[00:57] <dominicdinada> if i do a network boot would i be able to install a copy of natty via internet.... PC is headless, no cd, no floppy.... ya it has usb but the bios only supports usb drives such as floppy-cd
[01:03] <Smaug> hey all
[01:07] <Smaug> i have ubuntu 8.04.4 with python 2.5.2.    How should I go about installing python 2.7?  I am aware that I will need to have both 2.5.2 and 2.7 exist on the machine
[01:37] <adam_g> RoAkSoAx: https://code.launchpad.net/~gandelman-a/ensemble/fqdn-fix <-  this is small and adds the cobbler system name to cloud-inits metadata, which'll do the trick for now
[01:37] <adam_g> RoAkSoAx: i lost my scrollback of what you and smoser were pondering, so this is probably a hack until that gets done. but i believe it'll get us going in the meantime
[02:10] <dominicdinada> can someone help me setup a network boot ? ugh frusterating ?
[02:10] <Smaug> i have ubuntu 8.04.4 with python 2.5.2.    How should I go about installing python 2.7?  I am aware that I will need to have both 2.5.2 and 2.7 exist on the machine
[02:26] <toddnine> Hi guys.  Can anyone recommend a good replacement for AD?  We're using Apple's open directory at the moment, but we want LDAP for auth and Kerberos for network shares( with samba) that supports OS X, Ubuntu Workstations, and Windows 7 workstations
[02:26] <dominicdinada> samba itself
[02:28] <toddnine> dominicdinada: do you have any examples or doc I can use as a reference?
[02:28] <toddnine> I know how to use LDAP (I'm a java dev) and kerberos, but I'm not sure how to hook it all up as a service on a domain every computer can join
[02:29] <dominicdinada> toddnine: https://help.ubuntu.com
[02:29] <dominicdinada> atm i am working on something
[02:46] <toddnine> Network Authentication?
[02:46] <toddnine> https://help.ubuntu.com/11.04/serverguide/C/index.html
[02:48] <toddnine> Also, are there any pre-canned utilities to sync using the google sync tools for app customers?
[04:09] <linocisco> how to recover ssh on my server ?
[04:10] <linocisco> how to disable screen saver on my server ?
[04:28] <patdk-lap> how do you *loose* ssh?
[04:30] <airtonix> looking to get the mod_shared_roster_ldap installed on lucid server
[04:31] <linocisco> patdk-lap: ssh localhost on server is fine. ssh server-ip on server is fine. but from other computer using ssh client is not ok
[04:31] <linocisco> patdk-lap: after one unproper shutdown or forced shutdown or power loss , i found that the next day
[06:06] <stiv2k> ok
[06:07] <stiv2k> can someone help me figure out why i cant use 'freedns' as a protocol in ddclient?
[06:07] <stiv2k> it says
[06:07] <stiv2k> WARNING:  file /etc/ddclient.conf, line 5: Invalid Value for keyword 'protocol' = 'freedns'
[07:15] <jmarsden> stiv2k: Check the protocl names listed in /usr/share/doc/ddclient/README.gz -- are you sure your version of ddclient supports a protocol called freedns ?
[07:17] <stiv2k> jmarsden: i guess it doesn't, but googling suggests that >3.8.0 supports freedns
[07:17] <stiv2k> so im a little confused on that
[07:18] <stiv2k> i've been a zoneedit user for years and i'm getting sick of them constantly reducing the number of free features
[07:18] <jmarsden> Are you confusing a dynamic DNs provider name with a dynamic DNS protocol name?
[07:18] <stiv2k> jmarsden: what
[07:19] <stiv2k> no
[07:19] <stiv2k> well, zoneedit and freedns are both provider names
[07:19] <stiv2k> freedns uses their own protocol iirc
[07:29] <jmarsden> stiv2k: You can see what protocols ddclient works with by doing       fgrep '=> {' ddclient |sed -e 's/ => {//'
[07:29] <jmarsden> Or just read the Perl code :)
[07:30] <jmarsden> Make that    fgrep '=> {' /usr/sbin/ddclient |sed -e 's/ => {//'
[08:21] <tiger2wander> I'm recovering my ext4 partition with e2fsck -b block_number /dev/sdxx but after mount it has no file/folder inside mount point
[08:21] <tiger2wander> but `df` command show it is used 68% :(
[08:21] <tiger2wander> Could anyone help me to solve this? please!
[08:22] <tiger2wander> I'm using Ubuntu Server 10.04.3 amd64
[08:22] <tiger2wander> Tried almost superblock backup but no luck
[08:25] <CluelessPerson> I can't get php5 to work on my server. >.<
[08:31] <ersi_> tiger2wander: Have you checked if there's any LOST+FOUND folder where everything's stacked into?
[08:32] <tiger2wander> not yet, I keeping not touch anything until got more information
[08:33] <ersi_> that's wise
[08:33] <tiger2wander> before use e2fsck -b I was made a mistaken that act mkfsck -b (after run mkfsck -n to find superblock)instead e2fsck
[08:34] <CluelessPerson> php5 isn't processing, edditing the httpd.conf file in apache2 is blank, I'm new to linux and lost at this. >.>
[08:34] <tiger2wander> so that partition got formated as Ext2 :(
[08:34] <tiger2wander> I think Journal meta data was cleared by that command, right?
[08:36] <tiger2wander> just checked lost+found folder and see there are many file with prefix as #
[08:36] <tiger2wander> so how can I recovery from lost+found?
[08:37] <CluelessPerson> >.>
[08:38] <Daviey> Gooooooooooooood Day people.
[08:38] <tiger2wander> CluelessPerson, what's version of Ubuntu running on your server?
[08:39] <CluelessPerson> tiger2wander, 11.04
[08:39] <CluelessPerson> tiger2wander,   I believe.  Also, I screwed up the configurations for apache2 and completely reinstalled that.
[08:39] <CluelessPerson> but that works now, just not the php
[08:41] <tiger2wander> did you forget to install package: libapache2-mod-php?
[08:42] <CluelessPerson> tiger2wander,  .. >.> most likely.
[08:42] <CluelessPerson> here I go
[08:43] <CluelessPerson> tiger2wander,   alright, that installed, but I got several errors
[08:44] <CluelessPerson> tiger2wander, Syntax error on line 6 of /etc/apache2/httpd.conf: Cannot load /etc/apache2/modules/mod_php.so into server: /etc/apache2/modules/mod_php.so: cannot open shared object file: No such file or directory
[08:47] <CluelessPerson> tiger2wander, Ah, works now.
[08:47] <CluelessPerson> tiger2wander,  Thank you very much.
[08:47] <CluelessPerson> tiger2wander, checking with php.info()
[08:48] <CluelessPerson> tiger2wander, Thank you very much.  Everything works. :D
[08:50] <tiger2wander> CluelessPerson, you are welcome :)
[08:50] <CluelessPerson> tiger2wander, How does it feel to be very helpful?
[09:04] <tiger2wander> CluelessPerson, what's thing you are asking for?
[09:08] <CluelessPerson> tiger2wander, How does it feel to be very helpful?
[10:06] <jamespage> Daviey: would you be OK to sponsor a Jenkins upload for me later today - latest LTS bugfix release came out yesterday (does not need a FFE)
[10:07] <Daviey> jamespage: Maybe.. what is it worth?
[10:43] <jamespage> Daviey: hrm - or maybe not - pulls in a new dependency to fix a bug which we don't have packaged.
[10:48] <Daviey> jamespage: grr
[10:49] <jamespage> Daviey: I may just patch that out - the feature that uses it is borked in the current version so may be better to remove than get lots of bug reports.
[10:49] <Daviey> jamespage: check to see if it's worth the effort at this stage to start making changes.
[10:51] <jamespage> Daviey: think its minimal effort - just looking now
[10:53] <jamespage> Daviey: or maybe not - going to leave as is
[10:54] <Daviey> jamespage: ok :(
[10:56] <jamespage> Daviey: :-(
[11:58] <Daviey> jamespage: I just thought, it needs a minimal deb and a udeb, otherwise at build time - it's still going to link against the bloated library.
[11:59] <jamespage> Daviey: specifically which library are you using in the install?
[11:59] <jamespage> sorry - udeb
[12:00] <Daviey> jamespage: libxmlprpc-core-c3-0, that pastebin i gave you, i went back one too many rev's
[12:00] <jamespage> Yep - got that
[12:01] <Daviey> jamespage: dang, it would be so much easier to just build this from a seperate source package, but not ideal.
[12:04] <Daviey> jamespage: http://pb.daviey.com/GdAD/ , is a start.
[12:08] <jamespage> anyway - eat then scratch head
[12:08]  * jamespage lunch
[12:12] <rbasak> Is there a beta1 server iso image?
[12:12] <rbasak> I can't find one on cdimage
[12:15] <Daviey> rbasak: http://releases.ubuntu.com/oneiric/
[12:15] <rbasak> Thanks Daviey. I think I'm still as lost as ever!
[12:16] <Daviey> heh :)
[12:45] <tanathos> Hi, is there a way to scan port 25 to see if php sends emails?
[12:47] <jamespage> Daviey: I don't think you can strip down the udeb any further
[12:47] <jamespage> if you don't use one of libwww or libcurl - you don't get a client
[12:48] <jamespage> and the other disable options are for other parts of the syse,
[12:48] <pmatulis> tanathos: yes, use a packet sniffer like tcpdump
[12:48] <jamespage> all of those other deps are pulled in by libcurl
[12:48]  * jamespage goes to take a look there
[12:48] <tanathos> thanks
[12:48] <patdk-wk> tanathos, normally php doesn't send email via port 25, unless you programmed in smtp support yourself
[12:49] <Daviey> jamespage: ah!
[12:52] <jamespage> Daviey: which is a good example of how to build multiple versions of the same source code BTW
[12:54] <soren> Daviey, jamespage: udebs? I hope we're not still talking about Jenkins? :)
[12:55] <jamespage> now there is a thought....
[12:55] <soren> Otherwise, what are you smoking^Wup to?
[12:55]  * jamespage wonders whether libwww might be lighted on deps
[12:55] <soren> What's it for?
[12:56] <jamespage> its for the cobbler/orchestra hardware discovery/registration tool which runs from the installer
[12:56] <jamespage> we need a udeb for xmlrpc-c to chat with cobbler
[12:57] <jamespage> but ATM it uses libcurl3 which pulls in a load of other dependencies
[12:57] <zul> ewwww
[12:57] <soren> jamespage: Can't you fake it with straight curl or wget?
[12:58] <jamespage> soren: not sure - Daviey did you look at doing that first? ^^
[12:58]  * soren does not quite remember how complex Cobbler's RPC protocol is
[12:58] <Daviey> soren: generating xml via curl / wget isn't going to be clean
[12:59] <soren> Yes? You're not making a very convincing argument. :)
[12:59] <Daviey> using a token etc.. sure it's possible
[12:59] <Daviey> The cleaner solution seemed to be to use a C library for this :)
[13:00] <Daviey> soren: http://bazaar.launchpad.net/~davewalker/+junk/cobbler-enroll/files/6
[13:02] <zul> my eyes they burn
[13:11] <jamespage> Daviey: I think building a stripped down udeb for libcurl might be possible
[13:14] <Daviey> jamespage: rocking, but unless i am mistaken it needs a libcurl3-minimal*.(deb,udeb) ; to build against aswell.
[13:15] <jamespage> yep
[13:19] <Daviey> soren: Fancy fixing nova/test.py to support skip'ing of classes? :)
[13:19] <Daviey> it should be supported via unittest, but nova is hiding that functionality.
[13:25] <zul> Daviey: i think soren has gone afk
[13:26] <Daviey> bah
[13:51] <RoAkSoAx> smoser: ping
[13:52] <smoser> RoAkSoAx, yo.
[13:52] <smoser> i was looking for you.
[13:52] <smoser> :)
[13:52] <smoser> so where do we stand? i dont particularly like the "manage-etc-hosts".
[13:53] <smoser> i'd like the meta-data solution (populated by cobbler), or, save that, i think we could do a better job of getting a default hostname in cloud-init by explicitly parsing /etc/hosts.
[13:53] <smoser> (as the installer will have, i think, written an entry in /etc/hosts with what you give it for hostname)
[13:53] <RoAkSoAx> smoser: ok... but I was testing
[13:54] <smoser> yeah, thats fine.
[13:54] <smoser> and what did you find ?
[13:54] <RoAkSoAx> smoser: and if we set a --hostname for a system (machine.domain.com) then the hostname and domain name are passed as kernel arguments
[13:54] <RoAkSoAx> smoser: however, in the cloud init fix, you only fixed for hostname
[13:54] <smoser> well, thats because domain name is non-determinable
[13:54] <RoAkSoAx> smoser: that's why cloud-init keeps putting the domain name as ubuntudomian (or whatever cloud-init puts there)
[13:55] <smoser> localdomain.
[13:55] <smoser> the reason i booted on domainname is because the only thing i knew at the time to get it was running 'hostname -f'
[13:55] <RoAkSoAx> smoser: right, but if we tell cloud init to *not* set the domain when it is not passed on the config (as done with the hostname), then cobbler will automatically set the hostname
[13:55] <smoser> which explicitly says in the man page, DO NOT RELY ON THIS
[13:56] <smoser> right. and that will work. i agree.
[13:56] <smoser> but i think we have the chance to make cobbler look more like ec2, and i think we should take that chance.
[13:57] <RoAkSoAx> smoser: yeah we have the chance, but IMHO, we should make cobbler set the domain name either way
[13:57] <smoser> the problem we're trying to solve, though, that ensemble scripts use 'hostname -f' a lot, is really a problem.
[13:57] <smoser> if you've not read that man page, you shoudl read it, basically thats a bad idea.
[13:57] <smoser> :)
[13:57] <RoAkSoAx> smoser: hehe yeah I read
[13:58] <RoAkSoAx> smoser: but my point being is that regardless what type of integration we decide to do with cobbler, we should make sure that cobbler is able to set the domain name regardless of whether we are using cloud init or not, so that means, cloud init should not be changing the domain that's passed thru the kernel
[13:59] <smoser> cobbler should be able to set the domain name, yes, and apparently it can. so i have a hard time arguing with your point, and thus the solution of 'manage-etc-hosts'
[14:00] <smoser> i assume that in /etc/hosts of a provisioned system you have an entry for
[14:01] <smoser>  127.0.1.1 <HOSTNAME> <FQDN>
[14:01] <smoser> right?
[14:02] <smoser> actually, its probably : 127.0.1.1 <FQDN> <HOSTNAME>
[14:02] <RoAkSoAx> smoser: right, the domain is passed thru the kernel on the PXE file... which is the similar case as what happened with bug #838280 we just need to make sure that cloud-init doesn't change the hostname
[14:02] <RoAkSoAx> smoser: let me check
[14:03] <rbasak> What should the permissions/ownership of /var/lib/libvirt/qemu/? I think I've screwed mine up.
[14:03] <smoser> $ ls -l /var/lib/libvirt/qemu -d
[14:03] <smoser> drwxr-xr-x 5 libvirt-qemu kvm 4096 2011-09-12 19:59 /var/lib/libvirt/qemu
[14:04] <rbasak> Hmm, that's what I have
[14:04] <rbasak> I ran out of room in /var
[14:04] <rbasak> So I moved /var/lib/libvirt to /home
[14:04] <rbasak> And symlinked it
[14:04] <rbasak> (stopping libvirt-bin and restarting it after)
[14:04] <smoser> lxc ?
[14:04] <rbasak> But now I'm getting permission issues with sockets not able to be created in /var/lib/libvirt/qemu
[14:04] <rbasak> No just kvm
[14:04] <smoser> no. duh. kvm.
[14:04] <smoser> hm..
[14:04] <rbasak> (using libvirt)
[14:05] <smoser> i've seen other issues (lxc) with libvirt not liking symlinks
[14:05] <smoser> oh
[14:05] <smoser> i know why. i think
[14:05] <smoser> it might be app armor related
[14:05] <rbasak> Ah
[14:06] <rbasak> Yes, there's stuff in /etc/apparmor.d/libvirt
[14:06] <rbasak> (that refers to the old paths)
[14:06] <RoAkSoAx>  q/win 3
[14:06] <RoAkSoAx> errr
[14:09] <smoser> RoAkSoAx, /etc/hosts ?
[14:10] <rbasak> It looks like libvirt dynamically configures those files, and there's no configuration variable to set the path
[14:10] <rbasak> So I might be stuck :-/
[14:11]  * rbasak never expected to put virtual machines in /var
[14:11] <rbasak> (they belong to me, not the system!)
[14:12] <Daviey> rbasak: repossession of your /home.
[14:12] <RoAkSoAx> smoser: re-deploying a system. will take ~5 minutes
[14:13] <floown> hello
[14:13] <floown> I have a problem to receive mails truth my Ubuntu server
[14:14] <floown> # /usr/bin/mailq
[14:14] <floown> mailq: fatal: bad string length 0 < 1: myhostname =
[14:14] <floown> why I have this error please?
[14:14] <rbasak> Daviey: :)
[14:14] <jdstrand> when applying policy, apparmor resolves the symlinks. the that apparmor driver in libvirt uses realpath to avoid this problem
[14:14] <floown> I have read and apply this tutorial http://www.howtoforge.com/perfect-server-ubuntu-11.04-ispconfig-3-p6
[14:15] <jdstrand> you should look in /var/log/kern.log, dmesg or /var/log/audit/audit.log (if you have auditd installed) to see if you are getting apparmor denials. if not, then apparmor shouldn't be the problem
[14:16] <iggi_> Hello,
[14:16] <rbasak> jdstrand: thanks, I'm getting stuff like type=1400 audit(1316008945.990:24114291): apparmor="DENIED" operation="mknod" parent=1 profile="libvirt-32560cff-6b32-f675-d7cf-ba096875473e" name="/home/libvirt/qemu/oneiric-b1-test.monitor" pid=25261 comm="kvm" requested_mask="c" denied_mask="c" fsuid=133 ouid=133
[14:17] <jdstrand> right-- so that is the monitor file
[14:17] <rbasak> yeah, they all seem to be
[14:17] <rbasak> For pre-existing machines it can't bind to the socket
[14:17] <rbasak> For new machines it can't create the socket
[14:17] <jdstrand> (as opposed to the disk, which uses realpath)
[14:17] <rbasak> (incidentally, I see no socket)
[14:18] <jdstrand> yes, you won't see the socket because apparmor stopped it
[14:18]  * jdstrand looks at something
[14:18] <iggi_> I seem to be having a problem with bonding in Ubuntu Server 11.04. I was able to get two NICs bonded, but every time I reboot the machine one of the NICs goes into 100mbit mode (they are supposed to be gig)
[14:20] <jdstrand> rbasak: right, so that path is hard-coded by the monitor. moving all of /var/lib/libvirt somewhere else could cause a number of issues
[14:20]  * Daviey wonders if the thing jdstrand has turned to look at, is the same thing that is concerning me.
[14:21] <jdstrand> s/by the/for the/
[14:21] <jdstrand> Daviey: it isn't. I am waiting for kees to come online
[14:21] <smoser> RoAkSoAx, ... actually, platform.node() seems to be doing what i would have been doing.
[14:21] <Jeeves_> iggi_: Are they different chipsets?
[14:21] <rbasak> jdstrand: OK, thanks. It sounds like I can just move the images and symlink those, or if not I'll work around it. Thanks!
[14:21] <jdstrand> rbasak: all you should really *have* to move is the /var/lib/libvirt/images directory
[14:21] <iggi_> Jeeves_, Nope, both are the same and both are onboard
[14:22] <rbasak> jdstrand: yep, I'm with you
[14:22] <Jeeves_> iggi_: But to different switches?
[14:22] <smoser> err.. ug. i cant tell now.
[14:22] <iggi_> Jeeves_, both to the same switch, the swithc shows gig capability and auto negotiation on the port, but only 100mbit speed
[14:22] <smoser> but sometimes it will give a fqdn.
[14:23] <Jeeves_> iggi_: And if you switch the cables?
[14:23] <jdstrand> rbasak: that said, if you must move everything, you can adjust /etc/apparmor.d/abstractions/libvirt-qemu for your paths. you are also allowed to edit the /etc/apparmor.d/libvirt/libvirt-<uuid> file
[14:23] <iggi_> Jeeves_, that does pose a problem, since its an offsite machine, but it looks like that might be the next course of action
[14:24] <jdstrand> rbasak: the /etc/apparmor.d/libvirt/libvirt-<uuid> file is autogenerated only if it doesn't exist
[14:24] <smoser> so... i await access to your fresh installed box
[14:24] <jdstrand> rbasak: but don't edit /etc/apparmor.d/libvirt/libvirt-<uuid>.files cause it is dynamically generated on vm start
[14:24] <Jeeves_> iggi_: That's all I can think of,  really
[14:25] <jdstrand> rbasak: so that is a lot of info! :) I recommend putting everything back in /var/lib/libvirt, then adjusting your domain xml to put the disks wherever you want
[14:25] <rbasak> jdstrand: that's useful to know, thanks. Although I don't see any mentions of /var/lib/libvirt in /etc/apparmor.d/abstractions/libvirt-qemu, but moving everything but the images back seems to have worked
[14:25] <RoAkSoAx> smoser: ok so /etc/hosts sets 127.0.0.1 fqdn hostname
[14:25] <RoAkSoAx> smoser: and /etc/hostname is just hostname
[14:25] <rbasak> (so I've symlinked /var/lib/libvirt/images -> /home/libvirt/images but not touched the other directories)
[14:26] <smoser> RoAkSoAx, right it will always be.
[14:26] <jdstrand> rbasak: no, there is no mention of /var/lib/libvirt in /etc/apparmor.d/abstractions/libvirt-qemu. that is because the libvirt driver dynamically adds the monitor for that machine to the dynamic profile
[14:26] <smoser> so, i think cloud-inti could do a better job of falling back and not requiring network.
[14:27] <RoAkSoAx> smoser: right, yes, I believe we should still go ahead with your suggestion on doing it throuhg ensemble/cobbler for cloud-init
[14:27] <RoAkSoAx> smoser: however, we also need to make sure that cloud-init does not change what the kernel parameter is specifying
[14:27] <iggi_> Jeeves_, I think you are right, looks like the link is bouncing, must be a bad cable.
[14:28] <smoser> RoAkSoAx, i just don't like "kernel parameter is specifying"
[14:28] <smoser> :)
[14:28] <smoser> it makes it sound like there is a setting somewhere explicitly for fqdn
[14:28] <smoser> when there is not.
[14:28] <RoAkSoAx> smoser: there's a jhostname and a domain settings though
[14:29] <smoser> what do you mean by domain settings ?
[14:30] <RoAkSoAx> smoser append initrd=/images/oneiric-3-i386/initrd.gz  locale=  locale=en_US text priority=critical  auto url=http://192.168.123.2/cblr/svc/op/ks/system/node03 hostname=node03 domain=domain.com suite=oneiric
[14:30] <smoser> hm.. i didn't realize that 'domain' would be specific. i thought you were just appending to hostname.
[14:30] <smoser> http://paste.ubuntu.com/689237/ is what i think is reasonable to read /etc/hosts as a fallback.
[14:30] <RoAkSoAx> smoser: nope, that's why I was saying, cloud-init needs to respect that
[14:31] <smoser> well, its not *that* is what i disagree with
[14:31] <rbasak> Now if only I could get the BIOS to PXE boot. It seems to be ignoring the boot options :-/
[14:31] <smoser> cloud-init doesn't ever see *that*
[14:31] <RoAkSoAx> smoser: I agree, but yet again, it should not change it
[14:31] <smoser> in this specific situation, it would be ok to call 'hostname -f'
[14:32] <smoser> but what i don't like about that is the randomness. so i think i  just parse /etc/hosts to avoid hostname doing network operations if it would
[14:33] <RoAkSoAx> smoser: right, but that's why I'm saying, we can still use your approach to ensure that things are set correctly, but cloud-init should still respect what's passed to the kernel if no domain option has been passed to cloud-init
[14:34] <Daviey> rbasak: failing that, you could use ipxe on non-arm hardware
[14:34] <rbasak> Daviey: I'm starting off trying to get cobbler to build me an i386 server inside kvm
[14:34] <rbasak> Daviey: (so I can find my way around cobbler and know it works)
[14:36] <Daviey> rbasak: sounds wise
[14:37] <Daviey> rbasak: RoAkSoAx is the king for that
[14:37] <Daviey> smoser might have done it aswell.
[14:38] <Daviey> and smoser probably did it without using sudo.
[14:38] <smoser> well, i did it with being in libvirtd
[14:38] <smoser> and libvirtd group == root
[14:38] <smoser> so, it doesn't count.
[14:39] <smoser> rbasak, if you just want to play with cobbler....
[14:39]  * RoAkSoAx will brb
[14:39] <smoser> cobbler-devenv is reasonable as a place to look at a working setup.
[14:39] <rbasak> I seem to have broken it beyond where I was now. "Error starting domain: operation failed: failed to retrieve chardev info in qemu with 'info chardev'"
[14:39] <rbasak> smoser: I added myself to libvirtd
[14:39] <smoser> rbasak, that is coming from libvirt ? that error ?
[14:40] <rbasak> smoser: Yes. I don't think I changed anything. I've only been fiddling with the BIOS boot options in the guest
[14:40]  * rbasak has tried restarting libvirt-bin
[14:41] <smoser> rbasak, i
[14:41] <smoser> i'm not sure what is going wrong. is libvirt actually trying to start qemu/kvm ?
[14:41] <smoser> you should get a kvm.log somewhere from libvirt that might have some info
[14:43] <rbasak> aha
[14:43] <rbasak> kvm: -device virtio-net-pci,netdev=hostnet0,id=net0,mac=52:54:00:bc:56:31,bus=pci.0,addr=0x3: pci_add_option_rom: failed to find romfile "pxe-virtio.bin"
[14:43] <rbasak> So I did actually persuade it to pxe boot :)
[14:43] <jamespage> Daviey: I think this is just about as small as I can make it: http://paste.ubuntu.com/689248/
[14:44] <Daviey> jamespage: that is ldd of cobbler-enrol ?
[14:44]  * rbasak installs kvm-pxe
[14:44] <Daviey> rbasak: nooooooo
[14:44] <Daviey> rbasak: use ipxe
[14:44]  * rbasak purges kvm-pxe!
[14:44] <jamespage> Daviey: nope - thats libcurl
[14:45] <Daviey> jamespage: ok, rocking - thanks!
[14:45] <jamespage> so we would need udeb's for curl,xmlrpc-c, and libidn
[14:45] <rbasak> Actually that didn't help anyway, same problem. I suspect that a missing rom might be why it wasn't doing the booting but I think my info chardev error is a separate issue
[14:45]  * rbasak investigates further
[14:45] <jamespage> feels really messy tho.
[14:47] <Daviey> jamespage: if it compiles, shit it - is my mantra.
[14:47] <Daviey> erm, ship it.
[14:47] <Daviey> jamespage: really odd that we /need/ libidn.. :/
[14:48] <jamespage> difficult bit is how to make xmlrpc-c use the right version of libcurl - i.e. minimal
[14:48] <jamespage> otherwise we disable xmlrpc over a load of other protocols
[14:48] <jamespage> bah
[14:49] <Daviey> jamespage: libcurl isn't installed by default.. so if there is a new binary package, there is no impact.
[14:49] <jamespage> not sure I understand - or maybe phrased my concern correctly
[14:50] <Daviey> jamespage: Hmm. I understood your concern to be, if both variants, fat and thin are installed concurrently?
[14:50] <Daviey> that can't (or mustn't) happen
[14:51] <Daviey> Conflicts :)
[14:52] <jamespage> Daviey: lemme just gather my thoughts
[14:53] <caribou> This is not specific to the Server version, but has more impact on Server : Does someone knows why makedumpfile has become dynamically linked b/w Maverick and Natty ?
[14:53] <caribou> This renders kdump totally useless in Natty & Oneiric
[14:57] <jamespage> Daviey: so xmlrpc-c Build-Deps on libcurl3-openssl-dev
[14:59] <jamespage> We are proposing a libcurl3-minimal-dev + others; but thats pretty ugly for xmlrpc-c as switching the dependency will reduce what it can do?
[14:59] <Daviey> jamespage: Ah
[14:59] <Daviey> i see what you mean
[14:59] <jamespage> well at least I think thats the case
[15:00] <jamespage> I guess we could have another source package to build xmlrpc-c and hold two in the archive -one to provide the udeb
[15:00] <jamespage> and one as its
[15:00] <jamespage> /its/is/
[15:00] <jamespage> or we could do the udebs for the normal version of curl
[15:03] <Daviey> udebs for the nomral version oulls in ldap and kerbos, no?
[15:06] <jamespage> yes
[15:10] <robbiew> Daviey: yo!  what's up with the ARM netbook for rbasak?
[15:11] <Daviey> robbiew: I posted him the pandaboard, as that seemed more appropriate.
[15:11] <Daviey> robbiew: I thought we did dicuss this?
[15:11] <robbiew> we did...I though both was being sent...no big deal...I just knew you used yours for a picture frame
[15:11] <robbiew> or something
[15:11] <robbiew> lol
[15:12] <Daviey> robbiew: hah.. no, that isn't true.
[15:13] <Daviey> The netbook was more concerning to me, as i know it has my ssh and gpg key on it.. and i wasn't entirely sure just a rm was enough to stop rbasak retrieving it.
[15:13] <Daviey> :)
[15:13] <rbasak> :-P
[15:25] <smoser> RoAkSoAx, ok. so i think, this time for sure , that i have a good solution.
[15:25] <smoser> could you please try: http://paste.ubuntu.com/689276/
[15:25] <smoser> just run that on your installed node
[15:26] <smoser> and it should basically give the same thing as 'hostname -f' but without consulting dns
[15:26] <jamespage> Daviey: think I can get rig of IDN as well
[15:27] <jamespage> rig/rid
[15:27] <jamespage> typing bad today
[15:27] <Daviey> jamespage: Oh goody
[15:27] <jamespage> Daviey: still think we will need a xmlrpc-minimal source package tho
[15:29] <Daviey> jamespage: I think you might be right :/
[15:29] <lynxman> smoser: ping
[15:29] <smoser> lynxman, here.
[15:34] <RoAkSoAx> smoser: conference call at the moment
[15:35] <RoAkSoAx> smoser: where will that go?
[15:35] <smoser> into cloud-init. then it will fall back to using that if no local-hosname
[15:35] <RoAkSoAx> ok cool
[15:55] <TheEvilPhoenix> can anyone recommend some kind of web-based administration tool for an ubuntu-server?
[15:55] <TheEvilPhoenix> something like webmin but not
[15:55] <TheEvilPhoenix> because webmin is evil
[15:59] <jmarsden> !ebox
[16:00] <jmarsden> TheEvilPhoenix: Try Zentyal... but the real answer is to learn to admin a server using SSH and a shell :)
[16:17] <TheEvilPhoenix> jmarsden:  indeed, which I know how yo do.  I still want the actual GUI part (for quick tasks for which initiating an entire SSH connection isnt worth it)
[16:23] <robbiew> SpamapS: yo...so I can defer this, right? -> https://blueprints.launchpad.net/ubuntu/+spec/server-o-postfix-enhancements
[16:24] <rbasak> How do I point kvm/libvirt at an ipxe rom? I can see /usr/share/qemu/pxe-virtio.rom but no idea how I'm supposed to add it
[16:24] <robbiew> hallyn: I'm thinking this can be punted to P with possibly tyler helping? -> https://blueprints.launchpad.net/ubuntu/+spec/server-o-ecryptfs-testsuite
[16:24] <SpamapS> robbiew: I suppose so.
[16:24]  * robbiew pulls his leg back...and punts!
[16:25]  * SpamapS admires the hangtime
[16:25] <rbasak> It does seem as if I have /usr/share/qemu/pxe-virtio.rom but qemu is looking for /usr/share/qemu/pxe-virtio.bin
[16:26] <hallyn> robbiew: sounds good to me
[16:26] <robbiew> hallyn: cool, thx
[16:27] <hallyn> rbasak: can you (a) switch to ipxe for the moment, and (b) file a bug for kvm-pxe?
[16:27] <rbasak> Aha!  "ln -s pxe-virtio.rom pxe-virtio.bin" in /usr/share/qemu fixed it
[16:27] <hallyn> rbasak: qemu recently switched
[16:27] <hallyn> rbasak: kvm-pxe didn't get updated bc it's deprecated
[16:28] <hallyn> but we should updated it of course until it gets removed (even though it's in universe) or switched to using ipxe sources
[16:28] <rbasak> hallyn: I'm using ipxe; dpkg -S /usr/share/qemu/pxe-virtio.rom reports it's owned by ipxe
[16:28] <hallyn> rbasak: oneiric?
[16:28] <hallyn> that should have been updated
[16:29] <rbasak> hallyn: yes oneiric, ipxe	1.0.0+git-2.149b50-1ubuntu1 which seems to the latest that rmadison sees
[16:29]  * hallyn looking
[16:30] <hallyn> d'oh!  I thought Daviey had pushed that debdiff for ipxe
[16:30] <hallyn> hm, he did
[16:30]  * Daviey wonders if he is in trouble
[16:30] <rbasak> Shall I leave it with you? The symlink will do me for now.
[16:31] <rbasak> (or would you like a bug?)
[16:31] <hallyn> rbasak: please do file a bug if you don't mind
[16:31] <Daviey> hallyn: What is the problem?
[16:31] <rbasak> Daviey: ipxe drops in /usr/share/qemu/pxe-virtio.rom but qemu is looking for /usr/share/qemu/pxe-virtio.bin
[16:32] <hallyn> Daviey: I think I got confused with the .bin vs .rom .  Was it only qemu 0.15 that switches?
[16:32] <utlemming> rbasak: so, I successfully booted the OMAP kernel into qemu
[16:32] <rbasak> utlemming: cool!
[16:33] <utlemming> rbasak: I'm thinking that by tomorrow or Friday, we should qemu-able images being generated daily
[16:33] <utlemming> rbasak: I have a repeatable way of generating the disk format which includes uboot, etc
[16:33] <rbasak> utlemming: excellent!
[16:33] <hallyn> Daviey: yeah, so ipxe/debian/ipxe.links needs to switch back to being .bin not .rom for all
[16:34] <utlemming> rbasak: my problem is that I was using the OMAP4 uboot instead of the OMAP3, which was causing the kernel to be loaded in the wrong memory address space
[16:34] <hallyn> unless we quickly push qemu 0.15 :)
[16:36] <Daviey> hallyn: Having smoke 0.15, i wouldn't block that - but you need to be able to handle any technical issues that might arise.
[16:36] <Daviey> As in, do you have enough capacity?
[16:36] <hallyn> Daviey: we trust that upgrades will be coherent, and only have *.bin,
[16:36] <hallyn> Daviey: or should ipxe just create both *.bin and *.rom?
[16:37] <hallyn> Daviey: well, as I've said, I was using it for weeks on my laptop with no issues, so I'm pretty confident.  But no I don't have much spare time if things go kerblooy
[16:37] <rbasak> hallyn: https://bugs.launchpad.net/ubuntu/+source/ipxe/+bug/850154
[16:38] <Daviey> hallyn: yeah, i was using it the other day - and it 'seemed' to work for me ok.. but yes, the kicker is the fallout.
[16:38] <rbasak> hallyn: you might as well symlink them during a transition period so both *.bin and *.rom work I suppose.
[16:38] <Daviey> hallyn: hmm, do we need symlinks for both
[16:38] <Daviey> ?
[16:38] <hallyn> Daviey: http://people.canonical.com/~serge/ipxe-bin.debdiff
[16:39] <hallyn> no, we only need symlinks for one or the other.
[16:39] <hallyn> rbasak: thanks
[16:39] <Daviey> smoser: Have you sponsored a package yet?
[16:40] <hallyn> he sponsored one for me :)
[16:40] <hallyn> libcgroup iirc
[16:40] <Daviey> hallyn: you should be able to do your own signing :)
[16:41] <Daviey> hallyn: did you miss off the # for the bug closure on purpose?
[16:42] <hallyn> Daviey: oh, no.  I had it as '######' until rbasak hadned his bug over, then i converted it badly :)
[16:42] <hallyn> Daviey: i'm still waiting for pkg to build for a test, just to make sure
[16:42] <Daviey> k
[16:42] <hallyn> i don't know wtf is going on with my apt-cache.  slow right now
[16:42] <hallyn> oh, maybe it got corrupted with my unscheduled poweroff
[16:44]  * rbasak pulled the power on his laptop this morning. With no battery inserted :-/
[16:45] <hallyn> all i can figure is mine got too hot
[16:46] <hallyn> Daviey: updated http://people.canonical.com/~serge/ipxe-bin.debdiff . tested, works fine
[16:46] <hallyn> Daviey: don't know what you want with 'my own signing'.  Of course I signed it, but debdiff doesn't carry that...
[16:51] <Daviey> hallyn: sorry, i mean sign and dput :)
[16:51] <Daviey> (signing a package that you are not uploading seems overkill anyway :)
[16:51] <Daviey> smoser: Fancy reviewing and sponsoring hallyn's debdiff?
[16:52] <hallyn> Daviey: you think I have upload rights to that???
[16:53] <smoser> oksurefine
[16:53] <Daviey> hallyn: no, but you /should/ have upload access.
[16:54] <Daviey> hallyn: incidently, you do have upload access to etherboot :/
[16:55] <hallyn> yes but it is obsolete :)
[16:55] <hallyn> Daviey: yes, i'll try to get on the schedule for server upload rights in the next 2-3 weeks
[17:05] <Daviey> hallyn: cool.
[17:06]  * Daviey afk's
[17:06]  * koolhead17 is singing juju juju juju !!!
[17:10] <pmatulis> black juju?
[17:11] <TheEvilPhoenix> !offtopic
[17:19] <Ursinha> TheEvilPhoenix, juju is totally ontopic :)
[17:20] <koolhead17> Ursinha: +1
[17:21]  * koolhead17 looks at TheEvilPhoenix
[17:21] <Ursinha> if you don't know, you might! :) -> juju.ubuntu.com
[17:21] <koolhead17> http://twitter.com/#!/Canonical/status/113985182066937856
[17:22] <koolhead17> hehehe Ursinha :) when did this happen? and is channel name changed too?
[17:22] <Ursinha> koolhead17, happened today, it seems :)
[17:22] <Ursinha> the channel question I don't know
[17:23] <Ursinha> koolhead17, #juju has server people in it, #ensemble is empty, so I guess it's a yes :)
[17:23] <koolhead17> Ursinha: am allready there :D
[17:24] <talntid> anyone here think it's a decent idea, to store logfiles on a seperate server? every once in a while, my logfiles fill up my filesystem....
[17:24] <talntid> I have a virtualized environment, and 20 VM's...
[17:24] <talntid> I could start a log server, and nfsmount the logdir to the log server... ?
[17:25] <Ursinha> talntid, well... I'd rsync the logs every once in a while
[17:25] <Ursinha> might be safer
[17:25] <Ursinha> write locally then copy over
[17:26] <talntid> problem is, they have the ability to fill the filesystem pretty quickly..
[17:27] <talntid> from yesterday to today, i got 6.8gb of logs.
[17:27] <pmatulis> talntid: a log server is an old concept, you configure each log source by pointing it to the server
[17:27] <talntid> on just my webserver
[17:27] <pmatulis> talntid: you don't mount anything, it's built in to the log system
[17:27] <rbasak> I seem to have debian-installer exiting immediately and stuck in a loop because init keeps restarting it. Using a preseed file via cobbler. Is this likely to be a problem with my preseed?
[17:27] <Ursinha> pmatulis, how is that?
[17:27] <Ursinha> cool
[17:27] <talntid> syslogd?
[17:28] <pmatulis> talntid: we use rsyslog for a while now
[17:28] <talntid> when an application (apache, for example) asks for a logdir.. doesn't it write directly there?
[17:29] <pmatulis> talntid: see for a starter: http://www.rsyslog.com/sending-messages-to-a-remote-syslog-server/
[17:30] <pmatulis> talntid: apache is special, it has it's own logging system. maybe you can forward its messages too, i'm not sure
[17:32] <pmatulis> talntid: rsyslog is pretty powerful for remote logging, you can drop low priority messages if there are too many to handle, for instance
[17:32] <pmatulis> talntid: can also go directly into a d/b like mysql or postgresql
[17:33] <talntid> hmmm
[17:33] <talntid> the issue is never with the system logs...
[17:33] <talntid> always with apache or asterisk
[17:35] <pmatulis> talntid: well, asterisk should be an issue.  as for apache:
[17:35] <pmatulis> http://www.google.com/search?client=ubuntu&channel=fs&q=rsyslog+remote+apache+logs&ie=utf-8&oe=utf-8
[17:35] <pmatulis> *asterisk should not be...
[17:36] <talntid> roger that
[17:40] <rbasak> No joy with the default cobbler preseed on beta 1 either :-(
[17:52] <koolhead17> rbasak: what is the issue
[17:54] <smoser> RoAkSoAx, bug 850206
[17:55] <rbasak> koolhead17: I think the issue is that I'm booting a normal kernel inside kvm, so it can't find any disks. Is there a standard way to tell cobbler that I want the virtual kernel image? I can't seem to find anything other than a generic image in what cobbler has imported into /var/www/cobbler.
[17:57] <koolhead17> rbasak: https://help.ubuntu.com/community/Cobbler
[17:57] <rbasak> koolhead17: I could be wrong, what I'm seeing is that debian-installer keeps dying and being restarted by init.
[17:57] <rbasak> koolhead17: yeah I'm using that
[17:58] <koolhead17> so you have a profile ready with you and Vm on same network with boot from network enabled?
[17:59] <rbasak> koolhead17: yes, and it pxe boots ok and goes into the installer
[17:59] <rbasak> koolhead17: it's the preseed that seems to fail after that, at this stage debian-installer goes into a restart loop
[18:00] <rbasak> (after the installer has downloaded udebs)
[18:00] <rbasak> I think it's a partman issue
[18:00] <rbasak> and /proc/partitions is empty
[18:00] <rbasak> It's in kvm so I would expect to see a /dev/vda
[18:01] <koolhead17> rbasak: read the cobbler profile parameters your passing
[18:01] <koolhead17> and let pressed do auto partitioning
[18:03] <koolhead17> or to be precise dont use pressed file for first VM see how installation is happening, if its going well then introduce preseed and play with it :)
[18:27] <RoAkSoAx> smoser: cool,.t hanks!
[18:34] <Alison_Chaiken> Morning all.    I want to spin up a Lucid AMI in EC2 and save the result of my work in EBS.   I am confused because the list of AMIs on Canonical's website does not match what I see offered at amazon.com.
[18:38] <adam_g> RoAkSoAx: if metadata is generated outside of the orchestra provider, where is it going to come from?
[18:38] <RoAkSoAx> adam_g: meta-data is not really generated outside. It actually comes from the provider
[18:39] <adam_g> RoAkSoAx: currently it does, but the plan is to gut that and only provide cobbler with the user data to parts of the ks late command, no?
[18:39] <smoser> Alison_Chaiken, can you elaborate on that ?
[18:39] <RoAkSoAx> adam_g: it is not "really" as it grabs the UUID from cobbler and just creates the late command to pre-populate it
[18:40] <RoAkSoAx> adam_g: yes, the idea is only to pass user-data in base64 and we will store that into a ksmeta variable, which will be used on a late_command
[18:40] <RoAkSoAx> adam_g: and leave cobbler to provide the meta-data itself
[18:41] <adam_g> RoAkSoAx: the uuid ends up as metadata in /var/lib/cloud/seed/nocloud-net
[18:41] <Alison_Chaiken> smoser, I want to run a build system on Lucid.    I'm comparing https://console.aws.amazon.com/ec2/home?region=us-east-1#s=Images with "ebs/ubuntu-images/ubuntu-lucid-10.04-amd64" in the search box tohttp://cloud-images.ubuntu.com/lucid/20110913/
[18:41] <RoAkSoAx> adam_g: and the meta-data has the UUID of a system as instance-id
[18:41] <adam_g> RoAkSoAx: ok, so thats my question.. where does the meta-data itself come from, if not from juju?
[18:41] <RoAkSoAx> adam_g: meta-data only contains instance-id which *is* the UUID of a *cobbler* system
[18:41] <Alison_Chaiken> I was going to use the ubuntu.com site as a guide to pick the right AMI, but none of the ubuntu.com AMIs appear to be available from EC2 at the moment.
[18:42] <RoAkSoAx> smoser: I was thinking on maybe also using a python module to obtain the meta-data, what do you think? (I'm already using an orchestra python module to do some stuff with keys I generate for rsyslog)
[18:43] <Alison_Chaiken> Filling the search box as I have lists 8 AMIs on EC2, and I can't figure out how to choose among them.
[18:43] <smw> Alison_Chaiken, cloud.ubuntu.com/ami
[18:43] <Alison_Chaiken> The owner of all is 099720109477, which is Canonical?
[18:44] <smw> Alison_Chaiken, choose the ebs ami in the region you want.
[18:44] <Alison_Chaiken> http://cloud-images.ubuntu.com/ami is 404 for me.
[18:44] <smoser> Alison_Chaiken, http://askubuntu.com/questions/53582/how-do-i-know-what-ubuntu-ami-to-launch-on-ec2/53586#53586
[18:44] <Alison_Chaiken> So smw, all 8 are the same?
[18:44] <Alison_Chaiken> Thanks smoser, I'll look.
[18:47] <smw> Alison_Chaiken, what about cloud.ubuntu.com/ami?
[18:49] <rbasak> koolhead17: yeah, I was going to try and rip out the preseed next if I can figure out how (since the whole point of cobbler is to use one, it's there by default). I need to go now but I'll carry on tomorrow - thanks for your help!
[18:51] <Alison_Chaiken> Aha, ami-63be790a is on both ubuntu.com and EC2!    Thanks you guys.
[18:51] <smw> Alison_Chaiken, is it ebs backed? (can you make a micro instance?)
[18:53] <Alison_Chaiken> smw, this is my first use of EC2, and I don't know what a micro instance is.   EBS-backed, apparently.
[18:54] <smw> Alison_Chaiken, good. Many people start out without ebs-backed instances and it does not hurt them until much later.
[18:56] <Alison_Chaiken> "ebs" is in the image name, so as a n00b, I think it must be backed.
[18:56] <smw> Alison_Chaiken, sounds good :-)
[18:56] <Alison_Chaiken> As soon as I finish my build, I'll stop the instance and make sure it's still there right away.
[18:57] <smw> Alison_Chaiken, if you have the option to stop the instance it is the right one ;-)
[18:57] <skrite> hey all
[19:00] <skrite> hey all, i am looking for a simple to use means to view data bytes as they come in on a TCP socket.
[19:02] <RoAkSoAx> adam_g: btw.. i'll be changing everything to juju as well
[19:02] <RoAkSoAx> in the orchestra side of things
[19:07] <jdstrand> Daviey: kees and I talked. please see my comment in the bug
[19:11] <MasterRoot> hello all, does anyone know if mdadm will allow a RAID5 array to be migrated to a RAID0 array, or would I need to recreate the array?
[19:13] <cloakable> MasterRoot: recreate
[19:13] <cloakable> RAID5 is nothing like RAID0 at the disk level
[19:20] <MasterRoot> cloakable: I see, no problem. Thanks for confirming it for me :-)
[19:21] <Alison_Chaiken> smw, I've just been reading FAQ about "stop" vs. "terminate" and take your point.   Thanks again for your *helpful* advice.
[19:41] <Alison_Chaiken> I'm still a bit puzzled by the security groups.    Presumably the apt-gets I need to do are TCP, but what ports do I have to have open for them to work?
[19:41] <Alison_Chaiken> The security groups configures inbound access, so maybe closing all ports but ssh won't affect apt-get, as the request will be initiated inside the VM?
[19:44] <hazmat> Alison_Chaiken, access to external systems from an ec2 host are always fine.. security groups are inbound traffic filters
[19:45] <smw> Alison_Chaiken, you can get more help with Amazon specific stuff at #ubuntu-cloud and ##aws
[19:46] <Alison_Chaiken> Right hazmat, so my instance can always download files from other places, as long as it initiates the request it's considered "outbound"?
[19:46] <Alison_Chaiken> Thanks, smw.
[19:46] <hazmat> Alison_Chaiken, yes.. assuming its external to ec2.. if its too another ec2 host, then the other host's security groups come into play
[19:50] <Alison_Chaiken> Thanks hazmat.    I'm going to launch!
[20:12] <jj995_> hi all, I'm trying to migrate my NIS master to a new server.  I was reading the section "Chaning NIS master servers" in O'Reilly's Managing NFS and NIS, but the instructions aren't working (they seem outdated).  Any suggestions?
[21:18] <cr3> out of curiosity, I'm imagining it might be useful to have postgresql (or any database for that matter) running in the cloud where the data would reside in ebs. in such a case, should mounting, creating filesystem, configuring postgresql, etc. be handled by something like cloud-init and/or ensemble?
[21:19] <cr3> smoser: ^^^
[23:05] <hltan> hey has anybody got "sudo iptables -A OUTPUT -m owner --gid-owner fn -j DROP" to work?
[23:14] <hltan> exit