[00:00] <MoleMan2> errors=remount-ro sounds like a useful thing -_- will that run on any single disk read error?
[00:01] <MoleMan2> gartral: I presume you've tried top / htop to check if you can see what is running causing the load?
[00:02] <gartral> yes, htop show 20% CPU usage and less than 15% ram usage
[00:03] <MoleMan2> sarnold: theres very little important data on it TBH, most of the stuff that matters are probably configs and scripts I could copy off
[00:03] <sarnold> MoleMan2: ah, good. :)
[00:03] <MoleMan2> the only other thing I'd need is a backup of my Quassel db for IRC, that could take a while though :p
[00:04] <MoleMan2> everything else is 'media' on an external disk which I doubt will be affected, and is re-obtainable anyway :)
[00:04] <gartral> sarnold: also, ignore that 50c, byobu isn't reporting the right temps
[00:08] <MoleMan2> aargh, having VPN access would make this so much easier
[00:08]  * MoleMan2 reboots his router, hoping that doesn't get any worse
[00:08] <MoleMan2> oh wait, I can't. I can't access it without VPN to start with :(
[00:09] <gholms> This sounds like a job for dial-up!
[00:09] <gholms> Bah, wrong channel
[00:09] <gholms> ...though I guess it's appropriate for this one as well.  >_>
[00:09] <gholms> Sorry.
[00:13] <MoleMan2> right, sarnold, I'll ask now for when I've finished copying whatever I can think of, would I be better trying to force and fsck now before I reboot, or just go for a reboot and pray it comes back up? :p
[00:14] <MoleMan2> as the server is currently as good as useless anyway. so, I'll just grab some alcohol and forget about it if it all goes wrong
[00:14] <sarnold> MoleMan2: I think reboot and pray
[00:14] <MoleMan2> okay, thanks :)
[00:15] <MoleMan2> I'm SO glad I'm the type of person that just goes for 'whatever's going to happen will happen, why bother worrying or stressing' :P
[00:16] <sarnold> that helps :)
[00:21] <MoleMan2> oh, thats good. I can't connect to PostGreSQL, backing that up could be fun then
[00:22] <MoleMan2> screw it,
[00:22]  * MoleMan2 reboots
[00:23]  * MoleMan2 goes to get alcohol, and wait
[00:30] <MoleMan2> con...conn... connection refused
[00:31] <sarnold> refused? not timed out?
[00:31] <MoleMan2> not timed out
[00:31] <sarnold> well, okay, maybe it's still fscking.. no need to panic just yet..
[00:31] <MoleMan2> getting a response to ping, but PuTTY is giving connection refused
[00:31] <sarnold> wait, IP address, that means fsck should be done.
[00:34] <MoleMan2> I'll just sit here and pretend its all fine
[00:35] <MoleMan2> I don't seem to have rejoined IRC either
[00:42] <MoleMan2> yeah, anything I try to connect with gives me connection refused. SSH, SFTP (yes, I know they're the same) FTP, HTTP generally can't connect...
[00:42] <sarnold> damn :(
[00:44] <MoleMan2> yeah
[00:45] <MoleMan2> the annoying thing is, just about the only TimeSensitive thing I'd have been bothered about restoring is the db for IRC, as the longer thats down the more I miss from the backlog etc, but my backup is a few weeks old, and I couldn't connect to back it up earlier before I rebooted
[02:03] <avoine> smoser: hey, I wanted to know if your lxc-clone-readme.sh patch are still working with the latest lxc version in the daily build ppa?
[02:03] <uronu> how windows connect to ldap server without using samba? I tried pgina I can login with my ldap account but can't go through to the desktop(BSOD or Hang Up). Is there a correct configuration of pgina?
[02:04] <avoine> because I've tried it and it looks like the extra arguments like --userdata are not pass to the ubuntu-cloud-prep hook
[03:33] <avoine> smoser: nevermind I found my bug
[05:24] <toabctl> zul: filled a bug report for the getenv/getent typo in debian/cinder-common.postinst (see lp:1224275)
[07:07] <tibor> hello everyone. I just installed ubuntu 12.04 server on cloud service and wanted to make a JEOS installation. However the boot selection doesnt offer minimal install option as mentioned on ubuntu docs for JEOS. is that normal or do you suspect something wrong with the ISO?
[07:08] <jpds> tibor: JEOS just been gone for a while.
[07:08] <jpds> tibor: Usually an image on a cloud service will be as miniminal as it gets.
[07:10] <tibor> ok great, thanks for the info
[07:28] <tibor> does anyone have experience with debconf?
[07:28] <tibor> im trying to do following             sudo debconf-set-selections <<< 'backup-manager	backup-manager upload-dir string /var/archives/uploads'
[07:29] <tibor> it sets the settings
[07:29] <tibor> but at apt-get install  it doesnt respect the value
[08:36] <foo357> Hello, I have a ubuntu server which is having some troubles due to having a full boot sector. Doing updates with apt-get can't be done.
[08:37] <tibor_> yes, would have to clean unused packages from boot sector
[08:38] <tibor_> http://ubuntuforums.org/showthread.php?t=1435818
[08:38] <tibor_> last line
[08:39] <tibor_> or here the first answer http://askubuntu.com/questions/2793/how-do-i-to-remove-or-hide-old-kernel-versions-to-clean-up-the-boot-menu
[08:40] <RockinRoel> I've been told on #ubuntu-bugs that this would be a good place to find a sponsor to merge the fix for this bug? https://bugs.launchpad.net/ubuntu/precise/+source/openldap/+bug/1216650
[08:47] <foo357> I've got these linux image headers installed: http://pastebin.com/kd445GZj
[08:47] <foo357> How can I find out which ones are used?
[08:50] <tibor> uname -r
[08:51] <tibor> or you do it automatic
[08:51] <tibor> dpkg --get-selections|grep 'linux-image*'|awk '{print $1}'|egrep -v "linux-image-$(uname -r)|linux-image-generic" |while read n;do apt-get -y remove $n;done
[08:56] <foo357> tibor: ok, if I try to remove them through apt-get it complains about this: http://justpaste.it/ck4c should I just force it to continue?
[08:58] <tibor> yes go with apt-get -f install
[08:59] <tibor> ah but you probably cant run it because of lack of space
[09:00] <foo357> tibor: yes it fails, and adding --force-yes has no effect
[09:01] <foo357> tibor: can I somehow make it ignore those packages? "linux-server : Depends: linux-headers-server (= 3.2.0.43.51) but 3.2.0.53.63 is to be installed"
[09:06] <foo357> tibor: looks like the only option is to do some dirty manual removal
[09:06] <tibor> no you must remove it
[09:07] <tibor> did you try other suggestions from the links i posted?
[09:07] <foo357> tibor: maybe I can move the files of some old image header manually to free up enough space so that apt-get can do it's job sucessfully
[09:07] <tibor> both is fine
[09:07] <tibor> but manually removal doesnt change the grub settings and it will crash on reboot
[09:12] <foo357> tibor: all suggestions I see are using apt-get to remove stuff, but apt-get can't do anthing.
[09:13] <tibor> try apt-get autoremove (cleans up unused packages)
[09:15] <tibor> or u use dpkg to uninstall unused images (recommended, manual delete doesnt have all required steps) sudo dpkg --purge linux-image-2.6.28-11-server
[09:17] <tibor> or use aptitude: sudo aptitude purge linux-image-2.6.32-27-generic-pae
[09:20] <foo357> tibor: ok I simply used dpkg instead
[09:24] <foo357> tibor: ok /boot is freed up some now. But I'm still having an issue with that server package.
[09:28] <foo357> tibor: http://pastebin.com/x18g7WF1
[09:34] <tibor> try                               apt-get install --reinstall linux-server
[09:35] <tibor> do you have PAE package (32bit with >4gb ram) or 64 bit server? or?
[09:42] <foo357> tibor: no it fails, giving the same error
[09:45] <foo357> tibor: PAE package?
[09:51] <foo357> tibor: this is a 64 bit server so I don't think PAE is installed.
[09:53] <tibor> good
[09:53] <tibor> shouldnt be
[09:58] <foo357> tibor: how can I fix the problem with 'linux-server' and 'linux-headers-server'?
[09:58] <tibor> cant you reinstall both?
[09:59] <tibor> hmm
[09:59] <tibor> or use dpkg to reinstall it?
[10:02] <foo357> tibor: reinstalling both with apt-get fails.
[10:03] <foo357> tibor: how should I use dpkg?
[10:30] <tibor> dpkg -i packagename
[10:46] <siberiannerd> should i also write a bug report if expert install from ubuntu-12.04.3-server-amd64.iso was finished without warnings and after reboot i end up with grub's "error: file not found." ?
[10:47] <siberiannerd> that's somewhat obvious after fast checks that grub is broken
[10:47] <siberiannerd> same for alternate cd
[10:53] <rbasak> smoser: a thought. It would be really useful to me if /etc/issue listed eth0's address as standard on cloud images. Achieving this would involve overriding /etc/issue provided by base-files though. Do you think this would be useful to others?
[12:18] <smoser> rbasak, /etc/issue does not. but motd does, doesn't it?
[12:19] <rbasak> smoser: I think so. The annoying thing is that I need to log in to see the motd, and I often only have ssh access with cloud images. So it's an extra iteration to backdoor the image.
[12:20] <rbasak> smoser: btw, I have a new ubuntu-cloud-utils in my PPA awaiting build. I've taken care of all the packaging I think I can for now, so it's pretty much ready. Still plenty to do in the actual code, though.
[12:21] <smoser> rbasak. cloud-init writes network info to console
[12:22] <rbasak> smoser: remaining fixes to other packages needed: cloud-localds needs to be available in precise somehow, along with simplestreams of course. genisoimage needs to be depended on for cloud-localds (if that's not already done in upstream/saucy). simplestreams needs the gpg key packaged and working by default (not sure if you've done that already). THat's it I think.
[12:22] <rbasak> smoser: I can't necessarily always see the console. In my case, I can get VNC to the screen though.
[12:23] <smoser> you can tll cloud-init to write that to the screen. :)
[12:23] <rbasak> The TTY login clears it from the screen.
[12:23] <rbasak> So I end up with an instance that I would like to know the IP of, but have no way to find out.
[12:25] <rbasak> Given that: 1) cloud instances are generally ssh'd to only, with no password set; 2) this requires the IP, I think it would be useful. I can imagine a scenario where bare metal machines in a datacentre might have the same issue - an admin looks at the screen but can't log in that way because of the ssh key only requirement.
[12:25] <smoser> for 1. a cloud-instance writes its IP info to the console log.
[12:26] <smoser> for 2, i dont know.
[12:27] <smoser> i dont think it sounds like a terribly bad idea to have that information there.
[12:27] <rbasak> Yeah but the console log isn't always available. I think it makes sense to display IP information in multiple places so the admin can get it from where he needs to get it.
[12:28] <smoser> you'd hav to have some way to update /etc/issue.
[12:28] <rbasak> Yeah. That's a bit tricky.
[12:29] <rbasak> Perhaps from an if-up.d script, but it would need some coordination with base-files
[12:29] <smoser> well you could do it from the getty upstart job i think.
[12:29] <smoser> because i think thats just completely upstart managed.
[12:30] <smoser> ie, getty dies when you log out, and upstart starts it again
[12:31] <rbasak> Oh, that gives me an idea.
[12:31] <rbasak> PAM module.
[12:34] <smoser> rbasak, pam modules is how motd is updated.
[12:34] <smoser> but that requires login.
[12:35] <rbasak> I suppose another issue is that the IP could be out of date by the time it's ready. In theory.
[12:35] <rbasak> it's read
[12:35] <rbasak> Anyway, I intended that as an aside
[12:36] <rbasak> We need to sort out this cloud-utils/ubuntu-cloud-utils packaging
[12:36] <smoser> k.
[12:36] <rbasak> Do I need to create a new LP project?
[12:37] <rbasak> Or shall I leave it as a native package for now?
[12:51] <smoser> rbasak, i would do a new "upstream" for sure.
[12:52] <smoser> i know it seems like overhead.
[13:52] <Darkstar1> How can I increase the hard and soft limit for a user that doesn't have a shell?
[13:53] <Darkstar1> I tried editing the limits in limits.conf but that doesn't seem to work
[14:08] <psivaa> hallyn_: the lxc smoke tets for saucy is failing with similar issues as in bug 1207675 for the last two days. would help if you could take a look.
[14:13] <hallyn_> psivaa: trying to clear some final patches so we can merge 1.0-alpha1 into saucy.  let's wait and see if that merge resolves the issue
[14:14] <hallyn_> stgraber: setting up a test of dwight's monitor patch, hoping to send an ack in 15 ins
[14:14] <hallyn_> mins
[14:14] <psivaa> hallyn_: ack, thanks
[14:15] <stgraber> hallyn_: cool, thanks
[14:27] <JMichaelX>  i have a home server running ubuntu 12.04.3. it has 2 audio adapters, one is NVIDIA (digital), and the other is a c-media USB audio adapter. i configured alsa-base.conf to give index 0 to the USB adapter. audio is working, but i cannot adjust volume... using mpd or mocp, etc. any suggestions?
[14:37] <JMichaelX> would installing pulseaudio possibly fix this?
[15:18] <yolanda> adam_g https://code.launchpad.net/~yolanda.robla/charms/precise/gerrit/trunk/+merge/185289
[15:39] <smoser> hallyn_, trivial bug
[15:39] <smoser> https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/1224545
[15:39] <smoser> biggest change will be not installing 'gdisk' when you apt-get install lxc.
[15:41] <hallyn_> smoser: that's only for saucy?
[15:42] <smoser> yeah
[15:42] <hallyn_> stgraber: ^ when you merge 1.0-a1, can you make that change?
[15:43] <stgraber> hallyn_: sure
[15:44] <hallyn_> thanks
[15:46] <stgraber> smoser: I'll actually go with "cloud-image-utils | cloud-utils" if that's not a problem since I like having the package backportable without a change to all releases >= precise
[15:46] <nuclearbob> stgraber: if you have a chance, can you point me to any documentation you might have on running nested lxc in raring?  I've seen the post you have for precise with the custom apparmor profile, and I'm having trouble figuring how to use the one that ships in raring
[15:47] <stgraber> nuclearbob: essentially the same thing as that blog post except that you don't need to write the custom profile and can just set lxc.aa_profile = lxc-container-default-with-nesting in the config and be done
[15:47] <smoser> stgraber, yeah, that makes fine sense.
[15:49] <nuclearbob> stgraber: if I just use lxc-default-with-nesting, I get No such file or directory, and if I use the full path in /etc, I get Permission denied - failed to change apparmor profile
[15:51] <stgraber> nuclearbob: look closely at what I wrote above :)
[15:51] <nuclearbob> stgraber: thanks.  I'm bad at details sometimes :)
[16:49] <toabctl> I have a problem with pinning on precise together with the ubuntu cloud archive.
[16:49] <toabctl> see http://paste.debian.net/38979/ for my setup
[16:50] <toabctl> the problem is that single package selection does not work
[16:53] <rbasak> smoser: I'm struggling to get a cloud image to dhcp with a client-id of anything other than "ubuntu". Any tips? Do you know if I'm supposed to be able to change this by messing with userdata/metadata?
[16:53] <smoser> you can't change it.
[16:54] <smoser> this should be motivation to do it right :)
[16:54] <rbasak> What's "right"?
[16:54] <smoser> controlling a dhcp server.
[16:54] <smoser> and giving mac:ip address mappings definitively.
[16:54] <rbasak> I was hoping to be able to get the IP address out of dnsmasq.
[16:55] <smoser> the reason you can't change 'ubuntu' is that 'ubuntu' is the hostname in /etc/hostname of the un-modified image.
[16:56] <smoser> dhclient comes up in parallel with cloud-init finding the non-network datasource that you're providing it with.
[16:56] <smoser> and 'hostname' has already been called from /etc/hostname, so dhclient reads hostname() and sends it.
[16:56] <smoser> then cloud-init reads your data, changes /etc/hostname, calls hostname(new-hostname)
[16:56] <smoser> but too late.
[16:57] <smoser> the system you're designing is exactly how azure publishes its hostname :)
[16:57] <smoser> in azure, cloud-init has to ifdown and ifup after it sets the hostname, so that a dhcp request with the hostname provided by the cloud platform can then cause dns entries to be updated.
[16:58] <rbasak> Hmm.
[16:58] <rbasak> I have two routes I'm considering then.
[16:59] <smoser> ie, azure says to instance: "Your hostname should be YYY".  And then instance says to Azure (via dhcp) "My hostname is XXX".
[16:59] <smoser> and *then* after that dhcp request, 'host XXX' will resolve
[16:59] <rbasak> dnsmasq doesn't have dbus methods to query the mapping, it seems, but does have dbus signals on lease issue. So I could set up a listener and watch for that I guess.
[16:59] <rbasak> (for the dnsmasq case)
[16:59] <smoser> so if XXX != YYY, then YYY does not resolve.
[16:59] <smoser> even though the cloud platform knew that in the first place!
[17:00] <rbasak> Or I can watch console output, but that interferes with debugging/the developer a bit since I don't see a way to send console output to more than one place
[17:00] <smoser> what info does the dbus request get?
[17:00] <smoser> one other way is to use a callback / post home.
[17:00] <smoser> but then you have to run a service
[17:00] <rbasak> THe dbus signal will get a (ip, mac, hostname) triple.
[17:00] <rbasak> (s/hostname/client-id/ I guess)
[17:01] <smoser> that seems nice.
[17:01] <rbasak> It would be nicer if I could query the internal lease database rather than watch for changes.
[17:01] <smoser> yeah. agreed.
[17:01] <smoser> doesn't it keep a leases file ?
[17:01] <smoser> that you could just read?
[17:02] <smoser> do you have to be priviledged to listen ?
[17:02] <rbasak> Ah, yes. I didn't see it on first glance.
[17:02] <rbasak> It's in /var/lib/libvirt/dnsmasq/default.leases.
[17:02] <rbasak> For a first iteration, I could just hack a parse of that.
[17:02] <rbasak> I'll leave space to add other mechanisms later
[17:05] <smoser> rbasak, would you have to be privledged to lisen to the dbus signals ?
[17:05] <rbasak> smoser: I'm not sure
[17:06] <rbasak> smoser: I was just reading /usr/share/doc/dnsmasq-base/DBus-interface.gz which has no further infromation.
[17:07] <smoser> yeah
[17:09] <rbasak> smoser: every method I can think of to get the IP seems suboptimal in some way. So I'm just going to support multiple methods, with the caller specifying the method, and a default method that I might change in the future. So I'll just do a read from /var/lib/libvirt/dnsmasq/default.leases as an initial hacky method that will get us by for now.
[17:10] <smoser> seems an easy enough format.
[17:10] <rbasak> Since I can ask libvirt for the instance's eth0 MAC, it should only break if the user isn't using dnsmasq or if the dnsmasq leases format changes somehow.
[17:11] <smoser> right.
[17:11] <rbasak> EOD. Much progress today. Almost everything jju needs is implemented now I think. With this and managing volume deletions better implemented, I think it'll be generally usable.
[17:37] <mgw> Are there any firewall rules I need to be aware of with lxc+macvlan?
[17:37] <mgw> I have a rule like this at the top of the container's iptables: -A INPUT -m state --state RELATED,ESTABLISHED -j ACCEPT
[17:38] <mgw> but it's not catching established connections
[17:38] <mgw> instead it appears they're falling through to my last rule: -A INPUT -j REJECT --reject-with icmp-port-unreachable
[18:11] <adam_g> smb, around?
[19:44] <mgw> How can I load kernel modules inside an lxc container?
[19:45] <patdk-wk> you don't
[19:45] <mgw> I just discovered all the ipt_ modules are missing
[19:45] <mgw> ok, rather
[19:45] <mgw> how can I allow access?
[19:45] <patdk-wk> I don't believe you can
[19:45] <patdk-wk> load the modules on the host
[19:46] <mgw> hmm, maybe that's not my problem
[19:47] <mgw> i'm using macvlan networking
[19:47] <mgw> my —state RELATED,ESTABLISHED rule is not working
[19:47] <mgw> it works on the host though
[19:47] <mgw> but not in the container
[19:49] <mgw> patdk-wk: any idea what the problem might be?
[19:50] <mgw> I've tried both -m state and -m ctstate
[19:50] <mgw>     0     0 ACCEPT     all  --  *      *       0.0.0.0/0            0.0.0.0/0            ctstate RELATED,ESTABLISHED
[20:57] <mgw> any ideas why a rule like "iptables -m state —state RELATED,ESTABLISHED" is getting ignored inside an lxc container?
[21:05] <tom_ilsinszki_> I've installed the command line dropbox client and it works fine if I start it manually, however it is stuck on "Uploading…" if it is started on boot using the init.d script. I've followed this guide when installing: http://ubuntuservergui.com/ubuntu-server-guide/install-dropbox-ubuntu-server
[21:05] <tom_ilsinszki_> How should I find out what's preventing Dropbox from starting up properly?
[21:09]  * genii sips and ponders whether to trust a site named ubuntuservergui
[22:22] <gartral> hey all, what would cause a working, stable server's fs t suddenly go read-only?
[22:24] <gartral> http://paste.ubuntu.com/6099132/
[22:27] <zotta> Is there any reason why someone would use a perl script to fetch the exact same static html file every 30 minutes using a perl script?
[22:28] <sarnold> zotta: it'd give the html server an indication of when the client is alive and on the network..
[22:29] <zotta> Ok yes
[22:29] <zotta> but it's some stranger doing it
[22:30] <sarnold> maybe he intends to help you maintain the server? :)
[22:30] <sarnold> "hey zotta, it's me, looks like your server is down.."
[22:31] <zotta> that'd be strange
[22:35] <sarnold> hehe, yes :)
[22:56] <adam_g> roaksoax, around?
[23:49] <mgw1> ok, so i've been messing with macvlan stuff all day, I finally have something working properly, but not the way I want it. I want to host both lxc containers and kvm guests on the same host
[23:50] <mgw1> I'm working, at the moment, with a vlan interface — bond0.400
[23:51] <mgw1> I need a bridge associated with that vlan interface, for libvirt to use… but I also need to use bond0.400 as the link for my containers' macvlans
[23:52] <mgw1> however, using bond0.400 when there's a bridge generates an error (device or resource busy)
[23:52] <mgw1> and using the bridge itself as the link creates the problems with iptables I spent all day fighting with