[00:00] <MacRohard> https://launchpad.net/gpm
[00:10] <Technicus> I am investigeting . . .
[00:10] <davros-> evening
[00:10] <Technicus> Thanks for the suggestion . . .
[00:41] <Technicus> I have tried following this turorial: < http://mydebian.blogdns.org/?p=148  >, but the machine will not start and VirtualBox provides me with this message: "Failed to open a session for the virtual machine Video Surveillance System. Configuration error: Failed to get the "MAC" value (VERR_CFGM_VALUE_NOT_FOUND). Unknown error creating VM (VERR_CFGM_VALUE_NOT_FOUND).".  I do not understand how to connect to the server with SSH.  How can I do this?
[02:09] <Technicus> I have been trying connect a Ubuntu Desktop host and a Ubuntu Server VirturalBox gest on the same machine with no success.  I have been chatting on the #openssh channel and #vbox channel, I have tried the tutorial: < http://mydebian.blogdns.org/?p=148 >, and other suggestions . . . but I am obviously overlooking something very critical.  I don't know what to do at this point.
[02:10] <Technicus> How can I get this to work?
[02:13] <twb> Sorry, I don't support vbox.  Ubuntu recommends KVM and libvirt.
[02:19] <nandemonai> Technicus: Can the machines ping each other?
[02:19] <nandemonai> And openssh-server is installed on one or both?
[02:20] <Technicus> OpenSSH-Server is installed on the server.
[02:20] <twb> nandemonai: From the error he initially reported, it sounds like vbox is borking at the bridging level, irrespective of the guest OS configuration
[02:21] <nandemonai> Just saw the above post.
[02:22] <nandemonai> Technicus: That's a virtualbox issue for sure. How are the network interfaces configured in Virtualbox?
[02:23] <Technicus> I am trying to figure that out.
[02:23] <twb> There is probably a vbox channel
[02:23] <nandemonai> Technicus: Bridged, NAT, host only etc. Something isn't set right. I'd check the virtualbox forums and google your specific error.
[02:23] <Technicus> The host has an adapter called vboxnet0.
[02:23] <twb> Unless it's vbox ose it's not really our problem
[02:25] <twb> Oh, apparently there is no longer an "OSE"; there is just normal (open source) vbox, plus a proprietary Oracle "plugin"
[04:28] <crass> hallyn: I just checked up on the libvirt build and it failed becaus eo f the libnl3-dev missing dependency :(
[04:39] <hallyn> crass: gah. forgot the patch didn't touch debian/control.  new upload coming
[04:39]  * hallyn out
[04:53] <crass> hallyn: thanks, didn't expect you to be up :)
[05:00] <twb> dch; debcommit; dch -r; debcommit -r
[07:23] <linocisco> hi all
[07:24] <linocisco> how to setup/install ubuntu mail server for windows outlook clients ?
[07:30] <linocisco> hi all
[07:34] <twb> Well, you have three choices.  First, deploy a normal SMTP/IMAP setup and work out how to make outlook talk IMAP.  Second, deploy a propriatery MAPI server like Zimbra CS.  Third, deploy a webmail server and tell your users to suck it up and use a web browser instead of outlook.
[07:35] <twb> Oh, and of course fourth, just get rid of Windows.
[07:36] <linocisco> twb:  are you answering me ?
[07:36] <twb> linocisco: yes
[07:36] <jamespage> morning
[07:37] <linocisco> twb: thanks. Let me tell you have i have done. I have installed postfix. I have done dpkg-reconfigure postfix and enter settings. telnet myserver.org 25 and ehlo myserver.org is fine. what else I do ?
[07:38] <linocisco> twb: some said that I need to install dovecot and nmap to check ports.
[07:38] <twb> 25 is SMTP
[07:38] <twb> If you want them to be able to read mail (IMAP) as well as sending it (SMTP), you need an IMAP server -- dovecot is one
[07:39] <twb> IMAP leaves mail on the server; you could use POP3 instead, which downloads it to the desktop, but that is usually not what you want
[07:40] <twb> Rather: if you're an ISP you probably want POP3, if you're a corporation you probably want IMAP.
[07:41] <linocisco> twb: Let me recall theory, If I use IMAP,  all mail clients must be always online with ubuntu mail server. Otherwise, they can't read email if LAN cable or WIFI link is disconnected. Right ?
[07:42] <ersi> twb: He just wants a local server (He's been here for a while) and that server does not have internet connectivity.
[07:42] <ersi> linocisco: Well, setup both POP3 and IMAP and switch if users are being silly
[07:42] <twb> linocisco: with IMAP, the mail stays on the server.  Users can only access it when they are have access to the server.
[07:42] <linocisco> ersi: thanks eris. you always read my mind. knew half of my problems. thanks for coming
[07:43] <twb> linocisco: with POP3, they download it to their desktop or laptop, which means if their laptop explodes -- no more mail.  If they change laptop -- no more mail.
[07:43] <linocisco> twb:  can we have a settings to leave a copy of mails on server even though if we use POP3, right ?
[07:43] <ersi> linocisco: It's one of the things I'm good at, I can't always help with particular stuff - but my memory works wonders sometimes :P
[07:44] <ersi> linocisco: Yes, there is one such setting - but I wouldn't recommend using it. POP3 is a dumb protocol, and leaving copies on the server usually ends up in tears
[07:44] <ersi> like.. some silly user whining about three thousand duplicates because Outlook is a piece of.. goo.
[07:45] <twb> linocisco: you can, yes
[07:45] <twb> linocisco: the other problem is that outlook is REALLY SHIT, and AIUI only works properly if you use MAPI, which is Microsoft's mess alternative to IMAP
[07:46] <ersi> IMAP <3
[07:46] <twb> Even if the users are on Windows, I would probably try to get them onto tbird
[07:47] <linocisco> ersi: twb: I am not thinking only outlook. May Be thunderbird or any mail clients that will work on windows. because most users are always sticked to windows. But it would be nicer to work with Microsoft outlook because we already paid for office suite.
[07:47] <twb> tbird sucks pretty hard as well, but probably not as bad as outlook
[07:47] <ersi> linocisco: You can cache mail locally with IMAP, and make both Outlook shitpress and Thunderbird save copies locally in them and keep it on the server
[07:47] <ersi> Mail is hard, let's go shopping :-)
[07:47] <twb> mail is only hard because users won't learn mutt
[07:48] <twb> Sorry, let me rephrase
[07:48] <twb> mail is only hard because users
[07:48] <ersi> Amen.
[07:48] <linocisco> ersi:  you are funny. btw, let's get down to nitty gitty. other than postfix package. what else should I install ?
[07:48] <twb> First thing we do, let's kill all the users
[07:49] <ersi> linocisco: I usually use postfix with dovecot. Other than that, nothing more is needed - besides sasl-auth stuff
[07:49] <twb> ersi: ldap would be nice
[07:49] <ersi> yeah, but not absolutely necessary :)
[07:49] <twb> I also rolled out prayer, mainly because it was not PHP.  It looks OK if you tell it to use the cambridge "theme"
[07:49] <ersi> but yeah, indeed
[07:50] <twb> It's no gmail, but OTOH it actually works in w3m and lynx
[07:50] <ersi> I'm more of a roundcube kind of guy
[07:50] <twb> roundcube is PHP, so automatic fail in my book
[07:51] <twb> http://www-uxsup.csx.cam.ac.uk/~dpc22/prayer/
[07:51] <linocisco> ersi: namely please. dovecot-posfix or dovecot-common or ?? what ? please specific name please!!
[07:51] <ersi> meh, it's fancy and users love it
[07:51] <ersi> linocisco: Yes and yes
[07:51] <twb> My job is to prevent users from having what they want.
[07:52] <ersi> I'm not actually a sysadmin or have any kind of infra responsibility
[07:52] <ersi> I'm hired to break stuff :)
[07:52] <twb> I'm not allowed to set up nessusd yet :-/
[07:53] <linocisco> ersi:  yes and yes ?? only two packages (dovecot-postfix and dovecot-common ) ??
[07:53] <twb> linocisco: either just dovecot-postfix, which will try to set it up for you, or if that fails, then dovecot-imapd and postfix
[07:54] <twb> For me, dovecot-postfix did completely the wrong thing, so I just threw it out and did it right.  dovecot-postfix is only about twenty lines of config files anyway
[07:54] <linocisco> twb: so what is better? exim or courier or  .. ?
[07:55] <twb> linocisco: "dovecot-postfix" is a wrapper package.
[07:55] <ersi> linocisco: Nothing is better. It's a subject of taste.
[07:55] <twb> dovecot is fine; postfix is fine.
[07:55] <ersi> Pick one, stick with it until you're mad about it or got time to tinker
[07:55] <twb> dovecot and postfix are the ubuntu recommended choices
[07:55] <ersi> can't go terribly wrong with dovecot+postfix mix
[07:56] <ersi> I like that combo
[07:56] <linocisco> twb: thanks bro.  postifx = MTA , dovecot is for = IMAP+POP3 , right ?
[07:56] <twb> linocisco: yes
[07:57] <twb> I like postfix, especially compared to sendmail or exim.  Dovecot I pretty much just installed and then left alone, and it hasn't caused any trouble
[07:57] <twb> I'm still using procmail instead of sieve, tho
[07:57] <twb> Postfix is the LDA
[07:57] <linocisco> twb: thanks bro. I will google how to edit in dovecot to work with windows mail clients . If I failed, I would come back
[07:57] <linocisco> twb: that is another thing I dont know. what is procmail ?
[07:58] <twb> linocisco: also try #dovecot and ##windows
[07:58] <ersi> linocisco: It's a filtering software, sorta
[07:58] <twb> procmail is a thing where you can say stuff like "all mail from jerry, put in the jerry folder"
[07:58] <ersi> linocisco: Not necessary, you can look at it after setting up postfix+dovecot if you got the need :)
[07:58] <twb> Sieve is basically the same except it's built into IMAP
[07:59] <twb> Meaning that the MUA can do it, whereas ~/.procmailrc requires a shell account and knowledge of vi
[08:08] <linocisco> my second mail questions is concerning office deployment
[08:09] <linocisco> linocisco: in my remote offices, they used government ISP internet connection which allows only 8080 and 443 ports. Our enterprise mail sever use other port. so their only option is to use webmail only. I just want to setup POP3 mail clients for them. which support could ubuntu server perform to accomplish that ?
[08:10] <linocisco> [14:41] <linocisco> linocisco: in my remote offices, they used government ISP internet connection which allows only 8080 and 443 ports. Our enterprise mail sever use other port. so their only option is to use webmail only. I just want to setup POP3 mail clients for them. which support could ubuntu server perform to accomplish that ?
[08:10] <linocisco> in my remote offices, they used government ISP internet connection which allows only 8080 and 443 ports. Our enterprise mail sever use other port. so their only option is to use webmail only. I just want to setup POP3 mail clients for them. which support could ubuntu server perform to accomplish that ?
[08:12] <twb> You can't do POP3 over webmail
[08:12] <linocisco> I mean just to give them access to download emails and read offline like using pop3 clients. So they don't need to open email and wait for slow internet connection to check old mails
[08:12] <twb> OTOH you can run anything over any port, so you could just serve IMAP on port 8080
[08:13] <twb> I hear china are putting black boxes in all the cafe wifi access, too
[08:13] <lynxman> dannf: ping
[08:13] <lynxman> er sorry
[08:13] <lynxman> Daviey: ping
[08:13] <linocisco> twb: we have very slow internet with expensive charge. we are not china and worse than that.
[08:13] <twb> linocisco: where are you?
[08:14] <twb> myanmar?
[08:46] <eagles0513875> hey guys i am upgrading my server and i was just wondering in regards to dovecot + postfix all i need to do is backup the configs for each as well as my Maildir and then copy it over correct?
[10:55] <dogears> Hi!  Can I setup a DHCP server on eth0 on a laptop and have the wireless still function on dhcp
[10:56] <twb> Sure
[10:57] <twb> Unless you have a GUI and NM running, which will probably try to stop you because they're too damn clever for their own good...
[10:57] <twb> I know a bunch of people that set up their old laptops as routers or wifi APs
[10:57] <dogears> whats nm
[10:57] <twb> network manager
[10:58] <dogears> of course
[10:59] <dogears> I am trying to setup a laptop to have dhcp server to prepare new printers but need to have the laptop wireless still work.
[11:00] <dogears> I can start dhcp server but when I shut down the computer it fails to start on reboot
[11:00] <dogears> Can't figure
[11:00] <ichat> hi im looking for some 'guidelines'  to get the best performance out of  my  'server'  its running a 2.1ghz dualcore amd (64bit)  with  2x 1gb ddr2.   3x 1tb sata + 1x ide (20gb laptop drive 4200rpm so its slow).      its going to run  10.4  with  ebox installed...   my question is... how should i install...
[11:01] <ichat> should i run the 3 sata in software raid...
[11:02] <ichat> and if so.. is it wise to instal  the system /root   on it
[11:31] <jetole> Hey guys. Does anyone know which software it is that, when I log into one of my hosts, I receive an email that has a subject "*** SECURITY information for host.domain.com ***" and the email is just the hostname, the date and the logged in user?
[11:32] <jetole> Hey guys. Does anyone know which software it is that, when I log into one of my hosts, I receive an email that has a subject "*** SECURITY information for host.domain.com ***" and the email is just the hostname, the date and the logged in user?
[11:32] <jetole> oops. Sorry about posting that twice
[11:42] <trapmax> jetole: logcheck
[11:42] <trapmax> possibly
[11:45] <just-a-visitor> try man sudoers 5
[11:50] <RoyK> just-a-visitor: or man even, man 5 sudoers
[11:50] <RoyK> or just man sudoers :P
[11:50] <just-a-visitor> :)
[11:50] <diimdeep> hello, http://askubuntu.com/questions/59373/stream-a-random-audio-file-from-a-shared-folder-in-network
[11:51]  * RoyK just ordered 150 Hitachi HDS723020BLA642 drives
[11:58]  * jetole checks
[11:59] <jetole> no logcheck
[11:59] <jetole> default sudoers file
[12:01] <jetole> I'd really like to know what's sending the email. I have about 30 servers and just noticed it on these two but these servers have been around forever and I have never seen it before
[12:01] <jetole> though I may want to implement this on other systems once I know what it is
[12:03] <ersi> I'd check out the mail headers and see if it's smart enough to set some headers telling you WTF it is
[12:04] <jetole> I know on most of machines which also use ldap and sudo through ldap, it sends a email every time sudo is run but these machines are not using ldap and have a default sudoers file
[12:04] <jetole> ersi: I did. the only header I see is 'Auto-Submitted: auto-generated'
[12:04] <jetole> everything else is just standard headers
[12:04] <zul> morning
[12:05] <jetole> morning
[12:07] <jetole> ah crap. It is sudo and I just figured out why
[12:08] <jetole> the host is using an older domain which we just transferred to a third party server that no longer knows about a lot of these hosts so sudo says unable to resolve the host name
[12:08] <jetole> It's supposed to be changed to a different domain and that's why
[12:09] <jetole> thanks for the help guys. Now I gotta go hit the bank before I go into the office
[12:09] <jetole> later
[12:09] <ersi> jetole: Awesome that you found out. :-)
[12:09] <jetole> thanks to trapmax, just-a-visitor and ersi
[12:10] <jetole> +1 to just-a-visitor for giving me the right answer even when I didn't think that was it. lol
[12:10] <ersi> ^_^
[12:10]  * jetole runs to make sure I
[12:11]  * jetole runs to make sure I'm at work on time
[12:14] <just-a-visitor> :D
[12:29] <Daviey> Who wants to do a dhpython2 transition, and enable a test suite during the python build? :)
[12:30] <smoser> jamespage, or Ursinha can we get 20110831 cloud images on ec2 tests run please ?
[12:30] <zul> Daviey: ill do it
[12:30] <Daviey> smoser: Is that considered our candidate?
[12:31] <jamespage> smoser: ack
[12:31] <smoser> Daviey, unless you ahve a reason that it should not be
[12:31] <Daviey> smoser:  no, just wanted to know :)
[12:32] <Daviey> zul: bug #836677
[12:32] <zul> Daviey: funny i already knew which bug you were referring to
[12:33] <Daviey> zul: good stuff!
[12:34] <Daviey> zul: so ipy has been removed from nova in bzr?
[12:35] <zul> yep
[12:35] <rbasak> Is installing natty then do-release-upgrade the preferred way to install oneiric? I can't find any docs anywhere
[12:36] <Daviey> rbasak: That is useful as testing, at this point in the cycle.
[12:36] <zul> rbasak: easier to start off with oneiric
[12:36] <Daviey> rbasak: But you can just install from the b1 candidate iso :)
[12:36] <rbasak> OK so either :)
[12:36] <Daviey> yup!
[12:37] <Daviey> rbasak: fresh install, http://cdimage.ubuntu.com/daily-live/current/oneiric-desktop-amd64.iso
[12:39] <rbasak> Thanks Daviey. This is for a virtual machine, so is there a server ISO too somewhere? I'm still backing up my main machine and I'll probably go maverick->natty->oneiric for that!
[12:40] <Daviey> rbasak: sure.. just navigate the url above. :)
[12:40] <rbasak> Daviey: No server iso in that directory :-/
[12:40] <Daviey> rbasak: http://cdimage.ubuntu.com/ubuntu-server/daily/current/
[12:41] <rbasak> ah, ok!
[12:41]  * rbasak will hopefully stop getting lost soon :)
[12:42] <Daviey> rbasak: If you find out how, please let smoser know how.
[13:06] <zul> Daviey: stompy will need one as well
[13:08] <ivoks> SpamapS: hi; can we use 'start on stopped rc-sysinit' for upstart jobs that depend on not-yet-upstartified sysv jobs?
[13:10] <Daviey> zul: cool, can you update the status on the bug please?
[13:18] <davros-> this look like it will work? .....  http://pastebin.com/n3y6EfA1
[13:21] <zul> stompy has been taken care of as well
[13:21] <Daviey> zul: \o/
[13:49] <cr3> hi folks, anyone happen to know whether a patch for CVE-2011-3192 will be backported to Apache versions prior to 2.2.20, 2.2.19 is the latest in the repo.
[13:50] <_johnny> hi. anyone familiar with xinetd and stunnel? i've followed a few "how to"'s, but they all involve wrapping ssl around another connection, rather than binary execution - which i'm having a little trouble with. anyone tried something like this? (i have a xinetd service which runs a shell cmd, and want to convert this to a stunnel conf)
[13:51] <cr3> advanced search for that CVE under the ubuntu apache project in launchpad bugs, even those fix released, doesn't return anything
[13:53] <zul> cr3: ask the security folk
[13:53] <zul> cr3: but afaik its being worked on
[13:57] <jdstrand> cr3: sbeattie is working on updates for hardy - natty. I imagine once those are out he will either do or coordinate with the server team for oneiric
[13:58] <jdstrand> s/ do / do the patch himself /
[13:58] <lynxman> alright, who wants to have fun diagnosing a very nasty server bug with me? :D bug #818177
[14:11] <hallyn> smoser: libvirt is just TOYING WITH ME.  I've got /bin/mount wrappered to spit out some info.  Up ot the last time that /bin/mount is called during bootup, /dev/ptmx is a symlink.  But after i log in, it's not.
[14:12] <cr3> jdstrand: thanks, where can I look to keep updated about the state of the patch? might there be a bug or something I could subscribe to?
[14:13] <jdstrand> cr3: I imagine it is going out today, but sbeattie would be in a better position to say. so, either look at http://people.canonical.com/~ubuntu-security/cve/2011/CVE-2011-3192.html or www.ubuntu.com/usn (there is an feed on the latter)
[14:36] <Daviey> RoAkSoAx: I think the whole preseeds need review for cobbler TBH.
[14:36] <Daviey> (orchestra)
[14:36] <robbiew> Daviey: hey....you still using that Toshiba netbook to keep your feet warm?
[14:37] <Daviey> robbiew: Nah, i tried to see if it blended.
[14:37] <Daviey> (it doesn't)
[14:37] <robbiew> Daviey: can you send it to rbasak, then?
[14:37] <Daviey> robbiew: Yep.. not sure how useful it will be long term.
[14:38] <Daviey> rbasak: Can you let me know your address please?
[14:39] <Daviey> I don't plan to post it until tomorrow now.
[14:42] <airtonix> i think i've pretty much isntalled every package with the work xapian in it and yet haystack and xapian-haystack report that i don't have xapian installed.
[14:43] <RoAkSoAx> Daviey: I don't think there's much review to be done
[14:43] <RoAkSoAx> Daviey: it is just a simple non questions asked installation
[14:43] <RoAkSoAx> Daviey: though, minor stuff such as shostname needs to be handled
[14:49] <davros-> I hate dhcp3 so much right now
[14:51] <hggdh> SpamapS: why is it that, on RAID1 recovery I get the MDs as a mix of both disks?
[14:51] <hggdh> SpamapS: 3 mds; one has both disks active, one has disk1 active, and one has disk2 active
[14:52] <lynxman> hggdh: that sounds like a very bad md crash
[14:52] <lynxman> hggdh: it's normally either one disk or the other, mixup... eek
[14:54] <hggdh> lynxman: not really a bad crash. Just the standard ISO test ;-)
[14:54] <RoAkSoAx> lynxman: dude!!
[14:54] <RoAkSoAx> lynxman: can we do the thing tomorrow morning? I'm still fighting here with my fixes
[14:54] <lynxman> RoAkSoAx: don't worry, my machines are as dead as Zed
[14:55] <lynxman> hggdh: it still terrifies me :)
[14:55] <hggdh> lynxman: well, TBH, it also worries me... this is why I pinged Clint
[14:55] <RoAkSoAx> lynxman: hehehe ok
[15:02] <robbiew> zul: ping
[15:02] <zul> robbiew: whats up?
[15:03] <robbiew> zul: are u working on the cobbler/ARM bits?
[15:03] <robbiew> was thinking rbasak might be able to help
[15:03] <zul> robbiew: no RoAkSoAx got it working afaik
[15:03] <robbiew> RoAkSoAx: ^^?
[15:06] <smoser> hallyn, wow.
[15:08] <hallyn> smoser: all right, so i've only had two working ways to work around this.
[15:08] <hallyn> the racy one, is to have /etc/init/ssh.conf re-setup ptmx
[15:08] <hallyn> the other is to have libvirt not make ptmx a symlink, but a mount
[15:08] <smoser> ssh doing it is pointless
[15:09] <hallyn> yes, and racy
[15:09] <smoser> as its not only ssh that is affected
[15:09] <hallyn> but the first one :)
[15:09] <hallyn> ideally a systemtap taplet would show me where devpts is being mounted...
[15:09] <smoser> libvirt or lxc tools doing *anything* is really garbage to me.
[15:09] <hallyn> but i'd need linux-image-virtual-dbg which doesn't seem to exist
[15:10] <hallyn> kees ^
[15:10] <smoser> just like if my laptop bios decided it shoudl run 'ifconfig eth0 up' for my OS.
[15:10] <hallyn> smoser: i've tried having /sbin/MAKEDEV create ptmx the right way, but that wasn't enough
[15:11] <hallyn> right, ideally we find what, in our boot sequence, is creating ptmx the wrong way
[15:11] <hallyn> and fix that
[15:11] <hallyn> but failing htat, i'm goign to push a libvirt patch to do /dev/ptmx as bind mount isntead of symlink.
[15:11] <hallyn> (it's already doing symlink, so it's not like it's adding to hackiness)
[15:11] <smoser> hallyn, really, thats just going to break other things.
[15:12] <smoser> you're changing an expected environment
[15:12] <hallyn> what is going to break other things?
[15:12] <smoser> changing libvirt
[15:13] <smoser> you risk breaking other working containers
[15:13] <smoser> just to fix something you dont understand.
[15:13] <hallyn> I don't think so
[15:13] <davros-> wtf how can, address 192.168.1.10 not in subnet 192.168.1.0 netmask 255.255.255.255
[15:14] <hallyn> smoser: there are two ways to provide a correct /dev/ptmx in a container.
[15:14] <hallyn> of course i can just blame you for insisting on using libvirt-lxc
[15:15] <hggdh> davros-: well, with the netmask being all ones, it makes sense
[15:15] <smoser> you can blame ubuntu for choosing libvirt as the prefered hypervisor interface.
[15:15] <smoser> but blaming me is not going to do much good.
[15:15] <Daviey> zul: Was enabling the testsuite in python-dingus viable?
[15:15] <smoser> although i do agree with the choice to use libvirt
[15:15] <zul> Daviey: i totally missed that part
[15:15] <hallyn> smoser: it helps me ignore your annoying stmts about 'things i don't understand' :)
[15:16] <hallyn> anyway i'm off to try stap some more.  bbl
[15:16] <Daviey> hallyn: stapping is bad
[15:16] <smoser> hallyn, thats reasonable.
[15:16] <smoser> i really just dont think that changing a platform to suite an occupant is a good idea.
[15:16] <davros-> hggdh, what do u mean all 1's as in "1"92."1"68. etc
[15:16] <Daviey> smoser: I hate to do this, but i really do agree with you.
[15:17] <hggdh> davros-: netmask, not IP
[15:17] <just-a-visitor> Because netmask 255.255.255.255 or /32 specifies only 1 IP. You need at least /28 or 255.255.255.240. (In general: 255.255.255.0 will do.)
[15:17] <zul> Daviey: not really
[15:18] <Daviey> zul: uh?
[15:18] <davros-> hggdh, oops
[15:18] <davros-> thx
[15:19] <zul> Daviey: you asked me about the python-dingus testsuite being enabled and i just had a look at it and my response is not really worth it
[15:19] <jamespage> smoser, Daviey: ec2 testing completed - https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/
[15:19] <zul> Daviey: since python-carrot is going to go away soon anywyas
[15:19] <Daviey> zul: Ah.. do you want to comment on the MIR bug why?
[15:19] <zul> Daviey: sure
[15:20] <davros-> hmmm to bad it still gives out ip's that don't connect to the internet
[15:21] <zul> Daviey: commented
[15:21] <smoser> jamespage, do you have thoughts on https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/ARCH=amd64,REGION=eu-west-1,STORAGE=ebs,TEST=multi-part-ud,label=ubuntu-server-ec2-testing/lastBuild/console
[15:21] <smoser> no route to host ?
[15:21] <smoser> er... i guess it is UserInitiatedShutdown
[15:22] <smoser> ah. https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/ARCH=amd64,REGION=eu-west-1,STORAGE=ebs,TEST=multi-part-ud,label=ubuntu-server-ec2-testing/lastBuild/artifact/None/amd64/m1.large/ebs/i-ba545fcc/fd0b2556-95c8-4194-a41b-ecb77335c031-terminated.console.txt it seems to have not come back up after a stop and start.
[15:23] <smoser> oh wait. it never booted, even better.
[15:23] <jamespage> nope - it did boot
[15:24] <jamespage> phase 5/6 is restart and check then shutdown to complete test
[15:24] <smoser> you mean stop
[15:24] <smoser> stop and start
[15:24] <smoser> as opposed to reboot, right?
[15:25] <jamespage> it booted up first time OK
[15:25] <jamespage> tested OK
[15:25] <jamespage> but did not managed the stop/start that well - needed a kick as would not start
[15:25] <jamespage> it got there in the end (I think)
[15:25] <smoser> well, did it?
[15:26] <smoser> unless initramfs has a timeout before just rebooting, i dont know what would have done it.
[15:26] <jamespage> yeah it did - you can see the 'validated keys' message in the log - but it took a reboot
[15:26] <smoser> ah. so your test suite kicked it.
[15:26] <jamespage> and it took a long time to shutdown in phase 6
[15:26] <jamespage> thats the one
[15:26] <jamespage> thats the first time I've ever seen that error tho
[15:27] <smoser> is the "i'm going to reboot that instance for you" logged anywhere ?
[15:27] <smoser> yeah, its basically "can't find root". blkid /dev/xvda1 never came back.
[15:28] <smoser> but then, according to https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/ARCH=amd64,REGION=eu-west-1,STORAGE=ebs,TEST=multi-part-ud,label=ubuntu-server-ec2-testing/lastBuild/artifact/None/amd64/m1.large/ebs/i-ba545fcc/fd0b2556-95c8-4194-a41b-ecb77335c031-terminated.console.txt it never came back up after that.
[15:29] <jamespage> that might be a time issue on collecting console output
[15:29] <jamespage> INFO:root:Instance i-ba545fcc not responding after 1200 seconds - rebooting
[15:29] <jamespage> is the test framework kicking the instance
[15:30] <jamespage> INFO:root:Host ssh-dss key validated for instance i-ba545fcc (ec2-46-137-133-140.eu-west-1.compute.amazonaws.com) INFO:root:Host ssh-rsa key validated for instance i-ba545fcc (ec2-46-137-133-140.eu-west-1.compute.amazonaws.com)
[15:30] <jamespage> is it validating the signature of the server post stop/start
[15:33] <smoser> jamespage, and then https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/ARCH=i386,REGION=ap-southeast-1,STORAGE=ebs,TEST=simple-user-data,label=ubuntu-server-ec2-testing/lastBuild/artifact/None/i386/m1.small/ebs/i-d5e38880/
[15:33] <smoser> there is no terminated console ?
[15:36] <jamespage> smoser: one sec - still catching up
[15:36] <jamespage> looks like both eu-west-1 tests that failed failed in the same way?
[15:37] <smoser> i think so, other than the i386 did not get -terminated console collected
[15:37] <smoser> so we really dont know.
[15:37] <smoser> https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/ARCH=i386,REGION=eu-west-1,STORAGE=ebs,TEST=multi-part-ud,label=ubuntu-server-ec2-testing/lastBuild/artifact/None/i386/m1.small/ebs/i-de545fa8/cbfa90d3-2ab8-4b99-aa5f-36d7cba1b072-terminated.console.txt is very interesting. its the first time we've seen the url error on metadata service after i put in the 'ci-info:' information.
[15:38] <smoser> that ci-info information is printed on 'cloud-init start' job, which should only happen after eth0 is up.  but eth0 does not have a network address.
[15:39] <smoser> but it clearly didn't timeout, as 2.8 seconds or some such in
[15:40] <jamespage> did you see the kernel panic at the top of that log?
[15:40] <smoser> where?
[15:40] <jamespage> [7334662.954504] Kernel panic - not syncing: Attempted to kill init!
[15:41] <smoser> oh, yeah.
[15:41] <smoser> did the test suite kick that ?
[15:41] <jamespage> yeah
[15:41] <jamespage> the ap-southeast failure did not actually 'terminate' the machine per say
[15:42] <smoser> hmm.
[15:42] <jamespage> the entire test timed out and it scrubbed everything
[15:42] <smoser> strange.
[15:42] <Daviey> smoser: directly after the /root/dev/console bug
[15:42] <smoser> right
[15:42] <smoser> i just saw that.
[15:42] <smoser> so that is the bug.
[15:42]  * Daviey suspects race condition :/
[15:42] <smoser> and i'm happy to see that
[15:42] <smoser> as previously i'nd only seen it on kexec'd kernels.
[15:42] <Daviey> if only we could reliably trigget it :/
[15:43] <smoser> so for that log, it looks like:
[15:43] <smoser>  * bug 833783
[15:44] <smoser>  (which caused the kernel panic)
[15:44] <smoser>  * harness reboot
[15:44] <smoser>  * did not get networking, even though 'ifup eth0' finished.
[15:44] <smoser> is that right, jamespage i'm not sure about the harness kick.
[15:44] <SpamapS> ivoks: re the upstart question.. I'd do stopped rc RUNLEVEL=[2345] so you can also start when you switch from runlevel 1 -> 2
[15:45] <ivoks> SpamapS: ok, thanks
[15:45] <SpamapS> ivoks: make sure you stop on starting rc RUNLEVEL=[016] so you stop before the depended upon service.
[15:45] <jamespage> smoser: yeah - that looks right to me
[15:45] <smoser> then, *another* root/dev/console error
[15:45] <smoser> and success.
[15:46] <smoser> wow.
[15:46] <smoser> if at first you dont succeed.
[15:46] <jamespage> keep on rebooting!!!
[15:46] <jamespage> it will get there in the end!
[15:46] <SpamapS> hggdh: and re the RAID1 oddness.. that has not happened for me.. and I'm not sure how it would decide to choose the volumes that way ... at what step in the tests did you see this?
[15:48] <hggdh> SpamapS: I installed, booted with both disks, shutdown, removed first disk, rebooted degraded, shutdown, removed the second disk and readded the first, booted degraded, shutdown, added back the second, got it
[15:48] <hggdh> SpamapS: on AMD64 (KVM)
[15:49] <smoser> jamespage, is it possible to get a permalink to https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/ chart ?
[15:49] <smoser> ah. this looks like maybe https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/7/
[15:49] <jamespage> https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/7/
[15:49] <jamespage> snap
[15:49] <jamespage> thats the specific build record so should not change
[15:50] <SpamapS> hggdh: you broke your RAID badly
[15:50] <SpamapS> hggdh: I updated the test procedures late in the 10.10 cycle because they instructed you to do that.. booting with one up, then the other, corrupts the array consistently
[15:50] <SpamapS> hggdh: its an open bug that we don't have a good solution for, and can be avoided by not doing that. ;)
[15:51] <hggdh> SpamapS: ah. Then we have to update the instructions :-)
[15:51] <SpamapS> hggdh: I did!
[15:51] <hggdh> SpamapS: do you have the bug #?
[15:51] <SpamapS> bug 557429
[15:52] <SpamapS> hggdh: step 16 f could perhaps using some more clear wording.
[15:53] <hggdh> SpamapS: yes, looking at it now
[15:53] <hggdh> SpamapS: although I did not get the destructive effect on the bug
[15:54] <SpamapS> hggdh: We may need to more strongly state that this is a known bug that has no easy fix, and so that procedure should be avoided..
[15:54] <SpamapS> hggdh: you did, you just don't know it.
[15:54] <SpamapS> hggdh: dig through log files, you'll find some nastiness.
[15:54] <hggdh> will do :-)
[15:54] <SpamapS> Its *possible* that the writes were all nearly identical in alignment and placement, but given the way ext4 works, thats highly unlikely.
[15:55] <SpamapS> Have seen it corrupt stuff in /var/lib/dpkg .. /etc .. initramfs.. its just nasty
[15:59] <smoser> reported bug 838199
[16:00] <Daviey> smoser: ^^ might be a udev bug
[16:01] <smoser> i dont think so.
[16:01] <smoser> blkid blocked.
[16:01] <smoser> died on IO
[16:01] <smoser> kernel
[16:01] <smoser> you disagree?
[16:02] <smoser> if there wasnt the second failure, i would have really felt like attributing it to hardware failure on the host.
[16:04] <SpamapS> Do we have the dmesg/console output for that one?
[16:04] <hggdh> SpamapS: is http://testcases.qa.ubuntu.com/Install/ServerRAID1 better?
[16:06] <Daviey> smoser: things like http://git.kernel.org/?p=linux%2Fhotplug%2Fudev.git;a=commitdiff_plain;h=2c64f5898ccf7c2dfc330c6b21540d6080d37659 come to mind
[16:09] <smoser> Daviey, i'm confused.
[16:11] <Daviey> smoser: noted.
[16:11] <smoser> jamespage, so i think analysis was wrong for bug 838199
[16:11] <smoser> of the fialures at https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/7/
[16:11] <smoser> its the ap-southeast-1 and the top eu-west- that look alike.
[16:11] <SpamapS> hggdh: yes! thank you!
[16:12] <smoser> the bottom eu-west-1 is root/dev/console issue
[16:12] <jamespage> right
[16:18] <smoser> i see nothing at all in https://jenkins.qa.ubuntu.com/job/oneiric-server-ec2/7/ARCH=i386,REGION=ap-southeast-1,STORAGE=ebs,TEST=simple-user-data,label=ubuntu-server-ec2-testing/artifact/None/i386/m1.small/ebs/i-d5e38880/796746d6-f691-4b8c-b9c7-58c99b3605e4-stopped.console.txt
[16:18] <smoser> other than "The disk drive for / is not ready yet or not present."
[16:18] <smoser> and "The disk drive for /tmp is not ready yet or not present."
[16:18] <smoser> i'm really not sure how to handle that failure.
[16:36] <Daviey> lynxman: Have you fixed bug 818177 yet?
[16:37] <lynxman> Daviey: nope, I'm still running through it
[16:37] <lynxman> Daviey: thanks to apw and your suggestions I've reached a small impass
[16:37] <lynxman> Daviey: Looks like the culprit is mountall, it sits forever trying to remount the local filesystems, whereas remounting manually works
[16:38] <lynxman> Daviey: now the question is, how do I know how init is handling the local fs, and if so in upstart systems how can I modify its parameters to be more verbose
[16:38] <Daviey> lynxman: can you strace mountall ?
[16:39] <lynxman> Daviey: doing that right now
[16:40] <Daviey> lynxman: it might be an issue that foundations might be better to help with TBH.
[16:40] <lynxman> Daviey: I'll move it to ubuntu-devel then
[16:40]  * lynxman jumps again
[17:05] <foods> has anyone on the server team pushed anything to openstack/glance via the new gerrit?
[17:05] <foods> lol
[17:06] <adam_g> has anyone on the server team pushed anything to openstack/glance via the new gerrit?
[17:06] <RoAkSoAx> foooooooods..... I mean adam_g :) :P
[17:07] <zul> adam_g: nop
[17:08] <zul> adam_g: is it broken somehow?
[17:10] <adam_g> zul: it seems to be having trouble with my launchpad acct / openid URL and not letting me get anything into gerrit. i was wondering if someone else might be willing to push this  patch to get bug #828719 moving
[17:10] <Daviey> adam_g: Ask Monty to delete your gerrit account?
[17:13] <adam_g> Daviey: it seems lp is giving two different openid URLs for my account. im waiting to get a duplicate accounts merged into mine so the two urls hopefully do not exist
[17:15] <bluethundr> hello I am trying to update php on a lucid lynx server, but I am getting a gpg failure http://pastie.org/2460990
[17:15] <bluethundr> under centos I wqould just issue a yum --no-gpgcheck but is there a similar ubuntu oriented command? thanks
[17:16] <Daviey> adam_g: that is ..odd..
[17:16] <Daviey> two id's shoudn't cause issue :/
[17:17] <adam_g> Daviey: it might if gerrit is getting one by resolving your lp ID at login, and another by resolving your name at commit
[17:18] <Daviey> adam_g: Ahhh!
[17:21] <philsf> does anybody know if there's a webmail client that shows the threads in linear mode, a la gmail?
[17:21] <Pici> 'webmail client'?
[17:22] <philsf> Pici, webmail, imap client
[17:22] <philsf> (mistyped)
[17:22] <philsf> meaning, standalone imap client, NOT suid like openwebmail.
[17:24] <Technicus> Hello . . . how do I configure the firewall of the server edition?
[17:25] <Pici> philsf: I thought there was a way to do that it mutt, but I haven't used it enough lately to remember the keybindings/settings.
[17:27] <Myrtti> philsf: I have a feeling I've read somewhere that roundcube might have had something similar, but I really am not sure...
[17:27] <Myrtti> "Threaded message listing"
[17:28] <Myrtti> don't know if it's suid tho, *shrug*
[17:29] <Pici> oh. webmail.  I'm really not doing well answering questions today.
[17:29] <bluethundr> nm.. got it! all hail google! http://knowledgebase.cc/software/linux/php-linux/php-version-unter-ubuntu-updaten/   :)
[17:29] <philsf> Myrtti, roundcube is not suid, it's fine. It's actually my first choice for a webmail, for now.
[17:30] <cloakable> gmail interface is mine <.<
[17:33] <kirkland> adam_g: ping
[17:33] <kirkland> adam_g: I hear from agy that he has an "nagios-openstack-plugins" package, that does some nagios based monitoring of openstack
[17:34] <kirkland> adam_g: i thought this might be a cool thing to squeeze into oneiric
[17:34] <Daviey> kirkland: he made a package of it?
[17:34] <adam_g> kirkland: ya! i saw that too
[17:35] <kirkland> Daviey: that's what he said
[17:35] <adam_g> kirkland: they are simple checks that use existing plugins.
[17:35] <adam_g> kirkland: dunno if that should be packaged with openstack, or added to ensemble formulas via a nagios relation on each
[17:37] <kirkland> adam_g: let's package with openstack
[17:37] <kirkland> adam_g: i see, i misread
[17:38] <kirkland> adam_g: Daviey: yeah, i think we should jam that into our openstack packaging
[17:38] <kirkland> adam_g: Daviey: it's small/trivial enough
[17:39] <Daviey> Hmm.. i'm not sure TBH.
[17:40] <adam_g> kirkland: theres some discussion around those packages atm, actually
[17:40] <Daviey> Does that mean installing on all Orchestra clients?
[17:40] <Daviey> openstack or not?
[17:42] <hggdh> we need some help on the server ISO tests: http://iso.qa.ubuntu.com/qatracker/build/ubuntuserver/notcompleted
[17:45] <kirkland> Daviey: adam_g: i'd see us just shipping the nagios config in /usr/share/openstack/nagios or something
[17:45] <kirkland> Daviey: adam_g: and nagios would be a suggested package
[17:46] <adam_g> kirkland: personally dont know if it should be handled in openstack packaging or elsewhere, as this model wont scale well when we start tacking on collectd configs, ganglia, rsyslog, etc
[17:46] <kirkland> adam_g: fair enough;  openstack-monitoring-plugins
[17:46] <patrickmw> jamespage, python-jenkins project is working great.  Thanks for pointing me to it
[17:46] <kirkland> adam_g: as a new package/project
[17:47] <kirkland> adam_g: and hang -collectd, -ganglia, -rsyslog off of that as binary packages
[17:48] <RoAkSoAx> smoser: hjowdy!! does clout-init always changes the hostname to ubuntuhost if not specified in cloud-init config
[17:48] <smoser> no
[17:48] <smoser> 'ubuntu' is what is in /etc/hostname of a clean image
[17:49] <smoser> so on first boot, hostname gets set to that, then cloud-init comes by and sets it to the hostname portion of data from the metadata service
[17:49] <adam_g> kirkland: sounds better to me. it'll just dump some potentially useful stuff into a directory for users to cherry pick from?
[17:49] <RoAkSoAx> smoser: well I'm installing on bare metal and on first boot, cloud-init is changing the hostname to ubuntuhost
[17:49] <RoAkSoAx> smoser: 2011-08-31 13:22:53,906 - __init__.py[DEBUG]: handling set_hostname with freq=None and args=[]
[17:49] <RoAkSoAx> 2011-08-31 13:22:53,918 - cc_set_hostname.py[DEBUG]: populated /etc/hostname with ubuntuhost on first boot
[17:50] <RoAkSoAx> smoser: however, the hostname was passed to the kernel args via the pxe file
[17:50] <smoser> hm.. it might do that i guess if there is nothing in metadata. i'd have to look.
[17:50] <kirkland> adam_g: yeah
[17:50] <RoAkSoAx> smoser: so in user-data file I should just add
[17:50] <RoAkSoAx> manage_etc_hosts: True
[17:50] <RoAkSoAx> hostname: my-hostname-here
[17:50] <kirkland> adam_g: worst case, /usr/share/doc/examples
[17:50] <RoAkSoAx> fwereade: ^^
[17:50] <kirkland> adam_g: just somewhere discoverable in /usr/share
[17:51] <kirkland> adam_g: we can haggle over the final placement
[17:51] <adam_g> kirkland: sounds reasonable
[17:51] <kirkland> adam_g: but the key would be to establish and distribute some best practices, that are immediately usable
[17:51] <kirkland> adam_g: without being the be-all/end-all configuration of such tools
[17:52] <kirkland> adam_g: at least for now, users would have to opt-in, and enable them
[17:52] <adam_g> kirkland: i like that, then any formula can install that package and use the plugins without maintaining them within the formula.
[17:52] <kirkland> adam_g: if we get to a point where we have concensus on the monitoring best practices, then we can talk about enabling some of the by default (if we ever get there)
[17:52] <kirkland> adam_g: eggggzactly
[17:53] <kirkland> adam_g: just putting them *somewhere* would give us a base to improve upon
[17:53] <smoser> RoAkSoAx, in oneiric if you just set 'fqdn' to fqdn then you'll be set.
[17:53] <kirkland> adam_g: without that, everyone will write and rewrite their own nagios/collectd/rsyslog/ganglia modules and reproduce lots of work
[17:53] <kirkland> adam_g: with that shipped base, hopefully they'll start from there, improve upon it, and show us their diffs
[17:53] <smoser> RoAkSoAx, but please open a bug.. if it is not set, cloud-init should not go messing things up.
[17:53] <smoser> if not set and not available in metadata or cloud-config that is.
[17:54] <adam_g> kirkland: yah, sounds good
[17:54] <RoAkSoAx> smoser: will do
[17:54] <kirkland> adam_g: cool
[17:54] <RoAkSoAx> smoser: right but cloud-init should respected the parameter that is passed on the kernel to set the hostname either way, right?
[17:55] <RoAkSoAx> smoser: i'm gonna file a bug anyways
[17:59] <smoser> passed on the kernel?
[18:00] <RoAkSoAx> smoser: bug #838280
[18:01] <smoser> RoAkSoAx, it has nothing to do with bare metal
[18:02] <smoser> what does the meta-data include that you've seeded ?
[18:02] <RoAkSoAx> smoser: meta-data only has instance-id
[18:04] <smoser> yeah, that is the thing. since it is not there, cloud-init is falling back to a default. and it should instead fall back to not setting up hostname.
[18:05] <RoAkSoAx> smoser: right, but isn't it possible to check if the hostname has been set on a kernel parameter? If so, it should not try to set a hostname again?
[18:05] <smoser> RoAkSoAx, it wont have been set via a kernel parameter.
[18:05] <smoser> cloud-init will know nothing about that.
[18:05] <RoAkSoAx> smoser: I guess that the only "easy" way to address this is to set the hostname on the meta-data on the cloud-init meta-data generated by ensemble
[18:05] <smoser> as it runs on first boot after install
[18:05] <smoser> and you wont have that kernel parameter then.
[18:06] <smoser> RoAkSoAx, thats not so easy, and probably wrong, as why would ensemble know the hostname ?
[18:06] <smoser> it should not have to.
[18:06] <RoAkSoAx> smoser: in cobbler it kinda does
[18:06] <smoser> i'd rather it not, and rathre not solve it that way.
[18:06] <smoser> let me look at fixing in cloud-init.
[18:07] <RoAkSoAx> smoser: the thing here is that when ever you deploy with ensemble, the deployed machine ends up having its hostname as 'ubuntuhost' instead of the hostname that was told to the machine on the kernel parameter
[18:07] <RoAkSoAx> smoser: which means that on first boot, cloud-init changes the hostname from whatever it was set to something that it shouldn't (ubuntuhost)
[18:08] <smoser> RoAkSoAx, i understand that.
[18:08] <smoser> but onfirst boot, you will not have that kernel parameter.
[18:08] <smoser> so cloud-init could not even read it if it wanted.
[18:09] <RoAkSoAx> smoser: righyt, but i *think* it is writing the   /var/lib/cloud/data/previous-hostname with the correct hostname set by the kernel parameter
[18:09] <RoAkSoAx> smoser: i will have to doubel check that
[18:10] <smoser> RoAkSoAx, sure, thats fine, but thats only because it finds it there.
[18:10] <RoAkSoAx> smoser: right that's why I'm sauing, during installation the hostname was set, but when cloud-init initializes post-installation on first boot, it changes the hostname when it shouldn't have
[18:11] <smoser> i think we're in agreement.
[18:11]  * RoAkSoAx sometimes wonders if he expresses himself correctly :)
[18:13] <smoser> RoAkSoAx, can you confirm that it is setting it to the string 'ubuntuhost' ?
[18:13] <RoAkSoAx> smoser: trying to but ubuntu installations are failing for some kind of archive error
[18:14] <RoAkSoAx> smoser: give me a sec since I'm gonna try to do it again
[18:14] <adam_g> lynxman: ping
[18:14] <lynxman> adam_g: pongie
[18:15] <adam_g> lynxman: regarding bug #818177, is LVM involved on that system at all?
[18:15] <lynxman> adam_g: definitely
[18:15] <RoAkSoAx> smoser: btw... cloud init is still failing to import the PPA , but when I do it manually, it doesn't fial to import it
[18:15] <lynxman> adam_g: so far what we've seen is that udev is not mounting properly
[18:15] <lynxman> adam_g: does that ring a bell?
[18:16] <adam_g> lynxman: im seeing some kind of race here, not sure if its related. but only happens when there is are LVs present
[18:16] <lynxman> adam_g: so looks like its the same one
[18:16] <lynxman> adam_g: or could be
[18:16] <lynxman> adam_g: do you have a bug for that race?
[18:16] <adam_g> lynxman: no, i just hit it yesterday and worked around it but tried creating some LVs to see if it reproduces, and it does
[18:16] <lynxman> adam_g: so looks like we might be getting somewhere
[18:17] <adam_g> lynxman: my rootfs is on a raw partition /dev/sda5, i created a VG with /dev/sda7 and a LV.. now boot fails. from what i could tell just now, it looks like pvscan/vgchange was happening before the nodes were created for the PVs in /dev
[18:18] <lynxman> adam_g: which fits with what I'm seeing as well because udev somehow didn't get mounted on time
[18:19] <adam_g> lynxman: what do you mean by 'udev didnt get mounted on time'
[18:20] <lynxman> adam_g: in my case doesn't get mounted at all, but I think this is clearly the race condition
[18:20] <lynxman> adam_g: udev is needed before it kicks in
[18:28] <Daviey> bug 838298 is up for grabs for anyone that wants it.. (MIR blocker)
[18:29] <smoser> RoAkSoAx, i have no idea why cloud-init woudl fail import from the ppa.
[18:29] <smoser> can you show me a box without touching it and  let me look aroun d?
[18:32] <RoAkSoAx> smoser: i'm reinstalling on the test-rig
[18:32] <RoAkSoAx> smoser: will let you know once it is done
[18:39] <Daviey> hallyn: Can you create a ~ubuntu-virt vanilla PPA for qemu 0.15?
[18:39] <Daviey> zul: can you push your libvirt new version to that PPA?
[18:39] <Daviey> .. want to release note, suggesting people try them.
[18:39] <zul> Daviey: which one?
[18:40] <Daviey> zul: You re-merged?
[18:40] <zul> Daviey: i never got it built properly it always ftbfs
[18:40] <Daviey> ah.. not a good candidate then :)
[18:40] <zul> Daviey: i didnt think so :)
[18:40] <Daviey> zul: i thought it fixed issues for you?
[18:41] <zul> Daviey: supposedly...but new issues crept up will building it
[18:41] <zul> s/will/while/g
[18:41] <Daviey> lovely
[18:41] <zul> so ill leave that for hallyn in oneirc+1
[18:41] <Daviey> so i think we cn nack the idea of a new version. :)
[18:41] <zul> i menally already have
[18:42] <Daviey> heh
[18:42] <Daviey> zul: the issues you have.. do you have LP bugs tracking them, linking to upstream commits? :)
[18:42] <zul> no since its not in the archive yet
[18:42] <Daviey> uh?
[18:43] <Daviey> I mean the reasons you wanted a new version?
[18:43] <zul> Daviey: oh no i havent
[18:43] <hallyn> zul: didn't you have an arm patch for the libvirt candidate though?
[18:44] <hallyn> Daviey: what precise ppa name should i use?
[18:44] <zul> hallyn: yeah i do...we are talking about the merge i was woring on (0.9.4) not yours
[18:44] <Daviey> hallyn: I guess qemu015-testing ?
[18:44] <hallyn> Daviey: but then libvirt can't go in there :)
[18:45] <Daviey> well yeah... /me is thinking ON HIS FEAT
[18:45] <zul> or trying to :)
[18:45] <Daviey> The updated information that we are dropping the new libvirt, means we are only testing qemu.. right?
[18:45] <zul> no...we are dropping the libvirt i was working on...no hallyns
[18:46] <Daviey> hang on.. i thought hallyn was just doing qemu, and zul was doing libvirt.. am i wrong?
[18:46] <RoyK> anyone tested XEN with ubuntu yet?
[18:46] <zul> yes you are wrong hallyn was working on both qemu and libvirt
[18:46] <Daviey> RoyK: yes, some
[18:46] <Daviey> more testing welcome :)
[18:47] <RoyK> does it require virt extensions in the cpu?
[18:47] <Daviey> no
[18:47] <Daviey> hmm, for hvm it does(?)
[18:47] <RoyK> xen in mainline kernel is most appreachiated ;)
[18:48] <RoyK> I've been trying to uses paravirt ubuntu on xencenter and that's not a stroll in the park
[18:48] <RoyK> hopefully it'll be a bit easier with 12.04LTS when that gets around
[18:48] <Daviey> RoyK: only if people like you test it during development :)
[18:48] <RoyK> not that I want that old xencenter beast
[18:49] <Daviey> RoyK: if you want to test it on oneiric, that would float my boat.
[18:50] <RoyK> I guess we can release a blade from the xencenter and use that
[18:50] <RoyK> some dual xeon thing with 16GB RAM
[18:50] <RoyK> should be sufficient for testing :P
[18:51] <RoyK> Daviey: any idea if the ubuntu xen will support any redundancy?
[18:52] <RoyK> say, shared iSCSI storage for the VMs etc
[19:03] <Daviey> RoyK: No idea, sorry.  As this is the release we are re-introducing xen, it will have both bugs and feature issues.  Your input, right now, could make a difference :)
[19:05] <smoser> RoAkSoAx, i'm not sure where 'ubuntuhost' is coming from
[19:06] <RoAkSoAx> smoser: neither am I :S
[19:06] <smoser> oh. wait. yes i am.
[19:06] <RoAkSoAx> smoser: please enlight me
[19:07] <robbiew> RoAkSoAx: call time?
[19:07] <smoser>  /usr/share/pyshared/cloudinit/DataSourceNoCloud.py
[19:07] <RoAkSoAx> robbiew: yep
[19:07] <RoAkSoAx> robbiew: ready when you are
[19:08] <RoAkSoAx> smoser: ahh! I see... so we'd need a way to autodetermine the hostname given to the kernel
[19:08] <smoser> RoAkSoAx, well...
[19:08] <smoser> it will work if it didn't have that there.
[19:08] <hggdh> server ISO tests still to be done: http://iso.qa.ubuntu.com/qatracker/build/ubuntuserver/notcompleted
[19:09] <RoAkSoAx> smoser: right cause previous-hostname does not contain the hostname that was given to the kernel
[19:10] <Daviey> hggdh: why does iscsi always get left out?
[19:11] <hggdh> Daviey: IDK. Perhaps because not everybody has an iSCSI setup? But it is always iSCSI...
[19:11] <smoser> that is not getting set right.
[19:11] <smoser> let me look at this.
[19:12] <Daviey> hggdh: iscsi setup is just using two kvm / virtualbox / etc vm's?
[19:13] <hggdh> Daviey: I would guess it would work, yes
[19:13] <hggdh> Daviey: problem is time, for me
[19:13] <Daviey> hggdh: that is what the test case HOW TO suggests :)
[19:13] <Daviey> hggdh: ack
[19:14] <hggdh> Daviey: there is a finite number of VMs I can run at the same time ;-)
[19:14] <hggdh> *very* finite
[19:14] <Daviey> hggdh: I've had over 50 running on one machine before :P
[19:15] <SpamapS> Daviey: I was wondering why you were walking funny that day..
[19:16] <jetole> Hey guys. Does anyone know how to change the domain name on a system? I'm pretty sure it's supposed to be the first domain name listed in /etc/hosts but want to make sure
[19:17] <Daviey> SpamapS: hah
[19:18] <Daviey> jetole: /etc/hostname
[19:18] <jetole> Daviey: no the domain name. not the host name
[19:21] <jetole> according to the man page for hostname (man 1 hostname), which I am reading now, the dns domain part of a fqdn hostname is set by calling gethostbyname(2) on the set hostname so if the /etc/hosts file is called before the dns when resolving names then I can change it there
[19:21] <jetole> so I guess I was right, it's /etc/hosts
[19:23] <smoser> hallyn, what would you expect 'reboot' to do within a libvirt container ?
[19:40] <aubre> Hey, is there a patched apache coming out for lucid ?
[19:40] <Pici> aubre: For what purpose?
[19:43] <andol> Pici: https://issues.apache.org/bugzilla/show_bug.cgi?id=51714 I assume.
[19:47] <Pici> andol, aubre: looks like there is a bug filed, but no one is assigned to it: https://bugs.launchpad.net/ubuntu/+source/apache2/+bug/837991 :/
[20:02] <aubre> Pici: the apache team has fixed it, it just needs to be rolled out: 2.2.20 http://projects.apache.org/projects/http_server.html
[20:02] <SJr> I keep getting this on my server e-mailed to root
[20:02] <SJr> /usr/share/sendmail/sendmail: 1177: /usr/sbin/sendmail-msp: not found
[20:02] <SJr> how do I fix it?
[20:08] <hallyn> smoser: I'm not sure.  probably kill all tasks and then hang.  what does it actually do?
[20:08] <smoser> kill all tasks and then hang. i think
[20:11] <hallyn> smoser: there's currently no way, without introspection, to tell whether a container wanted to reboot or poweroff.  lxc looks a the container's utmp.
[20:12] <hallyn> dlezcano is pushing a kernel patchset to fix that.
[20:22] <smoser> RoAkSoAx, so the ony thing I can come up with is that the network isnt really all the way up when this is initially happening
[20:23] <RoAkSoAx> smoser: right, yeah I was thinking that might be related to a network problem
[20:25] <smoser> yeah, thats it.
[20:25] <smoser> completely
[20:25] <smoser> /etc/init/cloud-init-nonet.conf is at fault
[20:26] <RoAkSoAx> smoser: is there a way to make cloud init wait for network to be completely up?
[20:26] <smoser> fix the bug i think
[20:26] <smoser> :)
[20:29]  * Daviey hands RoAkSoAx a GOTO 10.
[20:31] <smoser> RoAkSoAx, http://paste.ubuntu.com/679198/
[20:31] <smoser> that is what we need for cloud-init starting too early
[20:31] <smoser> and i think the hostname bug is as easy
[20:33] <RoAkSoAx> smoser: cool
[20:43] <RoAkSoAx> smoser: are you commiting that fix?
[20:43] <RoAkSoAx> to your upstart job?
[20:46] <smoser> yeah. i committed to trunk and pushed.
[20:46] <RoAkSoAx> smoser: cool
[20:46] <RoAkSoAx> smoser: will this hit oneiric soon?
[20:47] <smoser> well, not till after archive opens obviously. but we'll get one there as soon as we can
[20:47] <RoAkSoAx> smoser: alright, cool,  as long as it is before the end of next month we are all good
[20:47] <RoAkSoAx> hehe
[20:50] <lynxman> Daviey: you're handing GOTOs now? wow
[20:54] <davros> do eth0 and eth1's mask have to be the same? eth0 is 255.255.252.0 eth1 is 255.255.255.0
[20:55] <lickalott> are you using them on different networks?
[20:55] <lickalott> *for
[20:57] <davros> no one is connected to the modem eth1 is the dhcp listen
[20:57] <davros> and it gives out ip's that dont have connection to the internet
[20:59] <lickalott> no router in between?
[21:00] <davros> no its a rack server 2 nic's in one box
[21:00] <lickalott> histoically eth0 = your NIC and eth1 = secondary or wireless interface
[21:00] <lickalott> ahhh
[21:00] <lickalott> then i would say yes
[21:00] <davros> hmmm ok
[21:03] <davros> wonder if that will fix the dead ip's lol
[21:15] <frewsxcv> i have ubuntu-server running on an ec2 instance. if i stop it, will /mnt erase?
[21:17] <SpamapS> frewsxcv: probably
[21:17] <SpamapS> frewsxcv: unless you've moved it to be on top of an EBS mount.
[21:23] <robos> hello: ubuntu in a production environment. Would you go with 11.04 or 10LTS?
[21:24] <robos> I'm thinking 10LTS and it's not even close. Thoughts?
[21:24] <Corey> robos: LTS.
[21:24] <Corey> robos: Otherwise I'd call you a fool.
[21:24] <robos> hah, okay. Double checking :-) thanks
[21:25] <Corey> 10.04LTS.
[21:28] <lenios> why do you think there are releases twice a year if you're a fool to go without lts?
[21:30] <patdk-lap> many reasons
[21:30] <patdk-lap> but unless there is something you specifically need in a newer version, using the lts is much better
[21:30] <patdk-lap> as your not forced to upgrade it every year
[21:32] <SpamapS> Corey: why would he be a fool for running 11.04 ?
[21:33] <medberry> support life on non-LTS is 18 mo.s IIRC.
[21:33] <SpamapS> Nobody asked how long the server ies expected to live.
[21:33] <lenios> true medberry
[21:33] <patdk-lap> don't most servers last till the hardware fails? :)
[21:33] <SpamapS> Or whether he plans to utilize continuous integration...
[21:34] <SpamapS> patdk-lap: not these days.. they have this thing called virtualization.. ;)
[21:34] <patdk-lap> ya, I run the vm's the same way
[21:34] <medberry> so vzn servers can last longer than the hardware that's running them (via migration)
[21:34] <patdk-lap> till the die a horrable death, just it comes less often :)
[21:35] <SpamapS> There are many who simply deploy constantly into test VMs .. and this includes testing on newer versions of the OS.
[21:35] <SpamapS> At my previous company it took 2 years of blood sweat and tears to get rid of RedHat 8.0 because people wanted to keep the same old servers alive until they died.
[21:36] <patdk-lap> heh, I was stuck on fc1 :)
[21:36] <patdk-lap> couldn't do much about it, new hardware got rhel5
[21:36] <patdk-lap> been replacing that as I migrate services around
[21:36] <SpamapS> The new model was simple.. maintain config management.. use it to actualize everything.. when a new OS release came out (CentOS 5 in this case) try out all known tests on it, and if they all pass, migrate by re-birthing all services.
[21:37] <SpamapS> so I don't know if I'd call him a fool for considering 11.04
[21:37] <SpamapS> Perhaps he was just inquiring about the quality of 11.04 vs. 10.04
[21:37] <SpamapS> not the life cycle. :)
[21:40] <hallyn> stgraber: if/when you get a chance, could you take a look at my proposed fix for 838380?  If there are no problems I'd like to see it pushed after beta release.
[21:41] <hallyn> haha, just realized the problem with the sshd lxc template in ubuntu.  hillarious
[21:42] <soren> hallyn: pray tell :)
[21:42] <jj995> I was reading http://serverfault.com/questions/190/zfs-vs-xfs/72743#72743 and saw "using lvm snapshots and xfs on live filesystems is a recipe for disaster especially when using very large filesystems." -- have anybody had good/bad experiences taking snapshots with ~8TB on LVM2?
[21:43] <hallyn> soren: it wants to bind-mount the template itself onto the containers /sbin/init.
[21:43] <hallyn> the template is under /usr/lib/lxc/tempaltes.  THe contaienr's rootfs is getting set up under /usr/lib/lxc
[21:43] <hallyn> so we've overmounted it with the root before we get around to bind-mounting it :)
[21:44] <Corey> SpamapS: I'm a big fan of long term stability for servers.
[21:44] <Corey> Particularly given some shops' reluctantce to touch things in production.
[21:46] <SpamapS> Corey: me too.. I think its the simpler approach for most cases.
[21:46] <SpamapS> Corey: BUT it has massive draw backs for disaster recovery and scaling.
[21:47] <SpamapS> Which is what CI does well
[21:47] <Corey> SpamapS: Right, but that requires an ephermeral nature to many of your VMs.
[21:47] <Corey> "This current one will be shut down and replaced in another two weeks so who cares about patching it" for instance.
[21:47] <Corey> Requires a bit of a different development mindset. :-)
[21:48] <hallyn> soren: (opened bug 838410 on it, wont' be fixing it today)
[21:48] <SpamapS> 2 weeks is a pretty crazy life cycle
[21:49] <SpamapS> 6 months is quite doable
[21:50] <SpamapS> Corey: the idea is that because you're duplicating the process with every change to the system, you don't care if it lives for 2 weeks or 2 years.
[21:50] <SpamapS> Having effective, realistic staging capabilities is pretty tricky tho
[21:52] <Corey> SpamapS: Yeah, it's not trivial for sure.
[21:52] <soren> SpamapS: It's well worth the effort IMO, though.
[21:52] <Corey> I find that LTS gives added flexibility for "bad practices." :-) I say this while working on an EOL'd environment.
[21:52] <soren> SpamapS: But yes, very much non-trivial.
[21:54] <SpamapS> I kept 6 redhat 8.0 servers, behind two firewalls (border and backend), alive until .. well.. crap.. I left in 2010 and they were still running.
[21:54] <alamar> :D
[21:54] <SpamapS> Don't tell me about EOL environments. ;)
[21:56] <TheEvilPhoenix> lol
[21:56] <alamar> Linux chupachups 2.4.17 #3 Tue Jan 15 16:31:01 CET 2002 i686 unknown
[21:56] <alamar> ;)
[21:57] <alamar> it'S scary what's out there..
[21:59] <SpamapS> alamar: dayum ;)
[22:15] <stgraber> hallyn: looks good. I'll try to remember to sponsor that tomorrow. If I don't, just poke me.
[22:28] <frewsxcv> does anyone know if the WSGI module is installed by default on ubuntu?
[22:28] <hallyn> stgraber: great, thanks
[22:35] <patdk-lap> heh, there are craploads of companies with nt 3.5 systems still up and running
[22:36] <patdk-lap> just a few months ago, someone came out and said they had a stock pile of EOL sql servers, some bank
[23:36] <elz89> frewsxcv: are you still around? I had to install WSGI module on Ubuntu Server 11.04 when I was setting up a firefox sync server a while back.
[23:39] <frewsxcv> elz89: i learned that mod_wsgi is not what i want. uwsgi is the officially supported module and is included by default with versions >0.8.6
[23:39] <frewsxcv> i grabbed nginx from the PPA
[23:44] <lynxman> RoAkSoAx: hey, adam_g fixed the problem we were having, the machine should be ready for getting onwards tomorrow morning :)
[23:51] <airtonix> i'm having some difficulty getting xapian installed and recognised by haystack. http://invisibleroads.com/tutorials/xapian-search-pylons.html << says install xapian-core xapian-bindings-python