud-init fail could be a bug in cloud-init [04:44] smoser or erichammond are the people to see about cloud-init. is this the latest official maverick AMI? [04:45] also, are you putting any user data in the launch? [04:50] nope, its the official cannocial one. [04:50] what does that mean? [04:51] flaccid: smoser or erichammond are the people to see about cloud-init. is this the latest official maverick AMI? [04:51] [2:45pm] flaccid: also, are you putting any user data in the launch? [04:51] Its the latest official and i am not putting any user data. [04:51] looks like 2 bugs [04:52] hmm [04:55] checked with 32bit AMI as well, exact same error. [04:56] if its the latest AMIs and you have not done anything additional, then submit bugs to ubuntu [04:58] let me launch one and so i can confirm it [05:06] Error404NotFound: i can confirm it. the messages we are seeing may be intentional, but i don't think that cloud-init-local should return non-zero if no user data is found [05:07] btw, with t1.micro instance there is not swap or /mnt ? [05:07] flaccid: great [05:07] i'm not familiar with mountall, but i've seen it stay resident and leak [05:07] and i don't see a need for it [05:07] Error404NotFound: correct - you may want to read the instance types page on the aws site [05:07] i also noticed these: [05:08] * Starting AppArmor profiles [80G 2011-03-28 04:00:52,257 - DataSourceEc2.py[WARNING]: unable to convert swap to a device [05:08] 2011-03-28 04:00:52,268 - cc_mounts.py[WARNING]: 'mount -a' failed [05:08] yup [05:08] that as well [05:08] not a very clean boot i must say [05:09] there is [ 0.210128] /build/buildd/linux-2.6.35/drivers/rtc/hctosys.c: unable to open rtc device (rtc0) but i know this bug and will be submitting a but on it soon so we can fix it in debian and ubuntu. archlinux fixed it ages ago. [05:09] it also adds /dev/sdb as /mnt in fstab even though t1.micro doesn't have it. [05:09] so please submit bugs for all of the above [05:09] heh, yeah that is a logical bug [05:10] in my debian images i use hd00 and a swap partition so none of this stuff occurs [07:41] Hello room [07:43] I'm in the process of setting up UEC on a HP Pavillion p6774y with a Realtek 8101E/8102E ethernet card and a RaLink Wireless N card, but have been unable to get an Internet connection to even run sudo update; sudo upgrade. The wireless abg works under Win7 but I've been unable to get it to work on CentOS 5.5, UEC latest and desktop Linux Mint (x64 for all installs). I would greatly [07:43] appreciate any help and/or insight the community may have to offer [07:48] DroBuddy: you might want to go to #ubuntu . sudo update and sudo upgrade are not valid commands [07:49] DroBuddy, lspci [07:49] Hrm, I may be mistaken on those commands as I have not had a chance to connect to the internet... So, I was simply trying to recall from the top of my head. But, nonetheless, thank you very much. I'll go give it a shot.. [07:49] the lspci is... [07:49] rtfm [07:49] Ah, sorry, I'm in the process of reinstalling.. Ner.. It's been a long day of messing with this. [07:50] I'm almost to the poing where I'm going to run ethernet down to my basement... [07:50] point* [07:51] DroBuddy, http://ubuntuforums.org/forumdisplay.php?f=336%C2%A0 [07:51] In the tutorials for setting up a cloud they should include like 7x12 backs of beer, at a minimum... But, I digress. [07:51] I'll check it out. [07:51] UEC is kind of not of any relevance until you have your network card working.. [07:52] Exactly :) [07:52] well eucalyptus is very buggy [07:52] whereis cloudstack is not. [07:53] flaccid, +1 [07:53] Can I have the cloud controller and a node installed on the same machine for testing purposes? I know CentOS install didnt complain, but logistically speaking... [07:53] DroBuddy: no. [07:54] The game plan is to have a server OS to host the VMs, as well as, my LAMPP stack. The VMs will only be used for mundane tasks like Modding, etc. [07:54] Hrm, damn. [07:54] I better start throwing some stuff together... [07:54] kind of defeats the purpose of a cloud if you aint going to put your lamp in there etc. [07:55] Well, the idea was to have LAMPP universally available when I load a VM for web dev, or asterisk, OpenCRM, etc. [07:55] The host would also be running openQRM, as well. That's what I plan on using to remotely initiate VMs [07:56] But, this is simply for testing more than anything [07:56] not sure what that means, but each instance can do whatever you want it to do [07:56] I'd like to see it work before I scale it. [07:56] ec2 works much better or if you want private cloud, cloudstack. but i wish you the best of luck with eucalyptus. [07:57] Basically, the server would always be on but the VMs would not necesarily always be running. As such, I want LAMPP running 24/7, irregardless of whether or not a VM instance is running [07:57] instances can run 24x7. [07:57] if the host OS crashes, so do instances. [07:58] Yeah, really, I'm trying to centralize my web development files, and have a few VMs for specialized tasks... But, I just started dabbling with this the other day [07:58] np [07:58] flaccid: Well, I understand that and I do not currently have any plans for redundancy, but this is just a test to see of it is the most efficent and economical way forward. [07:59] Sorry, I have a very slow Internet connection... Which also defeats the purpose of a cloud.. Other than private use. [08:00] the problem with eucalyptus is getting to work first and then having it be of some reliance. [08:00] Depends on how you look at it. In many cases, having only limited bandwidth is a reason *for* doing cloud. [08:00] soren: Exactly why I'm trying to centralize things on my side, really... [08:00] yeah its not like a cloud needs to be on the net if its for doing local tasks [08:01] That includes my PS3/Xbox360 Media server... lol; [08:01] God bless FOSS :) [08:05] Well, back to it. I'll be back when I get things set up which will hopefully be sooner than later [08:29] there is no god, certainly not one that blesses FOSS. [09:00] Morning everyone === daker_ is now known as daker [11:41] erichammond: you would have got yourself a cloud-init bug report :) === dendro-afk is now known as dendrobates === dendrobates is now known as dendro-afk [12:52] smoser, aroun? [12:52] d [12:57] koolhead11: just leave him a question in the channel [12:57] he'll get to it whenever possible .. others might be able to help too [13:08] kim0, i am not sure if am doing some stupid mistake [13:09] koolhead11: put your question in the channel man :) [13:10] kim0, i think i will do some brain storming b4 putting it. [13:10] * koolhead11 gives another 10 mins [13:10] yeah sure === uvirtbot` is now known as uvirtbot [13:18] kim0, so i had had not sourced my novarc and thats why i was not able to start image [13:26] yaay : [13:26] :) [13:38] koolhead11, here now [13:46] smoser, i was trying to run instance from the ttylinux and was not able to get the image [13:46] not able to get ? [13:48] "start" .. but I think it's started now [13:49] smoser, yes its started now [13:49] k [13:49] i have to source my novarc file === dendro-afk is now known as dendrobates === dendrobates is now known as dendro-afk [14:47] hi [14:47] this is neo [14:48] hi [14:48] jdfkasf === cmagina-afk is now known as cmagina [15:12] kirkland, do you have a pointer to a howto that describes doing an automated ubuntu install in kvm? preferrably, without requiring a network connection? [15:12] if the later possible? [15:13] aliguori: i gave this to hallyn last week, actually [15:13] aliguori: i can get it to you, but it does require a network connection [15:14] kirkland, okay, does it take a command line parameter like ks? if so, can you point it to local media (cd or floppy)? [15:14] or does it always do an http request for the answer file? [15:14] aliguori: that network connection, however, could just be the kvm host itself running apache local [15:14] aliguori: actually, yes, you could also build a floppy.img with the preseed [15:14] okay, excellent :-) [15:14] aliguori: here, let me give you this.... [15:16] aliguori: add these command line parameters: "priority=critical local=EN_US url=http://bit.ly/ubuntu-nqa" [15:16] aliguori: to an ubuntu server, alternate, or mini ISO [15:16] excellent [15:16] kirkland, thanks! [15:16] aliguori: note that the url param [15:16] aliguori: you can grab the preseed file I've put at that bit.ly link [15:16] aliguori: and host it local, on http://10.0.2.2/foo [15:17] aliguori: or put that into a floppy.img [15:17] aliguori: smoser has experience building a floppy.img [15:17] aliguori: with boot stuffs in it [15:17] aliguori: one sec [15:17] kirkland, i can google my way there, i've got scripts that do the floppy creation for fedora so it should be easy to adapt [15:17] aliguori: local= should be "locale=" [15:17] ah, that makes more sense :-) [15:21] aliguori: okay, ... here you go ... [15:21] aliguori: 1) boot server ISO [15:22] aliguori: hit ESC, then F6 for more kernel options [15:23] aliguori: add to kernel opts: "priority=critical locale=en_US url=http://bit.ly/ubuntu-nqa" [15:23] aliguori: install should proceed "nqa" no-questions-asked [15:23] aliguori: as for getting around that first ESC/F6, I *think* you can probably get around that by giving KVM a kernel, initrd, and kernel param options [15:24] aliguori: and use a floppy.img for those, plus the preseed file [15:24] aliguori: i'm asked the first part of that question a lot, so I'm going to go and make a blog post [15:28] so, would the thing to do be to codify the use of that preseed in a autotest/client/tests/kvm/steps/Ubuntu-11.04.steps file? [15:40] This page has info on booting kvm with direct kernel/initrd: https://help.ubuntu.com/community/UEC/Images/KVMKernelOptions [15:40] kirkland: ^ if you needed that [15:41] hmm .. that's probably more for booting uec images that installers [15:41] than* === dendro-afk is now known as dendrobates === koolhead11 is now known as koolhead11|afk === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates === hallyn is now known as hallyn-afk === dendrobates is now known as dendro-afk [18:22] aliguori: did you have a chance to play with that at all? [18:22] aliguori: does that get you anywhere further along the line? [18:24] not yet, but will in a bit [18:24] i just wanted to have an automated ubuntu install script for testing purposes [18:31] hallyn-afk: when you get back.. I'm interested in using lxc in a similar manner to schroot. Wondering if you have any experience doing that. [18:34] SpamapS: actually i'v enot quite left yet, gotta right about now, but [18:34] to use it as schroot, my plan was to use LVM snapshots [18:35] SpamapS: so yeah, let's chat when i get back [18:50] hallyn-afk: cool ttyl [18:50] mathiaz: ^5! howdy === daker is now known as daker_ [19:18] SpamapS: hi! [19:18] SpamapS: I've got a question related to memcache: [19:19] SpamapS: when it comes to clear memcache upon deploy new code what does make more sense: restart memcached or flush memcached? [19:21] mathiaz: flush [19:21] mathiaz: usually.. sometimes restart is better because it will force slab reallocation.. [19:22] mathiaz: depends on if you expect the new code to drastically alter the memory access patterns. [19:22] SpamapS: right - It seems that flush wouldn't expire keys set in the last second [19:22] mathiaz: and my info may be outdated.. I seem to recall reading that the slab allocation is more dynamic now. === dendro-afk is now known as dendrobates [19:23] mathiaz: in the past to make rolling upgrades less painful I had a 'cache_version' component to all cache keys, and if I wanted to force new caching.. I just bumped that.. so rolling upgrades would gradually populate the new cache [19:24] mathiaz: which was important, because we had jobs that would run for 30-40 hours [19:25] if the serialized data suddenly changed on them.. they'd freak out and have to be restarted === dendrobates is now known as dendro-afk === cmagina is now known as cmagina-afk === hallyn-afk is now known as hallyn [20:15] ubuntu-one svc unavailable? ouch .. [20:21] SpamapS: all right, so have you started anything? Did you have anything in particular in mind? [20:21] SpamapS: i think what i'd probably do is ship a new set of lxc templates [20:21] put the smarts in there [20:22] hallyn: no I've done nothing but I figured you would have thoughts [20:23] hallyn: my thinking is that the only difference is you need to run 'lxc-start /mounted/whatever/sbin/init' instead of 'chroot /mounted/whatever/bin/bash' [20:23] SpamapS: i'm thinking you have a canonical container for each release, i.e. clean-natty, etc, [20:24] SpamapS: they sit each on an LVM partition, and you create a working container by doing [20:24] hallyn: schroot does that [20:24] see schroot does *all* of this [20:24] lxc-create -n apache-natty -t natty-build - [20:24] yes, it does [20:24] but it doesn't give you network namespaces [20:24] but if you want to use schroot, just use it, don't argue with me :) [20:24] Right but we can build that in can't we? ;) [20:25] [20:25] i think kees was looking at that [20:25] I want an slxc basically ;) [20:25] pretty much what i was going to do [20:25] except based on the lxc code instead of based on schroot code [20:26] ah [20:26] I figured schroot already had most of the machinery necessary [20:26] anyway, i think we're best of with you following up on your direction, me on mine, then we conmpare notes [20:26] does schroot sitll require aufs? [20:26] one of my motivations is to avoid that [20:27] require, no [20:27] to modify schroot, the simple part would jsut be replacing the clone with clone(CLONE_NEWNS|CLONE_NEWPID|CLONE_NEWNET|CLONE_NEWUTS) [20:27] aufs is *awesome* tho [20:27] sure its slower.. [20:27] and can't do long filenames [20:27] the more complicated part would be the setting up of the network [20:27] but not having to rebuild the system to have LVM configured is a nice feature :) [20:28] yup, it's sort of why i haven't done this yet :) [20:28] now LVM really isn't necessary, [20:28] lxc-create right now takes a few seconds [20:28] but i'd still like it to be faster :) [20:29] well I'm less concerned about adding any speedto the equation. I just want instant containers so I can test upstart jobs without the b0rked up chroot support [20:30] SpamapS: ok, you already have that [20:30] just do as in http://s3hh.wordpress.com/2010/06/18/lxc-0-7-0/ [20:31] using '-t natty' in place of '-t ubuntu' [20:33] hallyn: hawt [20:42] SpamapS: all right, actually i think all that I want is a new 'clone' template [20:42] SpamapS: (which is just about written now, took all of 5 mins :) [20:43] SpamapS: so i can keep a 'nattydevel' container around, and quickly clone a new one-off container from that to build a package or reproduce a bug [20:43] gotta run downstairs to the other 'lab', bbl [20:45] hallyn: I'd really like for /home to be mapped in, and /etc/passwd to be copied in [20:45] SpamapS: oh, yes, /home will be mapped in, are you sure you want /etc/passwd? [20:45] SpamapS: actually so long as your prstine container does that, the cloned one will as well [20:56] hallyn: Yeah I want at least the non-system users duplicated so /home is usable [20:57] SpamapS: all right, i'll blog a workflow that i think shoudl work for you [20:57] still finishing up the script right now, really should go test vmware instead :) [20:59] hallyn: sweet [21:07] curse xfs's slow deletion [21:10] xfs rox [21:15] flaccid: :) i'm still using it... [21:16] as are many thousands :) [21:16] xfs_freeze is something other filesystems don't have [21:17] actually there was "fs_freeze" .. a generic one .. but it's probably not mature yet [21:17] but if the filesystem can't do it, it can't do it i.e. kernel level. that sounds like a userspace hack [21:18] http://manpages.courier-mta.org/htmlman8/fsfreeze.8.html [21:18] flaccid: I believe it is kernel level [21:19] not really. not for the filesystem itself and its only Linux Ext3/4, ReiserFS, JFS, XFS [21:19] ZFS owns all of these with zfs snapshots :) [21:21] yeah .. unfortunately crappy linux implementations [21:21] zfs snaps were jaw dropping for me :) [21:21] too bad they ended up @ Oracle [21:23] obino: would you be interested in answering questions like http://ubuntuforums.org/showthread.php?t=1716308 :) [21:24] freebsd runs in the ec2 cloud now thanks to colin@freebsd. working with aws and their ancient xen [21:25] using zfs on that ? [21:26] i think so. can't remember though [21:26] flaccid: I meant are you doing it [21:27] doing what exactly? [21:27] using zfs on freebsd [21:27] ENOCONTEXT :D [21:28] i would if i had a need [21:28] not sure why fbsd was brought up though [21:30] because that is where you do zfs snapshot goodness [21:32] or opensolaris :) [21:32] indiana was it .. [21:32] * kim0 remembers Oracle again .. bleh [21:33] opensolaris has muchly deteriorated [21:33] It's been killed I guess :) [21:34] Waiting for Solaris11 to come out .. and hopefully get a code drop [21:34] doesn't fit the public cloud at all though [21:35] public cloud builders ? [21:35] i don't see a public cloud supporting sparc.. [21:36] solaris on x86 is big too [21:36] hoping Illumos can bring that more light [21:37] days are gone for solaris [21:38] It's got a pretty decent kernel .. it looks mostly dead now, but will probably live in appliance form for years n years to come [21:39] zfs, dtrace, crossbow ..etc are all almost non existent (unfortunately) in Linux [21:40] this is why freebsd has the potential has its already in the cloud and on x86 [22:55] crossbow <3 [22:55] messing with crossbow tought me a crapton about networking [22:56] but not how to spell, s/tought/taught/ [22:56] hheee :) [23:13] SpamapS: http://s3hh.wordpress.com/2011/03/28/lxc-clone/ [23:17] whats lxc? [23:18] linux containers [23:18] like an enhanced chroot. emulate having a virtual machine without emulating any hardware and no guest os [23:19] ah i heard of these but not sure of what use they are [23:20] flaccid: that's what i say about computers :) [23:20] j/k [23:20] hehe [23:24] could you use a lxc as a replacement shell to give users a jail for ssh that is better than the average? === hallyn is now known as hallyn-afk === cmagina-afk is now known as cmagina