[00:15] <pukeko> hi there.. i have a Gbit rated lan, and am getting speed readings of approx 260mbits from XP clients when connecting to a local samba server ( at non peak time ) is this too low ?
[00:19] <qman___> pukeko, bitrate and throughput are two very different things
[00:20] <qman___> mediocre desktop computers can only transfer files in the 35-50MB/s range, older poorer performing ones lower
[00:20] <qman___> in ideal conditions, a windows vista/7 computer with a multi disk raid 0 can achieve 80MB/s from a linux samba server
[00:21] <qman___> the server must also have a multi disk raid to keep up with those numbers
[00:23] <hallyn> utlemming: smoser: stgraber: I may push http://people.canonical.com/~serge/lxc.debdiff  tonight.
[00:24] <pukeko> if i perform the same test (iperf) from another linux box im getting approx 700mbits -- am wondering if its the SMB 1.0 overhead ..
[00:25] <hallyn> (leaving tests running as i get some dinner)
[00:27] <stgraber> hallyn: "++    echo "Please login as Ubuntu." => shouldn't it be "ubuntu"?
[00:27] <hallyn> stgraber: I figured utlemming meant user Ubuntu
[00:27] <pukeko> should i set the Tcp Window to the same on the XP and the Linux ..?
[00:27] <hallyn> i.e., username
[00:27] <hallyn> i'll change it.  should i list the pwd too?
[00:28] <utlemming> hallyn: yup
[00:28] <stgraber> probably a good idea yeah
[00:28] <pukeko> if i do that i get around 700mbits also
[00:28] <stgraber> hallyn: I'm guessing this won't work until we have the new upstart though right?
[00:28] <hallyn> works fine :)
[00:28] <hallyn> the cloud images have lxcguest built in
[00:29] <stgraber> ah ok, so we'll break them when we remove lxcguest, good to know ;)
[00:29] <hallyn> in fact we'll have to think whether new upstart will cause it problems
[00:29] <hallyn> yeah
[00:30] <hallyn> i dno't know if automated cloud image generation will break, or just say "<shrug> no lxcguest, s'coo"
[00:30] <hallyn> utlemming: ^
[00:31] <utlemming> hallyn: I'm not sure I follow the concern
[00:31] <hallyn> utlemming: pretty soon lxcguest will be removed from precise
[00:31] <stgraber> utlemming: the lxcguest package won't exist anymore next week (hopefully)
[00:32] <utlemming> ah....that's not a problem
[00:32] <hallyn> sweet
[00:32] <utlemming> the package is being installed via the server^ pattern
[00:32] <utlemming> we are not explicitly installing it
[00:32] <hallyn> i don't know what that is
[00:32] <utlemming> apt package group
[00:33] <hallyn> ok, sweet.  thx.
[00:33] <stgraber> oh, which means that lxc will be automatically moved back to universe once lxcguest is dropped
[00:33] <stgraber> because currently the source and lxcguest are in main as lxcguest is in a server seed
[00:34] <stgraber> when that's no longer the case, it'll be fully moved back to universe and so anyone who's MOTU will have upload rights again
[00:34] <hallyn> are we ok with that, or do we beg for MIR for lxc?
[00:36] <stgraber> I think it'd make sense to have lxc at least be in the supported seed but that'll require an MIR
[00:37] <stgraber> I mean, we kind of hope people are going to use it in 12.04 otherwise we wouldn't put that much effort into it, so then I think it makes sense to have it as part of the LTS
[00:39] <hallyn> out for dinner, bbl
[01:26] <hallyn> stgraber: tests all pass.  did you ahve any more comments, or shoudl i push?
[01:26] <stgraber> hallyn: nope, that was the only thing I noticed
[01:26] <hallyn> ok thx
[01:36] <EvilResistance> anyone able to guide me through this?  http://askubuntu.com/questions/102381/configuring-ubuntu-server-to-act-as-a-gateway-similar-to-a-router
[01:41] <qman___> EvilResistance, assuming you have both interfaces configured already, eth0 is internet, eth1 is LAN, `echo 1 > sudo tee /proc/sys/net/ipv4/ip_forward; sudo iptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE; sudo iptables -A FORWARD -i eth0 -o eth1 -m state --state RELATED,ESTABLISHED -j ACCEPT; sudo iptables -A FORWARD -i eth1 -o eth0 -j ACCEPT`
[01:42] <EvilResistance> qman___, the configuration of eth1 is the main issue i'm having, it doesnt want to work :P
[01:42] <qman___> that is doing it live, to have said configuration automatic at boot requires a few config file changes
[01:43] <qman___> you'll have to be more specific
[01:43] <qman___> "it doesn't work" isn't enough to go on
[01:43] <EvilResistance> qman___, system isnt detecting the second networking card :P
[01:43] <EvilResistance> at least...
[01:43] <qman___> well, that's a hardware problem then
[01:43] <EvilResistance> not within networking... lspci has it, but its not being confiugred as an interface
[01:43]  * EvilResistance agrees
[01:43] <EvilResistance> qman___, alternatively, other question I had is related:
[01:43] <EvilResistance> how can i set up just the LAN part
[01:43] <mgw> any cobbler experts know what causes this on a cobbler replicate: Exception value: 'cobbler trigger failed: cobbler.modules.sync_post_restart_services'
[01:44] <qman___> ?
[01:44] <EvilResistance> qman___, that is to say, configure the box as a LAN controller, without having external gateway access
[01:44] <EvilResistance> and also, to provide IPs to each machine (also unconfigured on this box)
[01:44] <qman___> there's nothing to control
[01:44] <qman___> give it an IP
[01:44] <qman___> oh, you want to provide DHCP, you need a dhcp server
[01:44] <EvilResistance> mhm
[01:45] <EvilResistance> consider this is going to work with 45 systems, i dont want to be setting 45 IPs by hand :p
[01:46] <qman___> apt-get install dhcp3-server, or apt-get install dnsmasq
[01:46] <qman___> dnsmasq has the main benefit of also including DNS caching and automatic local DNS
[01:47] <EvilResistance> thanks.
[01:49] <qman___> for future reference, if you know what you need, but not which package to install, you can use 'apt-cache search x' to try and find it
[01:49] <qman___> among a larger list of things, 'apt-cache search dhcp server' returned both of the above results
[01:49] <EvilResistance> i see
[01:49] <Patrickdk> apt-cache search what I need
[01:50] <EvilResistance> qman___, last question, assigning an IP to the box itself for the LAN.  Can either dhcp3-server or dnsmasq do this automatically and give it, say, 192.168.1.1?  or does that get set up statically separate from the configuration of dnsmasq
[01:51] <qman___> your dhcp server must have a statically configured interface
[01:53] <qman___> actually that's not strictly true, in some rare instances
[01:53] <qman___> but for your purposes, and most all purposes, the DHCP server needs to already have a statically configured LAN interface in /etc/network/interfaces
[01:54] <qman___> the default gateway must always have a static IP
[01:56] <EvilResistance> indeed.  i assume i can configure the thing to hand out IPs from the 10.0.0.0/8 range, rather than 192.168.1.1?
[01:56] <EvilResistance> s/192.168.1.1/192.168.0.0\/16/
[01:57] <qman___> yes, you can use any range you want, though you should pick ranges in the designated private address spaces
[01:58] <EvilResistance> 10.0.0.0/8 is defined in the RFCs to be private-use if i'm not mistaken...
[01:58] <qman___> yes, it is
[02:00] <twb> 192.168/16, 10/8 and 172.16/12
[02:00] <twb> Also 169.something is reserved for the .local mDNS/DNSSD pseudodomain
[02:01] <qman___> 169.254/16
[02:01] <qman___> for APIPA
[02:02] <twb> avahi and all its friends need to FOAD tho
[02:19] <EvilResistance> alrighty, i'm going to go take a stab at this, i shall return and let you know of the results
[02:57] <EvilResistance> qman___:  i've installed dnsmasq, and it is running.  I've followed (as far as I can tell) the correct configuration modification to enable the integrated DHCP server, and assign from a range of 45 IPs in the 10.0.0.0/8 area, with a 12 hour lease: dhcp-range=10.0.0.5,10.0.0.50,12h.  I hook up one of the machines that i'm trying to get to work on this, and it doesnt receive an IP at all
[02:59] <qman___> EvilResistance, requisites: the server's interface must be up and configured, dnsmasq must have been restarted since the configuration change
[02:59] <EvilResistance> one second, *fixes something unrelated*
[03:05] <Resistance> okay, so now my system died >.> figures.
[03:06] <Resistance> ANYWAYS i've put this into /etc/network/interfaces, i'm not even sure this is correct for what needs to be configured: http://pastebin.com/BS7umd1q
[03:08] <qman___> that is not correct
[03:08] <qman___> gateway cannot be itself
[03:08] <qman___> you can omit a gateway if there is none
[03:09] <Resistance> that's what i wasnt sure about :P
[03:10] <Resistance> and still nothing :/
[03:12] <Resistance> of course, network manager on the system connecting to the DHCP thing doesnt give me detailed information at *all*
[03:14] <qman___> well, the mere presence of network manager can break everything
[03:14] <qman___> I assumed you were using a server
[03:14] <qman___> to eliminate network manager as the problem, 'sudo service network-manager stop'
[03:14] <qman___> other thing to check is the logs
[03:15] <Resistance> i am
[03:15] <Resistance> the server doesnt have network manager
[03:15] <qman___> and restart dnsmasq after you bring up your 10.0.0.1 interface
[03:15] <Resistance> the client system *connecting* to the server is
[03:15] <qman___> ah, ok
[03:15] <qman___> well, you can still stop network manager and use dhclient -v instead
[03:15] <qman___> that'll show you what's going on on the line
[03:15] <Resistance> but because network manager is a piece of [REDACTED], it doesnt give useful error messages
[03:15] <Resistance> ah
[03:16] <Resistance> ... only problem is that the system i'm trying to connect to the server is the system i'm talking from...
[03:16] <Resistance> and thats running off of a local wifi i have access to...
[03:16] <Resistance> if i stop network manager on here, the connection will die :/
[03:16] <qman___> unfortunately, yes
[03:17] <qman___> but network manager is pretty persistent
[03:17] <qman___> can't reliably dhclient with it running
[03:17] <qman___> can try though
[03:17] <Resistance> meh lemme load up colloquy off my iphone so i can continue to communicate here
[03:18] <Resistance> note my response speed will be far slower
[03:19] <Resist|WEB> okay, lemme bring down network-manager here...
[03:22] <Resistance> seems to be trying to broadcast to 255.255.255.255, but nothing seems to be responding...
[03:22] <twb> gateway line looks wrong
[03:23] <Resistance> there is no more gateway line.
[03:24] <qman___> ok, if nothing's coming back, then next step to try is to make sure you can communicate at all
[03:24] <twb> http://paste.debian.net/155527/ is how I'd do it
[03:24] <qman___> manually set an address with ifconfig or ip, and try to ping the server
[03:24] <twb> But you want a 0/0 route *somewhere*, it just makes no sense to have it routed to yourself
[03:25] <qman___> there's no internet in his scenario yet, because his second NIC doesn't work
[03:25] <twb> OK
[03:25] <qman___> but yeah, normally you'd set one
[03:29] <Resistance>  qman___ no respons... oh for the love of god...
[03:29]  * Resistance face desks
[03:29] <twb> Resistance: purge NM entirely
[03:29] <Resistance> I forgot to edit the firewall
[03:30] <qman___> that'll do it
[03:30] <qman___> stopping/starting network-manager works pretty reliably for my laptop
[03:30] <qman___> I often have to set it up as a router/gateway or bridge to troubleshoot stuff on the job, but also need network manager for wifi
[03:30] <qman___> and VPN
[03:31] <qman___> I know I don't NEED it for those functions, but it does them well and is way easier than doing it by hand
[03:33] <hallyn> zul: could you please syncpackage -d unstable numactl ?
[03:34] <arrrghhh> is there a way to find out what is the physical location of an image file ?  all i see is /dev/loop0 on /media/xda when i run 'mount'.  lsof didn't seem to find it either..
[03:34] <Resistance> bleh, still not working
[03:34] <Resistance> I'll fix it later, I have other stuff to do :/
[03:35] <twb> arrrghhh: losetup
[03:35] <twb> arrrghhh: /proc/mounts MAY tell you, I think that's new in 3.2
[03:35] <arrrghhh> ah, excellent
[03:35] <arrrghhh> 2.6.32 here lol
[03:36] <arrrghhh> sudo losetup -a did it.
[03:36] <arrrghhh> thanks
[03:36] <twb> mount -o loop uses losetup internally, you see
[03:37] <arrrghhh> interesting...
[03:37] <arrrghhh> the rabbit hole goes deep usually on linux.  especially ubuntu
[03:43] <hallyn> not to use arrrghhh's name in vain, but I just realized I lost an hour of last night's work by accidentally terminating an instance :(
[03:47] <arrrghhh> hallyn, i've been used for worse.
[03:50] <hallyn> even with wrath-of-khan-level gesturing?  good, cause it was painstaking ugly-patch-porting work...
[04:19] <cloneMX> hey all, someone alive?
[04:19] <twb> No
[04:20] <cloneMX> lol
[04:20] <cloneMX> someone not-bot alive?
[04:20] <arrrghhh> he's not a bot.
[04:21] <cloneMX> looks like
[04:21] <cloneMX> simple and clear
[04:21] <cloneMX> just a Not
[04:21] <arrrghhh> but he usually has a pretty good catalog of knowledge, he's answered more than a handful of questions i've had.
[04:21] <arrrghhh> er
[04:21] <arrrghhh> i guess i'm assuming he.
[04:21] <arrrghhh> meep.
[04:22] <kantlivelong> anyone here using apcupsd with a RFC1628 SNMP card?
[04:22] <arrrghhh> anyhoo, you got a question?
[04:22] <cloneMX> yep
[04:22] <twb> !anyone
[04:22] <cloneMX> well
[04:22] <arrrghhh> kantlivelong, i'm using apcupsd... but i don't think with that card.  sorry.
[04:22] <cloneMX> well cause polite first
[04:22] <kantlivelong> arrrghhh: its an ethernet based card :P
[04:22] <arrrghhh> i have a USB connection from the ups to the PC
[04:23] <kantlivelong> ive tried NUT and it seems good except one goof that it doesnt shutdown based on batt percent..
[04:23] <arrrghhh> hrm
[04:23] <arrrghhh> it's an APC?
[04:23] <kantlivelong> no
[04:23] <cloneMX> well mi problem is with some services at ubuntu 11.10
[04:23] <arrrghhh> oh
[04:23] <kantlivelong> RFC1628 compliant snmp web card
[04:23] <cloneMX> just stop sending bytes after a request
[04:23] <cloneMX> ssh and apache
[04:23] <kantlivelong> arrrghhh: which apcupsd says it supports
[04:23] <arrrghhh> hrm..
[04:24] <cloneMX> but the things is not with all host
[04:24] <cloneMX> just with some of it
[04:24] <kantlivelong> i hope its just my OS giving me the segfault for apcupsd:P
[04:24] <kantlivelong> (not ubuntu for the host node)
[04:25] <cloneMX> well think 2 problmems at the same time are not posible
[04:25] <cloneMX> lol
[04:27] <kantlivelong> i hope this works :)
[04:27] <kantlivelong> APCUPSD > NUT
[04:29] <cloneMX> guys you know if Ubuntu 11.10 got some kind of intruders network detector
[04:29] <twb> Is apcupsd APC-specific?
[04:29] <twb> I'm mainly usint nut because I don't know any better
[04:29] <twb> cloneMX: not out of the box.  Here is the one I use: http://cyber.com.au/~twb/doc/iptab.ips
[04:30] <twb> cloneMX: I also recommend running regular pentests from outside the network, e.g. openvas
[04:31] <cloneMX> well I was asking cause I dont know why the server is closing my network service conections like apache2 and ssh
[04:31] <cloneMX> but not from all Hots
[04:31] <cloneMX> hosts
[04:31] <cloneMX> : D
[04:31] <arrrghhh> have you checked the logs?
[04:32] <twb> cloneMX: probably because your switch is fucked or your NATing router is fucked, or you have a bad cable or similar.
[04:32] <arrrghhh> i would think all connections would drop if it was on the host side.
[04:32] <arrrghhh> no/
[04:32] <arrrghhh> ?
[04:32] <cloneMX> well im  using a brand new adsl modem
[04:32] <twb> cloneMX: do the normal fault isolation things, such as checking the logs (as arrrghhh suggests), turning other services off, turning debugging on/up, swapping out components
[04:32] <cloneMX> over a pppoe
[04:32] <cloneMX> no logs
[04:32] <cloneMX> no firewall
[04:33] <twb> testing components in isolation, e.g. connecting locally before connecting over the network
[04:33] <cloneMX> no info
[04:33] <arrrghhh> no logs?
[04:33] <arrrghhh> there's logs.
[04:33] <cloneMX> yep
[04:33] <arrrghhh> apache logs, ssh logs
[04:33] <cloneMX> no information over the logs
[04:33] <cloneMX> it just work from some hots
[04:33] <arrrghhh> hrm
[04:33] <cloneMX> and not from others
[04:33] <cloneMX> for example
[04:34] <cloneMX> now I can see the host from home
[04:34] <cloneMX> my adsl
[04:34] <twb> cloneMX: there is a limit to how much we can babysit you
[04:34] <cloneMX> lol
[04:34] <cloneMX> ty anyway
[04:34] <arrrghhh> cloneMX, first, quit using the enter key as  punctuation.  ask your question all in one line.
[04:35] <twb> cloneMX: it sounds like you're inexperienced enough that you really need someone on the ground who can isolate the fault.
[04:35] <twb> cloneMX: find your local linux user group (LUG) and ask them to recommend someone
[04:35] <cloneMX> kk
[04:35] <cloneMX> letme explain it
[04:36] <twb> You've been explaining the symptoms for 48 hours without getting anywhere.  Repeating yourself is not going to help.
[04:36] <cloneMX> im just did what you said
[04:36] <arrrghhh> oh i haven't been here to see it, sorry.
[04:36] <qman___> from what you've said, it sounds like a generic network fault, which could be your NIC, or a piece of hardware somewhere between the endpoints, or just the internet being the internet
[04:36] <cloneMX> from basics
[04:36] <arrrghhh> cloneMX, WAN residential connections are best effort.
[04:37] <arrrghhh> they have pretty much no QoS.  certainly they won't take anything that is passed thru your LAN
[04:37] <cloneMX> ok
[04:38] <qman___> home connections in general, and DSL in particular, are fairly unreliable
[04:38] <arrrghhh> yea... DSL suxxxxxors.
[04:38] <arrrghhh> i'm so very glad i vowed to never use it again :D
[04:38] <qman___> and if you're running on the ISP provided equipment, it's just as likely the box they gave you is crap
[04:38] <qman___> they do it all the time
[04:38] <arrrghhh> see the previous on dsl sucking ^^
[04:39] <cloneMX> ok
[04:39] <arrrghhh> cloneMX, you want a reliable host?  VPS.
[04:39] <twb> arrrghhh: what are you using instead, satellite down, PSTN up?
[04:39] <arrrghhh> lol@PSTN
[04:39] <twb> That's the only alternative to DSL in .au
[04:39] <twb> FSVO alternative = can't get DSL
[04:39] <arrrghhh> cable.. tis the best my area has to offer at the time being
[04:39] <arrrghhh> oh
[04:40] <arrrghhh> i forget about regional issues
[04:40] <arrrghhh> some areas have fiber
[04:40] <arrrghhh> some have... no choice
[04:40] <qman___> yeah, in the US cable is the best trade off
[04:40] <arrrghhh> indeed
[04:40] <qman___> FIOS is a better line, obviously, but is only available in select areas
[04:40] <arrrghhh> comcrast has run some fiber, and their connections are the fastest.
[04:40] <qman___> cable is widely available and fairly reliable as long as you have good equipment
[04:40] <arrrghhh> yup, no FiOS here.
[04:40] <EvilResistance> qman___:  i beg to differ on the FiOS thing
[04:40] <arrrghhh> qman___, always purchase my own.
[04:41] <arrrghhh> i wish that community internet would take hold
[04:41] <EvilResistance> qman___:  identical allowable internet bandwidth in the same location, homes literally next to each other, comcast has more net reliability than FiOS
[04:41] <arrrghhh> look up 'greenlight cable'
[04:41] <EvilResistance> at least out at my home
[04:41] <arrrghhh> i installed that head-end.  10mb up AND down for $40/mo
[04:41] <EvilResistance> out here where I am now, though, Comcast has the domination
[04:41] <arrrghhh> 100/100 for... i want to say $99/mo
[04:42] <cloneMX> lol
[04:42] <qman___> comcast also has a 250GB/month cap
[04:42] <qman___> FIOS does not
[04:42] <cloneMX> ok guys ty
[04:42] <arrrghhh> i know.  dicks.
[04:42] <Darkwing> I like my Cox cable... No cap, static IP address
[04:42] <Darkwing> 25up 10down
[04:42] <arrrghhh> wait, what?
[04:42] <twb> Darkwing: ah, but is it a static IPv6 addrses
[04:42] <arrrghhh> other way around i assume
[04:42] <qman___> for home lines anyway, business class comcast is better
[04:42] <arrrghhh> 25up lol
[04:42] <EvilResistance> qman___:  if available ;P
[04:42] <Darkwing> IPv4, but it works for a small home server :D
[04:42] <EvilResistance> business class isnt available out here
[04:44] <qman___> most of the customers I work with have comcast business lines, but at my house there is only charter communications
[04:44] <qman___> it's charter, dialup, satellite, or T1
[04:44] <qman___> no other options are available
[04:45] <twb> What's charter, a guy on a horse with a bag of USB keys?
[04:45] <qman___> cable, just a lesser known provider
[04:45] <qman___> costs more than comcast with less speed, but no usage cap
[05:16] <raddy_> Hello Everybody
[05:16] <raddy_> I installed postgresql 8.4 in ubuntu 10.04.
[05:17] <raddy_> But, /etc/postgresql/8.4/ folder not created
[05:17] <raddy_> Can anybody suggest what could the problem
[08:26] <lapsusbrutus> bat crashes on startup.. needs 9101 to function but fails to listen to this port. installed bacula also.  did same thing on a debian computer and no problem at all.
[08:49] <journeeman> Connection problems :/
[08:56] <journeeman> Sorry for flooding but, I'll repost the question - I recently installed Oneiric Server-amd64 on a Dell Poweredge 2950 which has an embedded ATI ES1000 card. I need to use virt-manager to manage a few VMs (having trouble with virsh) so, I installed xubuntu-desktop. On reboot, the screen just keeps flickering for a while before blanking out. The radeon driver is installed. There is no xorg.conf file
[08:56] <journeeman>  to edit in /etc/X11.
[08:57] <journeeman> Please help :)
[09:07] <henkjan> journeeman: you can connect with virt-manager on your workstation to libvirt on the server
[09:07] <henkjan> journeeman: or ssh -X to your server
[09:08] <henkjan> installing a graphical environment on your server to run virt-manager is not wat you should do imho
[09:09] <journeeman> Oh ok
[09:10] <journeeman> Will try them out. Thank you henkjan :)
[09:13] <chelz> journeeman: pretty sure you can use virt-manager to remotely connect to machines
[09:13] <chelz> File -> Add Connection -> Connection: [dropdown menu]
[09:13] <chelz> in that dropdown you can select various kinds of remote access
[09:14] <chelz> journeeman: so you really don't need to install something as heavy as a full DE
[09:14] <chelz> and even then, lxde is better than xfce, or dwn, ion, ratpoison, etc
[09:15] <greppy> ( as a screen and tmux user, /me <3 ratpoison )
[09:16] <chelz> dwm*
[09:57] <jkyle> hwo would I laod hte megaraid_sas driver in a preseed.cfg? my hands off install is not detecting the raid card correctly
[10:05] <RoyK> jibel: afaik that should happen automatically if the PCI ID of the board is known to the driver
[10:11] <Ruetobas> anyone using orchestra? just installed on an vm instance and the distro list is empty...
[10:46] <jkyle> RoyK: the driver is a megaraid_sas, it says it can't detect the drives. I load teh driver, and the install proceeds
[10:46] <jkyle> so, need to automate that selection
[10:46] <yahoo123> Good day to you all!
[10:48] <jkyle> tried: disk-detect/module_select select megaraid_sas
[10:57] <jkyle> actually, I think I got it...nice
[11:08] <Vivek> kirkland, roaksoax, SpamapS, Daviey : Anyone around ?
[11:17] <Vivek> roaksoax: The /etc/rsyslog.d/99-orchestra.conf is pasted here http://paste.debian.net/155555/
[11:46] <Daviey> uksysadmin: Can you raise a bug please?
[11:46] <uksysadmin> yeah will do
[11:47] <Daviey> Trying to work out why others haven't spotted this as yet.
[11:47] <Daviey> including our CI
[11:53] <soren> Daviey: What's this+
[11:53] <soren> ?
[11:54] <Vivek> Daviey: hI
[11:54] <Daviey> soren: bleed over from #openstack
[11:54] <Daviey> hey Vivek
[11:54] <Vivek> Daviey: I am still working on the DHCP issue from yesterday.
[11:55] <Daviey> Vivek: happy times!
[11:55] <Vivek> roaksoax has asked me for my 99-orchestra.conf file which was pasted earlier.
[11:55] <soren> Daviey: Ah, there. Awesome.
[11:58] <Vivek> My query was regarding creation of directories in the /var/log/orchestra/rsyslog directory corresponding to machines in the eth0 interface.
[11:58] <Vivek> In dnsmasq.conf also I have set up the interface eth0 to not give out any dhcp requests via eth0.
[11:59] <Vivek> eth0 is in bridged mode and the provisioning server runs inside a virtualbox . eth1 is the internal network range.
[11:59] <Vivek> The internal network range is something like 192.168.1.x
[11:59] <Vivek> eth1 is in internal networking mode.
[12:00] <Vivek> I would like to know why those directories named after 10.x.x.x I.P Addresses are created inside /var/log/orchestra/rsyslog.
[12:19] <Daviey> Vivek: i suspect they are created when the installer starts sending stuff back to the rsyslog server
[12:22] <Vivek> ok
[12:23] <Vivek> I understand that.
[12:23] <Vivek> Daviey: Can you point me to exact source code that causes this to happen ?
[12:27] <Vivek> Daviey: I would also like to know why there is dhcp and dhcp3 directories in /etc
[12:27] <Vivek> Which one does orchestra use ?
[12:31] <Daviey> Vivek: i think that is how rsyslog works.. you'd need to grep the rsyslog souce code for that.
[12:40] <journeeman> chelz, henkjan Thank you :)
[12:51] <Vivek> ok
[12:52] <Vivek> Daviey: There are no machines installed in the 10.x.x.x series
[12:52] <Vivek> via eth0
[12:53] <Vivek> All machines are commissioned via the eth1 interface with I.P Address 192.168.1.x series
[12:57] <Daviey> Vivek: i suspect it's something to do with your enviroment, virtualbox bridge?
[12:58] <Vivek> Yes
[12:58] <Vivek> eth0 is in virtualbox bridge
[12:58] <Vivek> eth1 is in internal network mode.
[12:58] <Vivek> eth0 has the 10.x.x.x series I.P Address assigned to it
[12:59] <Vivek> and eth1 has the 192.168.1.x series I.P Address.
[13:51] <zul> good morning
[14:01] <Daviey> zul: does bug 892754 impact us?
[14:01] <zul> looking
[14:02] <zul> Daviey: possibly
[14:04] <zul> Daviey: i think we should be fine
[14:05] <Daviey> zul: oh?
[14:05] <zul> we are already running 0.7.3 on precise
[14:05] <Daviey> zul: Precise has v7 of sqlalchemy, no?
[14:06] <Daviey> right
[14:06] <Daviey> "test suite doesn't work with sqlalchemy v7"
[14:06] <zul> Daviey: right but according to the pip-requires you need >= 0.6.3
[14:07] <zul> https://review.openstack.org/#change,3814
[14:08] <Daviey> zul: Are you following what i am saying?
[14:08] <zul> Daviey: i might be missing
[14:08] <Daviey> precise has 0.7 (aka v7) right?
[14:08] <zul> right
[14:08] <Daviey> "test suite doesn't work with sqlalchemy v7"
[14:08] <Daviey> *doesn't*
[14:10] <zul> oh suck i totally missed that :(
[14:10] <zul> yeah we need that patch
[14:30] <jjohansen> hallyn: so I have a bug or two that have slowed me down and I am to tired/stupid to see it atm.  I need to get a few hours sleep and then I get back to finding the problem and getting the ppa together for you.
[14:39] <hallyn> jjohansen: thanks
[14:45] <jcastro> jamespage: excellent blog post, I've totally ripped it off and resyndicated it everywhere
[14:45] <jamespage> jcastro: so I noticed - thanks!
[15:10] <hallyn> kirkland: byobu-tmux problem
[15:10] <hallyn> kirkland: log in as user1, run byobu-tmux.  life is good.
[15:10] <hallyn> kirland: log in as user2 in another window.  run byobu-tmux.  frowny face.
[15:11] <hallyn> kirland: 5929  mkdir("/tmp//tmux-1001", 0700)    = -1 EACCES (Permission denied)
[15:23] <tgardner> changing the orchestra.preseed to 'd-i  user-setup/encrypt-home boolean true' causes the installer to stop and complain that empty passwords are not allowed. Anyone seen this before ?
[15:24] <kirkland> hallyn: how odd
[15:24] <kirkland> hallyn: what's the uid of user1 and user2?  1000 and 1001?
[15:25] <kirkland> hallyn: did user1 DoS user2 by mkdir'ing /tmp/tmux-1001?
[15:25] <kirkland> hallyn: you can change the tmux socket path with the -S option (see tmux.1)
[15:29] <hallyn> kirkland: just reporting what i've found so far.  i'll test a fresh install later.  didn't explictly do any DoSing.  seems like byobu-tmux should pick a socket it can use automatically :)
[15:29] <hallyn> if i can reproduce on clean install i'll file a bug
[15:29] <kirkland> hallyn: yeah, I've talked to the upstream about this, it is actually pretty easy to annoy another user, though not completely DoS them
[15:30] <Vivek> kirkland: Hi
[15:32] <kirkland> Vivek: hi
[15:33] <Vivek> I am still trying to figure out the reason for those directories in /var/log/orchestra/rsyslog.
[15:33] <Vivek> kirkland: Does canonical have any case studies of Orchestra implementations ?
[15:38] <tgardner> jamespage, have you been testing orchestra preseeds with encrypted home directories? I'm finding that it stops the automated install.
[15:39] <kirkland> Vivek: not that I know of, though as I mentioned last week, I no longer work for Canonical, sorry
[15:39] <kirkland> Vivek: the /var/log/orchestra/rsyslog directories are the remote logging information from deployed clients
[15:39] <Vivek> Correct.
[15:39] <kirkland> Vivek: including the installation logs
[15:39] <Vivek> Correct.
[15:40] <Vivek> In my case I have used your installation instructions.
[15:40] <Vivek> 1) Set up a virtual machine in Virtualbox
[15:41] <Vivek> 2) eth0 is bridged to the external world and has an I.P Address in 10.x.x.x range.
[15:41] <Vivek> 2) eth1 has an internal network set up in the 192.168.1.x range.
[15:41] <Vivek> Nodes are getting commissioned without any issues.
[15:41] <Vivek> So far so good.
[15:42] <Vivek> Now in /var/log/orchestra/rsyslog/ I am getting directories of the form 10.x.x.x which should not happen as Orchestera is only supposed to commission nodes in the 192.168.1.x series.
[15:43] <Vivek> In dnsmasq I have used the interface=eth1 option so no dhcp requests should go via eth0
[15:43] <kirkland> Vivek: hmm, well, something's wrong then...  as that means that rsyslog is picking up remote system logging from 10.x addresses
[15:43] <Vivek> ok.
[15:43] <Vivek> Which according to me is a bug.
[15:43] <kirkland> Vivek: i really need to pass you over to Daviey and/or roaksoax for help on this
[15:44] <Vivek> ok np, I've had my conversations with them earlier today.
[15:44] <Vivek> Daviey, roaksoax : Are you folks around ?
[15:44] <Daviey> Vivek: I don't know what else i can add. Something is clearly posting logs back to the rsyslog, if we saw them, it would help.
[15:45] <Vivek> I can pastebin the logs np
[15:45] <Vivek> Which logs do you need ?
[15:46] <Vivek> Let me pastebin those logs.
[15:49] <Vivek> Daviey: http://paste.debian.net/155584/
[15:49] <Vivek> That is what is done.
[15:49] <Vivek> happening rather...
[15:50] <Daviey> Vivek: Golly.
[15:50] <Vivek> root@orchestra:/var/log/orchestra/rsyslog/2012/02/08/orchestra# ping 10.130.55.205
[15:50] <Vivek> PING 10.130.55.205 (10.130.55.205) 56(84) bytes of data.
[15:50] <Vivek> 64 bytes from 10.130.55.205: icmp_req=1 ttl=128 time=0.987 ms
[15:50] <Vivek> 64 bytes from 10.130.55.205: icmp_req=2 ttl=128 time=0.922 ms
[15:50] <Daviey> Vivek: So, orchestra is the hostname it came from.
[15:50] <Vivek> ok...
[15:50] <Daviey> i'm going to guess that is localhost of the rsyslog server
[15:50] <Daviey> Therefore, orchestra has done nothing incorrect.
[15:50] <Vivek> No
[15:51] <Vivek> 10.130.55.205 is not the localhost of the rsyslog server
[15:51] <Daviey> Vivek: you aren't listening. :)
[15:51] <Vivek> The I.P Address is 10.130.55.x but not 205.
[15:51] <Vivek> ok
[15:51] <Daviey> oh wait
[15:51] <Daviey> Vivek: what is 10.130.55.205
[15:51] <Daviey>  ?
[15:52] <Vivek> It is a machine on my corporate network trying to contact the orchestra server for some reason.
[15:52] <Daviey> So what is the problem?
[15:52] <Vivek> It is not  a node commissioned by orchestra
[15:52] <Daviey> Vivek: ignore orchestra.. this is rsyslog
[15:52] <Vivek> ok.
[15:53] <Daviey> Someone sat in a desk next to you tried to ssh to your orchestra server, right?
[15:53] <Vivek> Why are remote servers trying to log to the orchestra syslog ?
[15:53] <Vivek> I don't think so
[15:53] <Daviey> They are not..
[15:53] <Vivek> Unless the IT department is spying :)
[15:53] <Daviey> The log originated FROM localhost.
[15:53] <Vivek> ok
[15:54] <Vivek> I don't think any attempt was made to ssh to 205
[15:54] <Vivek> My concern is dnsmasq broadcasting still on eth0, a tcmpdump says no....
[15:54] <Daviey> Vivek: You have it the wrong way around...
[15:54] <Vivek> ok.
[15:54] <Daviey> .205 was trying to ssh into your orchestra server.
[15:55] <Vivek> Yes, I got it. That is how remote clients provide the logs to rsyslog.
[15:55] <Vivek> Why is .205 trying to ssh into the orchestra server ?
[15:55] <Daviey> Vivek: I can book a flight, and conduct a survey around your office if you want?
[15:55] <Vivek> lol
[15:56] <Daviey> Vivek: find out who has .205, and ask them?
[15:56] <Caribou> jdstrand: ping ?
[15:56] <Vivek> Yes have 6000 staff so that would not be feasible :)
[15:56] <Vivek> Anyways, thanks.
[15:59] <Vivek> Let me phrase the question in another way, when a new node comes up, is it possible to view that log some where on orchestra, rsyslog ?
[16:00] <Vivek> As far as I know it can be viewed in /var/log/orchestra/rsyslog/ but there I find those 10.x.x.x directories also.
[16:01] <Vivek> If i write to parse those logs and automate some action based on those logs which logs should I refer to ensure that a new node has come up ?
[16:01] <Vivek> write a script
[16:02] <Vivek> Or even pointing me to section of code that creates those logs would be appreciated...
[16:02] <kpettit> Anybody know of a good opensource drop box type of tool?  I need a way to let users upload large files and share them via link.  But we want it on our own systems
[16:02] <jdstrand> Caribou: hi
[16:02] <Caribou> jdstrand: morning
[16:03] <Caribou> jdstrand: I have a question for you that should be easy
[16:04] <Caribou> jdstrand: when doing NAT through iptables, what is needed to have tftpd packets accepted from the NATted network ?
[16:04] <Caribou> I read that nf_nat_ftp & nf_conntrack_ftp modules were needed
[16:06] <Caribou> jdstrand: right now, the tftpd daemon gets the request but refuses connection from the NAT address
[16:06] <smoser> rbasak, so for your private cloud issue, lynxman wants there to be an entry for `hostname` in /etc/hosts.
[16:06] <smoser> but we've had repeated issues with doing that.
[16:06] <jdstrand> Caribou: so nat clients are trying to connect to an un-natted tftpd?
[16:06] <smoser> one way it pisses off some thing, one way  it pisses off another.
[16:07] <Caribou> jdstrand: what do you mean by "un-natted tftpd" ?
[16:07] <lynxman> smoser: it's just so we're in line with the experience we deliver with the server install
[16:08] <rbasak> smoser: how about doing it if a hostname is explicitly specified in user-data, but not otherwise?
[16:08] <rbasak> Otherwise specifying a hostname with cloud-init is broken
[16:08] <lynxman> smoser: I recall we had a talk about this at the rally
[16:08] <smoser> rbasak, then just set 'manage_etc_hosts: localhost'
[16:08] <smoser> and you'll be happy
[16:08] <smoser> or even 'template' i think
[16:08] <smoser> http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/doc/examples/cloud-config.txt
[16:09] <Caribou> jdstrand: the tftp requests (actually PXE boot requesting pxelinux.0) coming from a 2nd NIC in a NATted subnet
[16:09] <jdstrand> Caribou: is the tftpd daemon on the same nat network as the clients?
[16:10] <Caribou> jdstrand: no, that's the issue
[16:10] <rbasak> Aha, I didn't know about this setting. Thanks!
[16:10] <jdstrand> Caribou: ok, so you see the nat address of the client in the tftpd logs, not the address of the router doing the nating?
[16:16] <koolhead17> hi all
[16:22] <Vivek> koolhead17: Hi
[16:22] <koolhead17> hi Vivek
[16:24] <Daviey> Vivek: I said, those locations get created by rsyslog itself.
[16:25] <Vivek> Daviey: I am trying to identify what those machines are by nmap
[16:25] <Vivek> Those machines that are trying to ssh
[16:25] <Daviey> Vivek: honestly, if you have 6000 staff - and have no way of tracking internal IP addresses back to an owner, you have MUCH bigger problems
[16:26] <Vivek> Daviey: Tracking them would involve coordinating with the Internal IT which is again going to be a PITA.
[16:26] <Vivek> Daviey: Let me do my homework first :)
[16:28] <Daviey> cool
[16:31] <koolhead17> so what are you tracking Vivek?
[16:31] <Vivek> Looks like IT monitoring me :)
[16:31] <Vivek> Need to figure that out mate.
[16:31] <Vivek> Had a talk with out manager.
[16:32] <Vivek> our*
[16:32] <koolhead17> are you downloding porn at work?
[16:32] <Tixos> hey
[16:32] <Vivek> No :)
[16:32] <Tixos> can someone tell me if this is a 'default' ubuntu user on a new server
[16:32] <Tixos> citadel:x:103:112:Citadel system user,,,:/var/lib/citadel:/bin/false
[16:33] <Tixos> googled the name, and it comes up with some admin panel ?
[16:33] <koolhead17> Tixos: where did you got the image? did you check MD5 checksum
[16:33] <Tixos> its a dedi, the provider installed it
[16:33] <Tixos> and trust me they are useless
[16:33] <koolhead17> Vivek: better work then downloading poer
[16:33] <Vivek> I got the nmap output, the OS is Microsoft Windows. with the open ports they are running.
[16:33] <rbasak> smoser: I've tried manage_etc_hosts: template and manage_etc_hosts: localhost but neither seem to have any effect. Is there something I'm missing?
[16:34] <koolhead17> Tixos: your at wrong place then
[16:34] <Tixos> not really
[16:34] <Tixos> look at my questions, its perfectly valid here
[16:34] <Tixos> also, citadel is in the repos, so if you dont want to help, thats fine
[16:34] <koolhead17> Tixos: the provider can remaster and upload the image. :)
[16:35] <koolhead17> afaik during deployment one has to create a user
[16:35] <Tixos> you havent even answered my question, im only assuming from your replies that its not default
[16:35] <koolhead17> Tixos: yes
[16:35] <Tixos> there is no home dir for that user
[16:36] <Tixos> no home DIR at all infact
[16:36] <Tixos> for a user that is...
[16:37] <koolhead17> Tixos: its a application
[16:37] <Vivek> koolhead17: The downloads I do are perfectly legal and done after managements approval. I don't know if using tor is the reason.
[16:37] <Tixos> its not installed though >    p   citadel-client                                            - complete and feature-rich groupware server (command line client)
[16:37] <koolhead17> no security threat. :) install server like mysql on the host and check /etc/passwd you will see
[16:38] <Vivek> In that case I am stopping that too now that IRC SSL ports are open.
[16:38] <koolhead17> Vivek: will talk tomorrow in office
[16:39] <Tixos> which is the correct way to create a user again, useradd or adduser :P
[16:39] <Tixos> ive been told off for this before
[16:39] <smb> zul, smoser So I finally figured out why there is bug 922486. I hope to have spelled out all my findings along with one proposal how to fix it. Based on testing it seems to work well but I guess someone with libvirt experience may want to make sure this is not too much of a hack.
[16:41] <Vivek> koolhead17: Yes, in that case IT will have to answer for disrupting my work :)
[16:41] <Vivek> koolhead17: Anyways tommorow.
[16:42] <Tixos> Fetched 15.4MB in 0s (21.2MB/s)  yum
[16:49] <zul> hallyn: ^^^
[16:50] <zul> smb: it looks fine to me, but i want hallyn's oppinon as well
[16:51] <Tixos> adduser or useradd?
[16:51] <smb> zul, Ah, ok. Sure. Sorry, forgot him as libvirt stuff reviewer
[16:52] <hallyn> smb: looking
  seems to make sense, not knowing any of that code.  if it fixes it, go for it :)
[16:55] <hallyn> zul: let's hope we don't forget that fix if we go with 0.9.9
[16:56] <smb> Tried to be careful and have it in the ubuntu patches section with that debdiff. Cannot upload it though. :)
[16:57] <hallyn> smb: no it's just that i already cut a 0.9.9 candidate, just have to not forget about your patch :)
[16:57] <hallyn> smb: you don't have upload rights?
[16:58] <smb> hallyn, nope. not beyond kernel (so only harmless things)
[16:58] <hallyn> heh
[16:58] <hallyn> smb: so that debdiff has been tested, and should be pushed?
[16:59] <smb> hallyn, Hm, I tested the version with a lot more debugging in it. Give me a sec, I have not yet tested the compile of the debdiff
[16:59] <hallyn> ok.  your comment in #6 also seems sensible, just based on a glance
[17:01] <smb> hallyn, Only that change I did never make to test it. Just looked a bit odd, and I was wondering why there was no real error message in the logs
[17:06] <smb> hallyn, Hm, think I need a bit more testing. Noticed some error messages when booting and am not sure they are because of my change...
[17:16] <smb> hallyn, Ok, seems the "ERROR:  Can't find hypervisor information in sysfs!" when booting in non-xen mode is independent to my change in libvirt. So it looks ok to upload it
[17:18] <hallyn> smb: the exact debdiff in comment 5, no changes?
[17:19] <hallyn> i'll toss in the new 'start on' at the same time then
[17:19] <smb> hallyn, yes exactly that
[17:20] <hallyn> ok, will push, thanks
[17:20] <smb> hallyn, thanks for uploading. I'll pick up the results tomorrow
[17:27] <hallyn> pushed - gnight
[17:29] <Vivek> laters
[17:29] <Vivek> &
[17:51] <SpamapS> jamespage: here?
[18:02] <hallyn> save the earth!
[18:02] <hallyn> oops, wrong chan :)
[18:02] <hallyn> i need to better control my windows! :)
[18:03] <hallyn> (you know, to reduce wasted packets, which waste energy)
[18:04] <acidflash> is anyone familiar with bcache?
[18:08] <webPragmatist> i'm trying to figure out how to install http://old-releases.ubuntu.com/ubuntu/pool/universe/t/ttf2pt1/
[18:08] <webPragmatist> it doesn't seem to be a part of a package in apt
[18:10] <hallyn> zul: have you had a chance to test libvirt 0.9.9?
[18:11] <hallyn> i'm going to fire off a qa-regression-test against it i guess
[18:15] <zul> hallyn: yeah i tested it a little last night, ill do some more tonight
[18:36] <utlemming> hallyn: it looks like you can't SSH into the cloud images when run under LXC
[18:37] <utlemming> hallyn: I'm looking now. The root cause is that SSH keys are not being generated on boot within LXC
[18:37] <hallyn> utlemming: it *might* be due to a missing upstart event bc of something that doesn't happen in containers (yet).
[18:49] <utlemming> hallyn: it looks like sense cloud-init is not seeing a data-source, it does not generate the SSH Keys
[18:51] <adam_g> utlemming: what does cloud-init use as a data-source when booting in a container?
[18:51] <adam_g> (ive got no idea, just curious)
[18:52] <utlemming> adam_g: it looks for the 169.???.???.??? address
[18:52] <adam_g> utlemming: thats the ec2 metadata service
[18:53] <adam_g> utlemming: when we boot cloud-init on hardware in teh CI lab, we're injecting metadata and userdata into /var/lib/cloud/seed/nocloud-net/ after installation, which gets picked up on first boot by cloud-init
[18:55] <adam_g> utlemming: http://paste.ubuntu.com/834299/ http://paste.ubuntu.com/834302/
[18:56]  * utlemming looks
[18:59] <utlemming> hallyn: I'll have to revist that script later, but I think we'll end up with ubucloud and ubucloud-local, where the ubucloud-local unclouds the image.
[19:02] <hallyn> utlemming: you mean the templates?  If so, we can just add '--local' to the template-specific flags
[19:02] <hallyn> though i'd rather to '--ec2', and have local be the default
[19:03] <hallyn> so, 'lxc-create -t ubuntu-cloud -n cloud1 -- --ec2'
[19:03] <utlemming> hallyn: yes...and I like that
[19:03] <hallyn> if no ec2, then always create the ssh keys, and don't waste my time looking for 169.*
[19:05] <utlemming> hallyn: although, I would have a generic label like "--no-cloud"
[19:05] <utlemming> er, "--cloud"
[19:05] <zul> hallyn: new libvirt looks ok so far
[19:08] <hallyn> utlemming: that's good.  how involved is the mangling we have to do in the rootfs?
[19:09] <hallyn> zul: qrt tests going ok so far.  though ih ad to edit the qatest/qatest.xml to not specify /usr/bin/qemu as the emulator
[19:09] <utlemming> hallyn: it should be easy
[19:09] <utlemming> hallyn: and less mangling than you'd think
[19:09] <zul> hallyn: ah cool
[19:10] <utlemming> hallyn: let me test my fix
[19:36] <hallyn> zul: actually, virsh appears to be hung doing an attach-interface (as part of the qrt)
[19:36] <zul> hallyn: thats not good
[19:37] <kraut> hi
[19:37] <kraut> i'm using a nfs-export from my ubuntu server on my dvb reciever for timeshift. it happens sporadic that the video freezes and i see at that moment a high usage of kworker
[19:37] <kraut> is this an issue in anyway?
[19:43] <stgraber> hallyn: ETA for upstart upload is 5pm eastern (so I believe 4pm for you)
[19:45] <hallyn> stgraber: exciting!
[19:47] <stgraber> hallyn: do you already have a lxcguest-less LXC source package?
[19:48] <hallyn> no
[19:49] <hallyn> stgraber: should we create an empty lxcguest package for it?  Or just yank it out of the source pkg?  or have lxc now conflict with old lxcguest?
[19:50] <stgraber> hallyn: well, upstart will conflict with lxcguest, so there's no way anyone could install the package ;)
[19:50] <stgraber> hallyn: I guess we should just drop it entirely from the source and update the template to only install it pre-precise
[19:50] <stgraber> then file a binary removal bug to get lxcguest out of the archive to avoid having it included in the cloud images
[19:52] <stgraber> that should take care of all the use cases I can think of
[19:57] <smoser> hallyn, did you hvae a questionon generation of ssh keys in cloud images ?
[19:59] <smoser> cloud-images don't have ssh keys generated, and don't generate them unless there is a instance-id found.
[19:59] <smoser> we maybe should fix that so that they generate the keys anywhay.
[19:59] <utlemming> smoser: I did...I'm patching the lxc-ubuntu-cloud to take care of that
[20:00] <smoser> wait...
[20:00] <smoser> no, lets do it right.
[20:01] <smoser> cloud-init should do that as leaving no ssh keys is quite generally useless.
[20:01] <smoser> what did you to disable cloud-init?
[20:01] <utlemming> smoser: I'm preseeding some meta-data
[20:02] <utlemming> smoser: and allowing users to define host name, instance-id and user-data
[20:02] <utlemming> smoser: which I think is valuable to the end user
[20:03] <smoser> yeah, that makes sense. but keys should get generated anyway.
[20:04] <hallyn> stgraber: were you queueing up any lxc changes right now?
[20:05] <stgraber> hallyn: nope
[20:07] <utlemming> hallyn: here's my patch for lxc-ubuntu-cloud, http://paste.ubuntu.com/834391/
[20:09] <hallyn> utlemming: ok, thanks
[20:09] <hallyn> smoser: ^ that fits in with what you want (always generating keys) too?
[20:10] <smoser> hallyn, but wait.
[20:10] <smoser> hostname ?
[20:10] <smoser> isnt that non-sensical?
[20:11] <smoser> or conflicting with an lxxc container setting ?
[20:11] <danp> yo! I'm wondering if the scripts used to build HVM AMI ami-976da7fe (099720109477/ubuntu/images-sandbox/hvm/ubuntu-lucid-daily-amd64-server-20110930-backport-kernel-oneiric) are available somewhere. I can't seem to find anything for that in the ec2-publishing-scripts or automated-ec2-builds repos /cc smoser
[20:12] <utlemming> afaik, it doesn't
[20:12] <smoser> danp, they are there.
[20:12] <smoser> "hvmify"
[20:12] <utlemming> danp: the EC2 registration tools are NDA for HVM though
[20:12] <smoser> thats the only difference between the hvm and the instance-store.
[20:12] <smoser> er.. i  meant to say hvm and ebs.
[20:13] <smoser> but, also, what utlemming said.
[20:13] <smoser> but more than the NDA bit is your account wont have credentials to say --virtualization-type=hvm
[20:13] <danp> you mean the ability to directly register images as HVM?
[20:13] <smoser> on a register.
[20:13] <smoser> yeah.
[20:13] <smoser> now...
[20:13] <smoser> if you want to try to be tricky..
[20:13] <danp> I don't believe I need that if I do some hackery and run CreateImage against a running/stopped instance
[20:13] <hallyn> smoser: utlemming: no, --hostname/-H doesn't conflict with an arg, but it does conflict with an option you can specify inthe config file
[20:14] <smoser> (i've wanted someone to do this, and only tried once or twice because failures cost $2.8 or whatever it is).
[20:15] <smoser> http://paste.ubuntu.com/834399/
[20:15] <smoser> danp, i recently did that on HP public cloud, which basically replaced a running oneiric instance with a precise instance.
[20:16] <smoser> you could follow that and get into a root fs from an ephemeral disk and populate the clean root ebs.
[20:16] <smoser> but the times i've tried, there was mysterious failure
[20:17] <smoser> hallyn, right. which doesn't make much sense. and cloud-init will read `hostname` fine. so i dont think you nee dthat
[20:17] <smoser> utlemming, you have funny indentation in that
[20:17] <danp> smoser: my plan has been to run a cc1.4xl using the AWS AMI, stop it, detach the root EBS, attach a prepped EBS at /dev/sda1 (can't attach to just /dev/sda), then run CreateImage against it
[20:17] <hallyn> smoser: keep the rest?
[20:18] <smoser> danp, yeah, that did not work for me.
[20:18] <smoser> it does seem like it should
[20:18] <smoser> and i'm interested in knowing if it does for you.
[20:18] <danp> could probably ask AWS for the ability to register HVM directly as well. but it would be nice to not need that
[20:18] <smoser> hallyn, well, do you pick up a dependency on 'bc' for that ?
[20:19] <smoser> danp, i honestly do not know anything about this, but i suspected when my start/stop/detach/attach/register failed, that the failure was intentional
[20:19] <hallyn> stgraber: note that we have things like net-device-added lo which still could stand to have lxcguest
[20:20] <stgraber> hallyn: not in precise though
[20:20] <smoser> hallyn, other comments on that patch there are to not use full paths to files
[20:20] <danp> smoser: yeah, my guess would be not being able to attach the partitioned EBS to /dev/sda. it shows up in the AWS AMI DescribeImage block device mapping as /dev/sda1 though. not sure
[20:21] <smoser> trust PATH
[20:22] <hallyn> smoser: utlemming: anything wrong with just uuidgen | cut -c -8 ?
[20:22] <danp> smoser: please forgive my bzr ignorance, I've run `bzr branch lp:~ubuntu-on-ec2/ubuntu-on-ec2/ec2-publishing-scripts` and `bzr branch lp:~ubuntu-on-ec2/vmbuilder/automated-ec2-builds` and I don't see hvmify in either of those
[20:22] <hallyn> stgraber: not for runlevel 2, but ther emight be something out there waiting for lo?  or not?
[20:23] <smoser> danp, its part of ec2-image2ebs
[20:23] <danp> got it. thanks!
[20:23] <stgraber> hallyn: all initscripts I saw so far are explicitly ignoring lo for net-device-up
[20:24] <hallyn> ok
[20:24] <stgraber> hallyn: I'd really like to see lxcguest go away completely, if that becomes a problem, I'd rather have the hack part of ifupdown/upstart than in lxcguest
[20:24] <hallyn> excellent
[20:24] <hallyn> then away it goes
[20:24] <stgraber> yeah!
[20:24] <hallyn> can anyone explain why i can read paste.ubuntu.com jsut fine, but when i hit 'downlaod as text' i have to go through auth?
[20:26] <stgraber> IIRC it's to avoid using paste.ubuntu.com for file storage
[20:26] <stgraber> by making it difficult to retrieve the raw data
[20:30] <danp> smoser: I see now. I was hoping to find a script that installed the backported kernel and all that as well. would that have been done with the way these tools were run to create that image?
[20:32] <hallyn> is uuid-runtime always present?
[20:33] <smoser> danp, well we dont install a backported kernel.
[20:33] <smoser> but just mount the image chroot (modify /etc/resolv.conf) and apt-get install stuff
[20:45] <danp> smoser: oh, but it looks like this is lucid with a kernel PPA added and linux-image-3.0.0-12-virtual installed from there. was wondering exactly how that was done so I could replicate it as closely as possible
[20:48] <smoser> ah. you booted the lucid sandbox one i uploaded once.
[20:49] <smoser> i created that like this:
[20:49] <smoser>  http://paste.ubuntu.com/834442/
[20:49] <danp> smoser: yeah. that's the only lucid HVM AMI available, right?
[20:49] <danp> outstanding! thank you very much
[20:49] <smoser> yes, but note the 'sandbox' in its name
[20:50] <danp> understood
[20:55] <adam_g> anyone happen to know the significance of the difference in the return code of the same 'ip addr add' command (2 vs 254) on two different systems: http://paste.ubuntu.com/834454/
[20:56] <adam_g> glancing thru the iproute source, i was only to find this :) http://paste.ubuntu.com/834456/
[21:02] <utlemming> hallyn: regarding uudigen....I really don't have much preference, just generating a random id
[21:03] <hallyn> utlemming: cool, will do the shorter thing then
[21:03] <hallyn> waiting to test (libc6 upgrade is slowing me down)
[21:45] <stgraber> hallyn: lxcguest-free world -15min ;)
[22:02] <stgraber> hallyn: it's upstart o'clock ;) uploading now
[22:05] <stgraber> hallyn: uploaded
[22:10] <axisys> how do I remove md3 and add the slices to md2 to increase the raid10?
[22:10] <axisys> md2 : active raid10 sdg1[4] sdc1[0] sdd1[1] sdh1[5] sdf1[3] sde1[2]
[22:10] <axisys> md3 : active raid10 sdh2[3] sdf2[1] sdg2[2] sde2[0]
[22:21] <hallyn> stgraber: sorry, how does http://people.canonical.com/~serge/lxc.debdiff look?
[22:23] <hallyn> zul: hm, that test also fails with 0.9.8
[22:23] <hallyn> zul: so, if libvirt is treating you all right today and tomorrow... maybe we should push 0.9.9.  <cringe>
[22:24] <zul> i havent had any problems with lxc i havent tried with kvm yet
[22:24] <hallyn> ok.
[22:25] <hallyn> utlemming: /etc/apt/sources.list has http://ubuntu-mirror.localdomain/ubuntu .  that's kinda useless in no-cloud situation
[22:25] <utlemming> hallyn: yes it does
[22:25] <utlemming> s/does/is
[22:26] <hallyn> what's the easiest way to have it use deafult mirror there you think?
[22:26] <hallyn>  /etc/cloud/templates/sources.list.tmpl
[22:26] <hallyn> ?
[22:26] <utlemming> hallyn: well, cloud-init is the one doing it, so I think a cloud-config line
[22:27] <hallyn> that would look like what?
[22:27] <utlemming> hallyn: one minute...
[22:28] <utlemming> hallyn: right under my "#cloud-config:" line, you want to add, "apt_mirror: <URL>"
[22:28] <utlemming> hallyn: methinks that reading the host machines /etc/apt/sources.lists and fetching the URL out of there would be the most prudent course of action
[22:33] <hallyn> utlemming: that seems a bit fragile (who knows which line to use).  There is a MIRROR option in /etc/default/lxc.  Use that if set, else use archive.ubuntu.com?
[22:34] <utlemming> hallyn: I like that idea.
[22:34] <CharlieSu> Hi all.  I'm using a preseed file to automate my Ubuntu installation for a bunch of computers.  Everything works wonderfully, but I'm looking for a strategy to give each computer a unique hostname automatically.  Anyone do something like this ever?  Possibly based off of the MAC address?
[22:35] <hallyn> utlemming: testing, will show you debdiff before i push
[22:35] <utlemming> hallyn: sounds good
[22:36] <hayer> How can I set up shared directories in pure-ftpd-mysql?
[22:44] <stgraber> hallyn: sorry for the delay, was dealing with upstart FTBFS because of the new libc ;)
[22:44] <stgraber> hallyn: looking now
[22:47] <hallyn> stgraber: fwiw i'm stil making changes to the lxc-ubuntu-cloud template, but not touch lxcguest
[22:49] <stgraber> hallyn: ok. The diff looks good, I'll just check with the actual code next to it though ;)
[22:51] <hallyn> stgraber: ?
[22:51] <stgraber> hallyn: right, looks good (wanted to check what else was in that else/elsif and make sure it was indeed only the code installing lxcguest)
[22:57] <hallyn> ah ok
[23:08] <stgraber> hallyn: new libnih was uploaded which once build will fix upstart's own build, so in an hour or so everything should be built
[23:08] <stgraber> hallyn: upstart built fine on i386 though, so you can use that for testing (once it's published)
[23:38] <hallyn> utlemming: http://people.canonical.com/~serge/lxc.debdiff   works for me (tm)
[23:40] <utlemming> hallyn: looks good to me
[23:41] <hallyn> stgraber: so I'm pushing lxc?
[23:42] <stgraber> hallyn: yep, I expect upstart amd64 to start building in the next 10min or so
[23:45] <hallyn> lxc is away
[23:53] <stgraber> hallyn: cool!
[23:58] <hallyn> stgraber: bug 929086 has me confused
[23:59] <hallyn> (it *is* listed in Depends)
[23:59] <stgraber> hallyn: yeah, doesn't make any sense :)