[00:25] <koolhead17> hi all
[00:29] <halvors1> I experiencing a issue here. When trying to setup IPv6 dhcp ("iface eth0 inet6 dhcp"). It doesn't work, and my mtu drops to 576.
[00:32] <halvors1> Anyone knows why?
[01:19] <halvors1> Anyone knows howto setup dhcp client to request IPv6 Prefix Delegation in /etc/network/interfaces?
[03:05] <Katronix> Hi all, can someone point me in the right direction as to how to do subdomains in bind? I can't seem to locate it
[03:12] <franklinux> hello my friens, i hava a problem!, my iptables dont functions, My browser says "connecting ..." and then get "the connection has been reset"
[03:14] <Katronix> franklinux I don't think anyone is actually here :(
[03:14] <franklinux> I have a squid server, the network has configured the proxy works, but when I create iptables rules to enable direct ips and open ports, the browser says "connecting ..." load and at the end says the connection has been reset
[03:15] <bitfury> hey guys, I just bought a 4 port PCI RS232 card for an ubuntu server but I'm not getting any connectivity using minicom
[03:16] <franklinux> Katronix, you can help me???
[03:16] <nibalizer> I'm here
[03:17] <Katronix> I might be able to, put your iptables on to pastbin
[03:17] <Katronix> nibalizer do you know anything about manually adding a subdomain to bind?
[03:17] <bitfury> pastebin: http://pastebin.com/nVH6Mh1Y
[03:17] <franklinux> ok
[03:17] <nibalizer> Katronix: I can go look at my bind config if you want
[03:17] <bitfury> I have an onboard serial port which works fine (ttyS0)
[03:17] <nibalizer> this should be absurdly well documented online
[03:18] <Katronix> nibalizer if you know where that would be fine, I find a ton of stuff about adding a domain, but not subdomains
[03:18] <nibalizer> are you trying to delegate to another server?
[03:18] <Katronix> nibalizer no same server
[03:18] <nibalizer> http://www.zytrax.com/books/dns/ch9/subdomain.html
[03:18] <nibalizer> does that help?
[03:19] <mardraum> yes, $ORIGIN is what you need
[03:19] <mardraum> also, probably #bind (and some google-fu as hinted)
[03:20] <franklinux> Katronix, http://pastebin.com/aVqaQW3t
[03:20] <Katronix> nibalizer so basically the name of the subdomain followed by IN A and the IP?
[03:20] <franklinux> there are my iptables
[03:21] <franklinux> Katronix, there are my iptables: http://pastebin.com/aVqaQW3t
[03:21] <Katronix> franklinux is this your result of using iptables -L ?
[03:22] <nibalizer> bitfury: is therer linux support for your pci card?
[03:22] <nibalizer> do you have whatever driver you need for it loaded?
[03:22] <franklinux> Katronix, there are the results with iptables -L
[03:22] <franklinux> Katronix,  http://pastebin.com/ACYmGDMG
[03:24] <franklinux> Katronix, For example, I try to open a site that is under https:// and not let me, I say connecting .... load and at the end says "the connection has been reset"
[03:24] <bitfury> nibalizer: I guess it's not compatible, bought this card: http://www.geeks.com/details.asp?invtid=1001-0168-000A
[03:24] <bitfury> :(
[03:25] <Katronix> franklinux does it have an SSL on it?
[03:25] <bitfury> I blindly bought it..
[03:28] <franklinux> Katronix,
[03:28] <franklinux> Katronix, no, that page is external, but I try to open or under enable iptables and I can not, I get in the browser is "connecting ..." for a while, and finally says "the connection has been reset"
[03:29] <Katronix> you might need to tell it what to do with traffic on the web ports
[03:32] <franklinux> Katronix, How can you to go it?
[03:33] <Katronix> franklinux honestly I don't worry about it much as I'm behind a hardware firewall
[03:34] <Katronix> nibalizer do you know of something I can use on the server to tell me how much bandwidth I've used? though I doubt I would ever come close to the 10 tb they give me, would be nice to keep track of it
[03:36] <nibalizer> ifconfig shows you
[03:36] <nibalizer> RX bytes:607371618079 (607.3 GB)  TX bytes:366623558692 (366.6 GB)
[03:37] <Katronix> nibalizer would it know when to reset?
[03:37] <nibalizer> i'm not sure how smart it is about that kind of stuff
[03:37] <Katronix> k
[03:37] <nibalizer> if you could reset it, a cron to run it montly would be cool
[03:37] <Katronix> yeah
[03:37] <nibalizer> or simply a cron to take the numbers and write to a file
[03:37] <nibalizer> maybe a little script to do some math
[03:38] <Katronix> true
[03:38] <nibalizer> or maybe this is the wrong tool for the job
[03:38] <Katronix> it would depend on if ifconfig is all time, or just since last boot lol
[03:41] <nibalizer> iptraf is cool too
[04:02] <Katronix> well I seemed to have solved the subdomain issue, now just need to figure out getting apache to work with it.. but for now sleep
[04:03] <Katronix> nibalizer thanks for the help
[04:58] <esuave> how come when i recompile apache with a newer version of openssl installed it keeps showing the old version is installed? can anyone help me out please?
[10:26] <RoyK> anyone here that knows if iscsiadm and friends support using an iscsi hba and actually offloading the iscsi parts to it, rather than just using it for transport
[10:26] <RoyK> ?
[10:26] <Pupeno_W> Is there some way to specify what to do on reload in upstart? I want to send HUP to a specific process.
[10:42] <atul_> ola zul
[12:23] <koolhead17> hola zul
[12:23] <zul> hi
[12:24] <koolhead17> how is your knee sir ?
[12:25] <zul> koolhead17: fine thanks for asking
[12:26] <koolhead17> adam_g, are you awake?
[12:38] <ironm> Hello. I am not sure if it is a bug in ubuntu-server or just wrong settings on my fresh ubuntu-server 12.04 box. I use encryption for the boot drive and also an additional encrypted volume due to the entries in /etc/crypttab and /etc/fstab (see. http://paste.debian.net/181178/ ). When I open the luks volume manually with "tcryptsetup luksOpen /dev/sdb secret" he passphrase input is "hidden". During the boot the passphrase input for th
[12:38] <ironm> e additional volume is "echoed" to the display what I don't want.
[12:39] <ironm> is it a know bug or do I need to modify the setting? Thank you in advance for any hints.
[13:37] <blizzkid> Hi all, does anyone have a (link to a) good description of how the mechanism works that mounts a USB-key in /media/[USB-key-ID]? I can't seem to find any *decent* documentation about it.
[13:39] <lordievader> blizzkid: Not entirely sure, but I think udev deals with those kind of things, perhaps that is a start.
[13:40] <blizzkid> lordievader: nope, that was my initial thought.
[13:41] <lordievader> blizzkid: It also does not give you a hint to something that udev calls to do it for him?
[13:42] <blizzkid> lordievader: nope. AFAIK it used to be that way, but seems things have changed. Nothing in udev points me even slightly in the right direction.
[13:53] <halvors> Seems like opendchub is no longer avaliable in the Ubuntu repositories. Is there an alternative DC Hub i can use?
[13:59] <glosoli> Is there any easy way of having mail server, some kind of all in one DEB Package from APT ?
[14:03] <ironm> glosoli, have you checked postfix?
[14:03] <_ruben> install "Mail server" using tasksel?
[14:04] <glosoli> ironm: too complated to configure
[14:04] <glosoli> _ruben: hmm will it configure everything ?
[14:04] <glosoli> most of the things I meant
[14:04] <_ruben> glosoli: define "everything"
[14:05] <glosoli> like configure needed parts, and I will be left of only setting domains and then creating addresses and emaisl
[14:05] <_ruben> glosoli: i'd expect so, never tried it myself tho
[14:06] <ironm> glosoli, the default mail server of ubuntu is postfix
[14:07] <glosoli> ironm: which requires a lot of configuring by hand
[14:07] <ironm> glosoli, most of configuration questions have to be answered during the installation
[14:08] <ironm> glosoli, you have to know how to use text editor however .. and understand what you are doing
[14:11] <glosoli> ironm: what about smth like cyrus ?
[14:11] <ironm> more complicated than postfix ... even it is not real smtp-server
[14:11] <ScottK> sudo apt-get install dovecot-postfix amavisd-new-postfix will get you 99% of the way there, however amavisd-new-postfix has a signficant bug, so if you're on 12.04, want the version from precise-proposed.
[14:12] <glosoli> ScottK:  well I had hard time with dovecot before...
[14:13] <ScottK> dovecot-postfix will do a lot of the configuring for you.
[14:13] <ironm> glosoli, that is true .. dovecot is a bit hard to configure (due to the understanding what is there going on ... )
[14:15] <ScottK> And ironm is right, you'll need to understand what's going on.  Those packages are just to get you started with a generally working configuration.
[14:15] <ironm> ScottK, I guess that amavisd is a antivirus filter software .. right?
[14:16] <ScottK> That pulls in, in addition to itself, clamav and spamassassin and together they do A/V and spam detection.
[14:16] <ironm> thank you ScottK
[14:17] <ironm> I don't use any (as I use linux everywhere on my desktops)
[14:17] <glosoli> ok thanks..
[14:20] <glosoli> ironm: in btw, what's wrong with Cyrus
[14:20] <glosoli>  ?
[14:20] <ironm> glosoli, it is not up to date anymore
[14:21] <ironm> in my opinion ...
[14:24] <ironm> glosoli, you have two main choices 1. postfix and 2. sendmail. all other stuff is nor really that what I want.
[14:24] <glosoli> ironm:  Dunno, I had so much time wasting tryting to configure either, documentations are not up to date
[14:24] <glosoli> tutorials too
[14:25] <ironm> and sendmail is more complicated to configure
[14:25] <ironm> glosoli, yes .. the documentation is a bottleneck
[14:26] <glosoli> ironm: Can you point me to any good resources for Postfix… ?
[14:27] <ironm> glosoli, http://www.postfix.org/documentation.html
[14:27] <ironm> http://www.postfix.org/docs.html
[14:28] <ScottK> The Ubuntu Server Guide has good step by step instructions.
[14:29] <glosoli> So basically to setup something like Roundcube for working and etc, I will need Postfix, RoundCube itself, and some web interface would be cool for creating emails
[14:29] <glosoli> what could you recommend
[14:29] <glosoli>  ?
[14:33] <ironm> glosoli, in my opinion web-interface for email-client is not recommended due to security reasons. I use icedove myself (with dovecot/postfix)
[14:33] <mardraum> roundcube is php software, so you'll need postfix, dovecot (or similar), apache (or similar) and some php provider
[14:33] <glosoli> mardraum:  I have seen it's possible to set up it iwth Cyrus hm
[14:34] <mardraum> why would you
[14:34] <glosoli> Things like dovecot drives me nuts.
[14:34] <mardraum> things like?
[14:34] <ironm> glosoli, you are not alone ;) ... <glosoli> Things like dovecot drives me nuts.
[14:35] <glosoli> mardraum: like being some kidn of package meant for linux hackers to configure, spending days by days for daily user tryed to get it working withou ability to find any proper tutorial or documentation
[14:36] <mardraum> you were probably trying to setup some dumb design with vhosts and mysql and some random library a dude with a blog told you that you totally needed
[14:37] <glosoli> mardraum:  might be, the official documentations are even more mess
[14:37] <mardraum> the dovecot config file is entirely readable
[14:37] <glosoli> readable for Linux hackers as I mentioned above
[14:37] <mardraum> linux hackers?
[14:38] <glosoli> the people who are a lot of into console
[14:38] <mardraum> if you find dovecot hard, you may be in ther wrong channel, just sayin.
[14:39] <mardraum> the*
[14:40] <glosoli> ah ok, I forgot that' sorry  if some documentation is too hard to follow, it's likely user who is stupid :)
[14:40] <glosoli> Anyway thanks for help :)
[14:42] <ironm> glosoli, you have two IRC channels: #dovecot and #postfix
[14:42] <glosoli> ironm: Tryed out they "awesome" support already
[14:42] <glosoli> :)
[15:08] <pmatulis> glosoli: i'm not sure what you meant by 'daily user' but setting up a mail infrastructure for the first time will require a fair amount of study and perseverance
[15:11] <glosoli> pmatulis:  yeah, I understand that  : >
[15:19] <pmatulis> glosoli: it's a strange paradox that the technology that an end-user first latches onto (easy from their pov) is a technology that is actually one of the most challenging from an admin's pov
[15:19] <pmatulis> (technology being 'email' in this case)
[15:42] <Kingsy> does anyone know of a working ppa that has the latest version of xdebug in it ?
[15:50] <Daviey> smoser: uho.. am i the new jamespage?
[15:50] <Daviey> daviey@smithers:~$ ssh-import-lp-id davewalker
[15:50] <Daviey> WARNING: Invalid keys at [https://launchpad.net/~davewalker/+sshkeys]
[15:51] <Daviey> (lucid)
[15:51] <jamespage> Daviey, lol
[15:51] <smoser> Daviey, you'll need to file a bug on windows 8, which i presume is your new OS.
[15:51] <smoser> or, lucid sucks. we released a new LTS. you should use it.
[15:51] <Daviey> smoser: No, it's the new OSX.
[15:52] <Daviey> smoser: I demand te full 5 year support.
[15:52] <Daviey> (only updated this box from hardy a few weeks ago :)
[15:52] <Daviey> smoser: it's one of the build boxes for mythbuntu... and smithers doesn't look happy.. http://smithers.mythbuntu.org/
[15:54] <smoser> Daviey, we can SRU that to lucid if you'd like.
[15:54] <smoser> and i will again state, that i found the checking of keys to be stupid and did not write it.
[15:57] <Daviey> smoser: but what is my issue?!!??!!?!
[15:57] <tgm4883> Daviey, you have invalid keys
[15:57] <tgm4883> ;)
[15:58] <smoser> Daviey, its either the blank lines or the ^M
[15:58] <Daviey> bah
[15:58] <smoser> or the fact that you have a public key whose private key is widely distributed in there.
[15:59] <smoser> yeah, i remember ...P4WX3f0bEmkkluw== well.
[15:59] <smoser> ok, i just wanted to see if i could see you jump. thats not the case (that i know of).
[15:59] <RoyK> Daviey: perl -pe 's/\r//g' -i somefile # :-þ
[16:02] <jdstrand> adam_g: hi! do have any idea why I might be seeing this in /var/log/keystone/keystone.log (set logging to DEBUG) when using euca-describe-instances:
[16:02] <jdstrand> (eventlet.wsgi.server): 2012-07-30 10:59:16,602 DEBUG 127.0.0.1 - - [30/Jul/2012 10:59:16] "POST /v2.0/ec2tokens HTTP/1.1" 404 274 0.008214
[16:02] <zul> jamespage: what do i need to do to the MIR to get it on your list?
[16:03] <jamespage> zul, if its on this list http://reports.qa.ubuntu.com/reports/ubuntu-server/release-bugs.html
[16:03] <jamespage> I have one eye on it
[16:03] <zul> jamespage: ok how do i get it on yruo list
[16:04] <jamespage> zul, do you want me to pickup the openvswitch MIR?
[16:04] <zul> jamespage:  sure if you want
[16:04] <jamespage> zul, well I'm not sure I 'want' but I 'can'
[16:04] <jamespage> :-)
[16:04] <zul> jamespage: sure you can :)
[16:04] <jamespage> lol
[16:05] <jamespage> zul, it looks like its be assigned to jdstrand anyway - what needs doing to it?
[16:05]  * jamespage looks
[16:05] <zul> jamespage: looks like security review
[16:05] <jamespage> zul, agreed
[16:05] <jdstrand> zul, jamespage: it's in progress. if I can get openstack working on quantal I might be able to get somewhere
[16:07] <jdstrand> adam_g: keystone service-list and keystone endpoint-list look ok
[16:08] <jdstrand> adam_g: ah! I think I found it. nm
[16:08] <Daviey> jdstrand: I really think openvswitch should be considered for MIR outside of openstack.
[16:08] <Daviey> It has real life use cases not directly associated with openstack.
[16:09] <ScottK> "Has real life uses" isn't a criteria for main.
[16:09] <jdstrand> Daviey: that's fine, but quantum wants it and that is presumably the main use case (at least as described to me)
[16:09] <jdstrand> (both openvswitch and quantum are in progress)
[16:13] <Daviey> ScottK: Please feel free to inspect each of my sentences, and misinterpret them as you see fit.
[16:13] <ScottK> Thanks.
[16:17] <jdstrand> ok, I think I finally have my openstack issues worked out
[16:17]  * jdstrand crosses fingers
[17:07] <glosoli> how can I get the date of package installation
[17:07] <glosoli>  ?
[17:14] <RoyK> glosoli: I don't think that's recorded, but if you haven't done an "apt-get clean", the packages should be in /var/spool/cache/apt/archives, and the file date there should say when the package was downloaded, which may be the same as when it was installed
[17:15] <glosoli> eh
[17:15] <glosoli> it's a pity then
[17:15] <RoyK> heh - running apt-get clean regularly?
[17:16] <glosoli> dunno I don't remember might have run some weeks ago
[17:16] <glosoli> :D
[17:19] <glosoli> RoyK: ah wrong dir it was /var/cache/apt/archives
[17:21] <RoyK> glosoli: ah - sorry
[17:34] <smoser> stgraber, around ?
[17:34] <stgraber> smoser: yep
[17:35] <smoser> i'm having an issue with reoslvconf wrt overlayroot
[17:35] <smoser> /etc/resolv.conf is not getting updated
[17:37] <smoser> stgraber, could you take a look at ubuntu@ec2-107-22-60-153.compute-1.amazonaws.com and see if anything sticks out to you as broken ?
[17:37] <smoser> (you should be able to get in with stgraber@castiana)
[17:38] <stgraber> smoser: so resolvconf is working properly but nothing has populated resolvconf... (/run/resolvconf/interface/ is empty)
[17:38] <stgraber> checking if I can figure out why
[17:39] <smoser> so without the overlayroot, it works fine.
[17:41] <stgraber> smoser: I think I see the problem, you're mounting the overlay after /run was mounted, it should be the other way around
[17:41] <smoser> well that is what i thought was the issue to, but i dont know why that would be.
[17:41] <smoser> the overlay is the root filesystem
[17:41] <smoser> (set up by initramfs)
[17:42] <stgraber> according to /proc/mounts, / was mounted after /run, so if the resolvconf/interface entry was added to /run before / was mounted over it, it'd be hidden, explaining the current behaviour
[17:43] <RoyK> sounds like insanity to me...
[17:44] <stgraber> smoser: how likely am I to loose access to the instance if I kill and respawn dhclient?
[17:44] <smoser> no. magically not. and you will get the update.
[17:44] <smoser> (i tried that)
[17:44] <smoser> i think you'r eright aobut /run.
[17:45] <smoser> i'll have to muck around in the initramfs
[17:45] <stgraber> smoser: what we do in casper is setup /root as overlayfs, then use "mount --move /run/ /root/run" before the pivot_root + init call
[17:46] <smoser> which is what the initramfs ususally does
[17:47] <smoser> stgraber, thanks for pointing me in the right direction. i can likely figure it out.
[17:48] <stgraber> smoser: np. let me know if I was wrong and it needs more investigation on resolvconf's side
[17:52] <zastaph> I installed xorg, and ran sudo xstart .. does that mean x server is running?
[17:53] <zastaph> because an app that depends on it says it isnt
[17:55]  * genii-around ponders why this app which needs X is running on a server
[17:57] <zastaph> http://cutycapt.sourceforge.net/
[17:57] <zastaph> "You cannot use CutyCapt without an X server"
[17:58] <zastaph> does it even make sense to use x server on ubuntu server ?
[18:03] <genii-around> zastaph: Not usually, since the idea is it is supposed to be a box dedicated to to doing something like feeding out pages or files or data...so the client is where the gui is. And then you ssh in and administer, or use web-based control like ebox
[18:03] <genii-around> In thin-client server scenario different of course
[18:04] <zastaph> will try from desktop then
[18:14] <smoser> stgraber, your argument was that /proc/mounts eclared that /run was mounted before /, right?
[18:15] <smoser> but that is "normal"
[18:17] <stgraber> smoser: hmm, indeed, I guess /proc/mounts won't change ordering after the --move call...
[18:17] <smoser> right.
[18:18] <smoser> so i dont think that that was the issue.
[18:18] <smoser> its possible (but hard to tell) that /run is not really a tmpfs, but it is surely writable
[18:20] <stgraber> smoser: can you pastebin /proc/<pid of dhclient>/mountinfo?
[18:20] <stgraber> that should make it possible to figure out the stacking of the various fs and know exactly what /run is supposed to be at that point
[18:21] <smoser> http://paste.ubuntu.com/1119965/
[18:24] <skrite> hey all.
[18:24] <stgraber> hmm, that looks good
[18:24] <moxie_man> I have a dd-wrt router and a linux mint server which has a heavy network load. I want to dynamically throttle back the linux mint server /only/ when other computers on the network experience low throughput.
[18:24] <stgraber> smoser: can you get me access to that instance again?
[18:25] <smoser> well, that one is dead :)
[18:25] <smoser> but to another, sure.
[18:25] <moxie_man> I am wondering if there is a direction you could point me in
[18:25] <skrite> is it possible, with one sql-server, to run cluster tables (ndb) and another type of database engine (innodb or myisam) in the same sql server? i only need to cluster often-used tables
[18:26] <smoser> stgraber, ubuntu@ec2-50-16-102-212.compute-1.amazonaws.com
[18:29] <stgraber> smoser: missing my ssh key apparently?
[18:30] <smoser> try again
[18:31] <moxie_man> the keyword I was looking for was "prioritize"
[18:31] <stgraber> smoser: ok, worked now
[18:32] <stgraber> smoser: hmm, what happened to /var/log/upstart in there? :)
[18:32] <smoser> what do you mean?
[18:32] <smoser> its a new instance.
[18:33] <stgraber> yeah, I guess I'm too used to systems producing some kind of job output, so finding a system with missing or empty /var/log/upstart just looks weird
[18:33] <jdstrand> aha!
[18:34] <jdstrand> Daviey, adam_g: so, if I do 'keystone ec2-credentials-create' and then 'keystone ec2-credentials-list', I can see my new credential. if I reboot then do 'keystone ec2-credentials-list', then it is gone
[18:37] <smoser> stgraber, what is it that actually /etc/resolv.conf . is it /etc/resolvconf/update.d/libc ?
[18:37]  * jdstrand files a bug
[18:38] <Daviey> jdstrand: we don't do any reboot testing really :/
[18:38] <zul> but we do test :)
[18:39] <stgraber> smoser: yeah, in theory you dhclient should trigger /etc/dhcp/dhclient-enter-hooks.d/resolvconf that creates a file in /run/resolvconf/interface and triggers resolvconf which will then call /etc/resolvconf/update.d/libc to generate /run/resolvconf/resolv.conf
[18:40] <jdstrand> well, at least I know what the problem is
[18:40] <stgraber> smoser: in a regular instance, do you usually get dhclient to log to syslog? (just noticed it didn't in the overlay one)
[18:40] <jdstrand> now I can work around it be creating them each time
[18:41] <adam_g> jdstrand: that may not be a bug, ec2-credentials may be meant to be epehemeral
[18:41] <adam_g> im not sure, tbh
[18:41] <jdstrand> adam_g: they persisted on precise
[18:41]  * jdstrand just verified that
[18:42] <smoser> stgraber, checking
[18:42] <jdstrand> but, I'll see if I can find a config option
[18:43] <smoser> stgraber, http://paste.ubuntu.com/1120012/
[18:44] <smoser> thats a fresh pure instance
[18:44] <smoser> /var/log/syslog
[18:44] <stgraber> ok, so dhclient always fails to log to syslog in the cloud instances, not specific to overlay
[18:44] <jdstrand> this kinda implies they are persistent as well: "Migration of users, projects (aka tenants), roles and EC2 credentials is supported for the Essex release of Nova. To migrate your auth data from Nova, use the following steps:..."
[18:46] <adam_g> jdstrand: what is your ec2 driver set to in keystone.conf?
[18:46] <jdstrand> keystone.contrib.ec2.backends.kvs.Ec2
[18:46] <jdstrand> (whatever the default is)
[18:46] <adam_g> jdstrand: set that to ...sql.Ec2
[18:46] <jdstrand> (this is not an upgrade-- I abandoned that idea and this is a fresh nistall with just 12.10 packages)
[18:47] <adam_g> jdstrand: kvs = in-memory key value store, totally lost on restart, not sure why it exists
[18:47] <jdstrand> adam_g: oh wow-- or that it is the default! :)
[18:48] <adam_g> jdstrand: yup
[18:48] <jdstrand> adam_g: that was it exactly. on reboot, they are still there
[18:49] <adam_g> jdstrand: i believe we carried a patch on precise to set all backends to sql by default. we'll do the same this time around, im sure
[18:49] <Daviey> adam_g: that was the only option until you added persistence, no?
[18:49] <adam_g> Daviey: for the service catalog, yea. well, the other option for persistence is a flat config file
[18:49] <jdstrand> adam_g: fyi, bug 1031012. I updated with your find
[18:50] <adam_g> jdstrand: thanks
[18:50] <jdstrand> adam_g: thank you! :)
[18:50] <stgraber> smoser: can you do a test boot with the file I put in /etc/dhcp/dhclient-enter-hooks.d/run? I used /tmp for the debug data but I'm not sure it's going to be writable whenever dhclient will get spawned but at the same time I don't quite trust /run at that point either ;)
[18:52] <smoser> stgraber, you can do whatever you want there.
[18:53] <smoser> if you want something to stick across boot, edit it in /media/root-ro, use 'sudo overlayroot-chroot vi /etc/dhcp/dhclient-enter-hooks.d/run'
[18:53] <jdstrand> adam_g: not sure what you think about this, but I set:
[18:53] <jdstrand> [logger_root]
[18:53] <jdstrand> level=DEBUG
[18:53] <jdstrand> in logging.conf
[18:53] <smoser> (i can do it if you'd like, but feel free to own that instance. i'm debugging on another)
[18:53] <jdstrand> adam_g: I found it hard to debug keystone with WARNING
[18:53] <jdstrand> adam_g: 2 cents
[18:53] <adam_g> jdstrand: agreed
[18:53] <stgraber> smoser: ok, will do that then
[18:56] <stgraber> smoser: abusing the kernel log buffer as a way of getting reliable storage ;)
[18:57] <smoser> stgraber, well, hmm..
[18:57] <smoser> maybe this is not unrelated.
[18:57] <smoser> but in the initramfs ii'm wiring to /dev/.initramfs/overlayroot.txt
[18:57] <smoser> or something
[18:57] <smoser> ie, using the devtmpfs mount for that.
[18:58] <stgraber> smoser: based on my debug, it looks like dhclient is ran in initramfs before anything is really mounted?
[18:58] <stgraber> smoser: http://paste.ubuntu.com/1120037/
[18:58] <smoser> lsinitramfs /boot/initrd.img-3.5.0-6-generic | grep dh
[18:58] <smoser> seems to imply otherwise
[19:04] <smoser> stgraber, ^
[19:06] <stgraber> yeah, not sure what's going on, but the /proc/mounts is definitely completely wrong at that point in the boot sequence
[19:08] <stgraber> adding some more debug in there, that kernel log is starting to get quite long ;)
[19:12] <smoser> stgraber, resolvconf.conf (from upstart) is expected to run before anything else, right?
[19:12] <smoser> i logged command and parameters for the files shown http://paste.ubuntu.com/1120065/
[19:13] <smoser> and thats the order they're getting called in.
[19:13] <stgraber> smoser: in most cases it'll be run before anything else, but it shouldn't be a requirement or that'd cause a race
[19:14] <stgraber> though it indeed looks like that's the reason of the failure...
[19:15] <stgraber> resolvconf: Error: /run/resolvconf/interface either does not exist or is not a directory
[19:15] <stgraber> smoser: ^
[19:18] <smoser> stgraber, so 'start on mounted MOUNTPOINT=/run' (/etc/init/resolvconf.conf) must not block
[19:19] <stgraber> smoser: yeah, that's it... adding "mkdir -p /run/resolvconf/interface" to /etc/dhcp/dhclient-enter-hooks.d/resolvconf fixes it. Not sure it's the right fix though
[19:19] <smoser> so that job doesn't block
[19:19] <smoser> and /etc/init/networking.conf is then running
[19:20] <smoser> before /etc/init/resolvconf.conf ever ran
[19:20] <r3dLunchb0x> how can I setup out put of someones home directory when they login or make it part of their prompt?
[19:21] <stgraber> smoser: only the start on *ing (starting, stopping, ...) are blocking and they're only blocking for something that depend on them, which isn't the case of networking.conf or network-interface.conf
[19:21] <smoser> um..
[19:21] <smoser> man mounted
[19:21] <smoser> that should block
[19:22] <smoser> so maybe mountall is confused by the overlayfs mounts.
[19:23] <stgraber> that or it's not networking.conf bringing the interface up
[19:23] <stgraber> "block" in upstart term means blocking the emission of the started event, so it's only blocking stuff that depends on it
[19:23] <stgraber> network-interface.conf only depends on the kernel
[19:26] <stgraber> hmm, that's a bit wrong actually, network-interface.conf can only be triggered once upstart-udev-bridge is started
[19:26] <stgraber> and upstart-udev-bridge is "start on starting udev" and udev is "start on virtual-filesystems"
[19:27] <smoser> stgraber, well, networking.conf is start on local-filesystems
[19:27] <smoser> which, per man page, cannot occur until after virtual filesystems
[19:27] <smoser> which shouldn't occur until after MOUNTED=/run
[19:27] <smoser> (or i thought that is what was guaranteeing this)
[19:28] <off_om> hi
[19:28]  * stgraber tests to know which of network-interface.conf or networking.conf is configuring eth0 (or if it makes any difference)
[19:29] <off_om> anyone  configured successfully with ddcliente joker.com? ubuntuserver 12.04
[19:29] <off_om> *
[19:34] <smoser> stgraber, i think you're right. its probably network-interface.conf that is bringing it up.
[19:35] <stgraber> smoser: apparently. I disabled /etc/init/networking.conf and I'm still getting the same behaviour
[19:36] <smoser> /sbin/resolvconf is just assuming that IFACE_DIR="${RUN_DIR}/interface" exists.
[19:36] <smoser> and the only thing thats going to make that is /etc/init/resolvconf.conf
[19:38] <stgraber> smoser: can you diff "mountall -v" between an overlay and standard instance?
[19:39] <esuave> so how can i create an image of a running ubuntu server VM? i wanna create a USB restore install
[19:39] <smoser> you mean just run ?
[19:39] <smoser> stgraber, you just want hte output of those ?
[19:39] <esuave> basically just wanna have a custom build of ubuntu to deploy to any server
[19:39] <smoser> or you want output of mountall --debug
[19:40] <stgraber> smoser: comparing the output of "mountall -v" should let you know what's considered local, virtual, ... by mountall
[19:47] <smoser> stgraber, normal: http://paste.ubuntu.com/1120127/  overlayroot: http://paste.ubuntu.com/1120128/
[20:08] <smoser> stgraber, i'm pretty sure its just a race condition.
[20:09] <smoser> here is mountall --debug output from overlay: http://paste.ubuntu.com/1120170/ , normal: /tmp/mountall-debug-normal.txt
[20:09] <smoser> the idfference is that in the overlayroot case, / is mounted readwrite (i wasn't the overlayfs mount read-only).
[20:10] <smoser> and that is causing a different path through different events, exposing the resolvconf race condition.
[20:17] <stgraber> hallyn: pushed a simple bugfix to ubuntu:lxc (wrong echo call in lxc-start-ephemeral). Won't upload for something that trivial though.
[20:22] <stgraber> smoser: sounds like a bug that we should be discussing with slangasek at least. One idea would be to change network-interface.conf to include "and mounted MOUNTPOINT=/run" and then change resolvconf to include "or starting network-interface"
[20:22] <smoser> bug 1031065
[20:22] <stgraber> which should ensure that resolvconf is always started before an interface is brought up and should also make sure we won't start resolvconf before /run is mounted
[20:25] <Daviey> jdstrand: 1031063.. what CPU do you have?
[20:25] <jdstrand> kvm_intel
[20:25] <jdstrand> I am trying with nested=0 now
[20:26] <Daviey> jdstrand: Well, we nest by default.. and that is what it is trying
[20:27] <jdstrand> yes, but I tried to override that
[20:27] <jdstrand> libvirt_type=qemu is what I thought that would do
[20:28] <Daviey> jdstrand: i /think/ qemu==kvm
[20:29] <Daviey> if you don't have kvm kernel support, then it should fallback to qemu.
[20:29] <jdstrand> well, I couldn't load kvm_intel in the guest at all, even if the host has nested=1. this might end up being a won't fix or a kernel bug
[20:30] <jdstrand> 'vmx' would show up in the virsh capabilities of my openstack guest
[20:30] <jdstrand> so it was probably fooled
[20:30] <jdstrand> even though I couldn't load kvm_intel in there
[20:30] <Daviey> jdstrand: what is the host CPU model?
[20:30] <jdstrand> i7
[20:30] <Daviey> jdstrand: ahhhh.. i bet you don't have the kernel extras?
[20:31] <jdstrand> kernel extras?
[20:31] <Daviey> jdstrand:  in the instance, are you runnig -virtual or -generic?
[20:31] <Daviey> precise or quantal? ;)
[20:32] <jdstrand> oh, probably generic
[20:32] <jdstrand> yes
[20:32] <jdstrand> -generic
[20:32] <Daviey> gah, dammit.. the virtual kernel didn't include the kvm kernel support, and needed the -virtual-extra's package
[20:33] <jdstrand> precise host (-generic), quantal openstack guest (-generic) trying to start an instance
[20:34] <jdstrand> Daviey: should my openstack server being running -virtual?
[20:34] <jdstrand> (ie, it is a guest)
[20:35] <Daviey> no
[20:36] <Daviey> well, if it is.. you need the extras package to modprobe kv,
[20:36] <Daviey> kvm*
[20:45] <jdstrand> Daviey: I'm confused-- my host is precise. it is running an openstack quantal server
[20:46] <jdstrand> Daviey: I installed in the openstack quantal server linux-virtual and linux-image-extra-virtual, but it doesn't seem to do anything
[20:47] <jdstrand> ie, I only have /lib/modules/3.5.0-6-generic, nothing else
[20:47] <jdstrand> if I try to remove -generic, it tries to remove everything
[20:47] <Daviey> jdstrand: no.. sorry, perhaps i wasn't clear
[20:47] <jdstrand> I cannot load the kvm_intel module in the openstack quantal server, even though my host used nested=1
[20:47] <Daviey> -virtual is a minimal kernel with less support.
[20:47] <Daviey> -generic is ideal
[20:48] <jdstrand> ok
[20:48] <Daviey> jdstrand: So, your host machine is precise, and you are running quantal in kvm, outside of openstack.
[20:48] <jdstrand> so reloading with nested=0 and then booting the openstack quantal server resulted in it not working either
[20:48] <Daviey> In that VM, you are running nova?
[20:49] <albert23> with generic you need -extra too? (linux-image-extra-3.5.0-6-generic: /lib/modules/3.5.0-6-generic/kernel/arch/x86/kvm/kvm-intel.ko)
[20:49] <jdstrand> Daviey: yes. virtualized openstack, as per https://wiki.ubuntu.com/SecurityTeam/TestingOpenStack
[20:49] <jdstrand> albert23: I have that file, I can't load it
[20:50] <jdstrand> FATAL: Error inserting kvm_intel (/lib/modules/3.5.0-6-generic/kernel/arch/x86/kvm/kvm-intel.ko): Input/output error
[20:50] <Daviey> no, -extra is JUST for virtual to give you a fuller kernel
[20:50] <jdstrand> I see
[20:51] <jdstrand> let me double chek my virtualized precise openstack could launch an instance
[20:52] <Daviey> jdstrand: in precise, $ grep -i ^KVM_NESTED /etc/default/qemu-kvm
[20:52] <Daviey> KVM_NESTED=" nested=1"
[20:52] <Daviey> right?
[20:53] <jdstrand> yes, my precise openstack launches it fine
[20:53]  * jdstrand checks
[20:54] <jdstrand> Daviey: yes (this is a default install)
[20:54] <Daviey> hmm
[20:54] <jdstrand> I think it might be a kernel issue
[20:54] <Daviey> So to clarify, precise host.. launching precise openstack, it works as expected... launching quantal openstack, it fails?
[20:54] <jdstrand> the precise openstack can load the kvm_intel module fine
[20:55] <jdstrand> Daviey: yes. <myhost> -> <precise openstack> -> instance = instance runs
[20:55] <jdstrand> Daviey: yes. <myhost> -> <quantal openstack> -> instance = instance fails
[20:55] <jdstrand> s/yes.//
[20:56] <jdstrand> <precise openstack> has kvm_intel loaded
[20:56] <jdstrand> <quantal openstack> does not
[20:56] <Daviey> jdstrand: suck.  This isn't something that is currently tested...
[20:56] <jdstrand> (and I can't)
[20:56] <jdstrand> this sounds like maybe hallyn's area-- I know he tests qemu quite a bit and thought it was with nested virtualization
[20:56] <Daviey> So this isn't really an openstack issue as such.. (although, ungracefully failing is).. but a direct kvm/kernel issue?
[20:57] <jdstrand> Daviey: that is what I am thinking now
[20:57] <jdstrand> I thought 'qemu' meant, you know qemu
[20:57] <jdstrand> :)
[20:58] <jdstrand> Daviey: but there is likely another bug there where openstack doesn't fallback to qemu when the cpu doesn't support it, but I understand that is not a priority
[20:58] <Daviey> jdstrand: Well, you can blame me for exposing it by default.. https://launchpad.net/ubuntu/+source/qemu-kvm/1.0+noroms-0ubuntu7 ;)
[20:59] <jdstrand> Daviey: it works beautifully on precise :)
[20:59] <Daviey> jdstrand: well, so.. lets look at this another way.. If i start a quantal kvm instance on precise, i should be able to modprobe kvm.. but currently failing?
[21:00] <aarcane> I'm trying to decide between ssmtp and nullmailer for my LAMP server, and I remember the primary difference is that one works in offline mode by storing messages to send later, and the other works only when the target SMTP server is readily available, but I don't remember which is which.  Does anyone know?
[21:00] <jdstrand> Daviey: kvm*_intel*. kvm loaded fine
[21:01] <Daviey> jdstrand: right
[21:03] <Daviey> jdstrand: amd64 all the way, right?
[21:04] <jdstrand> the host and guest are both amd64 installs, yes
[21:05] <Daviey> huh
[21:05] <Daviey> ubuntu@server-378:~$ sudo apt-get install qemu-kvm ; sudo modprobe kvm-intel
[21:05] <Daviey> FATAL: Module kvm_intel not found.
[21:05] <Daviey> confirmed.
[21:06] <jdstrand> Daviey: that is a different error than I get
[21:06] <jdstrand> here it finds the module
[21:06] <jdstrand> but can't load it
[21:06] <Daviey> jdstrand: Ah, IS pulled the rug out from under me..
[21:07] <Daviey> this is an AMD server.
[21:07] <jdstrand> Daviey: fyi, bug #1031090
[21:08] <Daviey> jdstrand: apt-get install linux-image , in guest.. doesn't give anything?
[21:09] <jdstrand> Daviey: I didn't have linux-image installed. I just installed it and it worked fine
[21:10] <jdstrand> (ie package install, not kvm_intel loading, which still fails)
[21:10] <jdstrand> dmesg is silent on the issue
[21:15] <r3dLunchb0x> trying to setup quotas for user home directories on 12.04, but not working.
[21:15] <Daviey> jdstrand: confimed, it's an intel only issue
[21:18] <jdstrand> Daviey: I guess it is good that it is confirmed, unfortunately for me, the combination of the openstack and the kernel bug means I can't launch instances :\
[21:18] <jdstrand> Daviey: how important is nested virtualization to you guys?
[21:19]  * jdstrand is trying to figure out if he is on his own here...
[21:19] <Daviey> jdstrand: well, it really makes life good.
[21:19] <Daviey> but not vital
[21:19] <Daviey> jdstrand: I've been using AMD as of late, so not noticed it.
[21:19]  * jdstrand nods
[21:19] <Daviey> canonistack region 2 seems to be AMD, region 1 Intel
[21:25] <jdstrand> Daviey: yeah, reloading kvm_intel nested=0 and booting my quantal openstack vm, I see no kvm module loaded, but <domain type="kvm">
[21:29] <Daviey> jdstrand: okay, well.. i'm not sure if it's an openstack bug, or a libvirt bug then.
[21:29] <Daviey> hmm
[21:29] <jdstrand> I agree (I don't know either)
[21:29] <Daviey> with nested on the host OFF.. it should fall back from kvm to qemu under the scenes.. does it not?
[21:30] <Daviey> ie, it should start an unaccelerated guest.
[21:30] <jdstrand> that is what I am testing and didn't think it did-- but I might have messed something up. hold on
[21:30] <Daviey> s/guest/instance
[21:31] <jdstrand> ok <domain type="kvm"> is confirmed
[21:31] <jdstrand> but no guest. let me see why
[21:32] <jdstrand> 2012-07-30 16:31:13 TRACE nova.compute.manager [instance: f90f97cc-3d75-48d0-9eab-b2aa2aae8ee2] libvirtError: internal error no supported architecture for os type 'hvm'
[21:32] <jdstrand> Daviey: it is not falling back
[21:32] <Daviey> *sigh*, the kernel team don't have a git bisect script.. :/
[21:33] <Daviey> jdstrand: Well, the libvirt xml generation was re-written for quantal.. seems this *could* be a regression.
[21:33] <Daviey> or.. still, a libvirt one.
[21:33] <Daviey> or even a qemu one
[21:33] <Daviey> Or i have NFI.
[21:34] <Daviey> :)
[21:34] <jdstrand> shoot, it could be qemu-kvm too-- depending on how it was invoked
[21:34] <jdstrand> heh
[21:34] <jdstrand> good times
[21:35] <Daviey> traditionally, we stopped the lbvirt xml generation not include the line.. allowing libvirt to fill it in, depending on what it felt best suited.
[21:36] <jdstrand> libvirtd.log is not super helpful. it clearly wants pm-utils
[21:36] <jdstrand> oh, wait
[21:37] <jdstrand> nm
 is all it has
[21:39] <jdstrand> let's see if I can force it
[21:53] <jdstrand> Daviey: I think it might be a libvirt thing
[21:54] <jdstrand> Daviey: if I copy/paste the cml from nova-compute.log I can shove it into precise's libvirt no problem
[21:54] <jdstrand> Daviey: xml
[21:55] <jdstrand> Daviey: however, if I take the same xml, I get an error on quantal's libvirt
[21:55] <jdstrand> Daviey: if I change this line:

[21:55] <jdstrand> to be:
[21:55] <jdstrand> <type arch='x86_64'>hvm</type>
[21:55] <jdstrand> Daviey: then it works on quantal
[22:19] <Daviey> jdstrand: good find!
[23:18] <JonEdney> Anyone familiar with a way to uninstall sendmail?  I installed dovecot and postfix, and sendmail is causing a problem, but apt-get remove isn't working, says no sendmail package.
[23:23] <lamont> postfix delivers a binary called "sendmail", as required by policy.  It also forces the removal of sendmail (via Conflicts: sendmail)
[23:24] <JonEdney> I see, I kinda figured it was something attached somewhere; I had to stop the service 'sendmail' to continue troubleshooting.
[23:26] <ScottK> JonEdney: sudo apt-get remove postfix will actually remove it.
[23:29] <JonEdney> ScottK - I think I need postfix; I'm learning working on my VPS.
[23:30] <ScottK> So if you think you don't need sendmail, you're probably not looking at the problem correctly.
[23:30] <ScottK> I'd suggest "sendmail causing problems" was a symptom, not a cause.
[23:31] <JonEdney> Thats what Im trying to get to the bottom of the problem; I installed ISPConfig on my VPS the other day, and the mail server is giving me access errors when trying to send mail, someone said sendmail needed to be removed after I showed my logs, and thats where I'm at.
[23:33] <ScottK> No, your problems started at "I installed ISPConfig".
[23:33] <JonEdney> I'm hearing that also tbh - of course it was after the fact.
[23:33] <PatrickDK> you should ask the ispconfig people how to fix your issues
[23:34] <PatrickDK> as help in here, while *will fix* the issues, will probably break ispconfig
[23:34] <ScottK> These huge mega-control systems don't work well and AFAIK, most any of them that have been packaged for Debian/Ubuntu have been removed as irredeemably buggy.
[23:34] <JonEdney> Yeah thats what I'm doing also PatrickDK - this is a learning curve for me, I'm considering getting rid and starting fresh.
[23:34] <ScottK> Zentayl (or something similar) is AFAIK, the only exception.
[23:35] <PatrickDK> the biggest issue is, if something goes wrong, you have no place to go
[23:35] <ScottK> JonEdney: You'll learn a lot more and have a more secure/stable system if you do it without such a package.
[23:35] <PatrickDK> you need to learn it all anyways
[23:36] <JonEdney> Yeah, and thats my goal is to learn and continue to do so.  ISPConfig just seemed like a neat installer.
[23:36] <JonEdney> Thanks for the tid-bits guys.
[23:36] <PatrickDK> there is the joined dovecot-postfix package