[00:00] <renagadex> I'm trying to start a *simple* webchat, but I'm stuck at implementation.... I don't know how exactly to move past polling.
[00:00] <renagadex> Any suggestions?
[00:01] <cloakable> Look at a php jabber client?
[00:01] <renagadex> will do
[00:12] <Vivek> Any developers around at these times ?
[00:14] <Vivek> I am facing a situation where dnsmaq is serving dhcp requests on eth0 when it is not supposed to do that, in /var/log/orchestra I have directories created with the I.P Address range of eth0
[00:14] <Vivek> I am running virtual box with eth0 in bridged mode and eth1 in internal networking mode.
[00:15] <Vivek> Orchestra configuration should only permit dhcp requests via eth1 and not eth0
[00:15] <Vivek> in /etc/dnsmasq.conf I have made the change interface=eth1.
[00:16] <hallyn> Vivek: check with ps -ef if that file is beign used.  last i tried i had to edit /etc/default/dnsmasq instead.
[00:17] <Vivek> hallyn: What changes did you make there ?
[00:17] <Vivek> Any howto or doc pointing to this would be most welcome.
[00:18] <hallyn> Vivek: this is just something I found earlier today.  adam_g should have a better idea of whether it's supposed to be like that.  (I'm guessing it's a bug)
[00:19] <hallyn> Vivek: but, you can check /etc/init/dnsmasq.conf to see how /etc/default/dnsmasq is used.  (i'd be more helpful, but right now can't)
[00:28] <hallyn> Vivek: use DNSMASQ_OPTS="--interface=eth0" or whatever
[00:32]  * hallyn out
[00:48] <nOStahl> hey guys, fresh install of 11.10 server here
[00:48] <nOStahl> I cant seem to set public/static ip on it though
[00:48] <nOStahl> been all over google nothings working
[00:48] <nOStahl> dhcp works fine
[00:49] <nOStahl> someone help me figure out what im missing ?
[00:49] <cloakable> What are you putting in /etc/network/interfaces?
[00:50] <nOStahl> one of the static ips    for address
[00:50] <nOStahl> 255.255.255.0 for submask
[00:50] <cloakable> Uh-uh. Can you pastebin it for me?
[00:50] <nOStahl> gateway  is public ip of the router
[00:51] <twb> cloakable: oh noes then you might get his public IP!!1!
[00:51] <cloakable> twb: apparently so! Given he's put the public ip for gateway
[00:51] <cloakable> nOStahl: You need to put the LAN ip of the gateway for IP, for one.
[00:52] <cloakable> nOStahl: for gateway address, that is
[00:52] <nOStahl> k
[00:53] <nOStahl> what is the broadcast line anyways?
[00:54] <cloakable> Not usually needed to be set, defaults to 255.255.255.255
[00:54] <cloakable> Which is fine
[00:54] <cloakable> nOStahl: pastebin yet?
[00:54] <twb> cloakable: I believe interfaces(5) needs the address and at least the netmask or the broadcast line
[00:55] <cloakable> twb: address minimum
[00:55] <nOStahl> im running back and forth from the terminal heh
[00:55] <twb> That must be new
[00:56] <twb> It used to crack the shits around debian 6
[00:56] <cloakable> it also needs iface <interface> inet static
[00:57] <cloakable> But I can't see nOStahl's interfaces file due to not being psychic.
[00:57] <twb> cloakable: are you looking at a 0.7 ifupdown package?
[00:57] <cloakable> twb: yup
[00:57] <cloakable> 11.10
[00:58] <nOStahl> http://pastebin.com/0q9XdRrT
[01:00] <cloakable> Okay, that network line isn't needed
[01:02] <nOStahl> and my public ip's start with 99 so its not a 192 ip
[01:02] <cloakable> http://pastebin.com/27ujEJVH
[01:03] <cloakable> Is it behind a NAT?
[01:04] <cloakable> Also in that case, keep the broadcast line
[01:04] <nOStahl> how about the resolv.conf
[01:04] <nOStahl> can it just have nameserver 208.67.222.222 in it?
[01:04] <nOStahl> or does it need the search line?
[01:05] <cloakable> It can just have the nameserver
[01:05] <cloakable> Doesn't your gateway run dns?
[01:06] <nOStahl> aye
[01:06] <nOStahl> but its att
[01:06] <nOStahl> heh
[01:06] <nOStahl> the dsl at my office some times pops up dns errors cant find www.google.com heh
[01:07] <cloakable> Aha
[01:07] <cloakable> You can also put google dns
[01:07] <nOStahl> so i'll try it with the router nameserver
[01:07] <nOStahl> see if i can do sudo /etc/init.d/networking restart after that
[01:08] <cloakable> I only run static interfaces on my router personally. Everything else is either dynamic or static dhcp leases
[01:08] <nOStahl> I dont like the interface of the att modem
[01:09] <nOStahl> lets see if it blows up
[01:09] <nOStahl> one sec
[01:10] <nOStahl> rtnetlink error
[01:11] <nOStahl> file exists
[01:15] <nOStahl> any ideas
[01:16] <cloakable> Hrm, not really.
[01:16] <cloakable> I've used static dhcp leases for ages >.>
[01:20] <Vivek> hallyn: ok
[01:20] <nOStahl> hrmm
[01:21] <nOStahl> i dont even see any place in my att modem to set static
[02:09] <roaksoax> kirkland it is already but probably in upgrades is not being done
[02:29] <twb> Sanity check: in /etc/ppp/peers/foo if it says usepeerdns, that should overwrite resolv.conf, right?
[02:29] <twb> resolvconf (the package) isn't installed, but resolv.conf isn't being overwritten when the PPPoE handshake finishes.
[02:31] <twb> Never mind, it was being overwritten
[02:36] <twb> So the *actual* problem is this: pppoe is writing a correct resolv.conf.  Something *else* is repeatedly blowing it away and putting in broken entries.
[02:43] <maxb> Is NetworkManager installed/running? That's the only thing other than resolvconf that I know of, which rewrites resolv.conf
[02:52] <twb> Good question; it isn't.
[02:52] <twb> I *think* the problem was actually, ironically, /etc/init.d/pppd-dns, which moved resolv.conf.pppd-backup over the top of resolv.conf
[02:53] <twb> I don't know yet why that would have triggered but it's the only explanation I can find
[02:56] <_dinsdale> thinking about starting a linux softwareraid with a seagate caviar black and barracuda green. Would anyone recommend against this choice or will the probably slower green drive be a bottleneck for the main drive?
[03:02] <twb> _dinsdale: don't use green drives in a server
[03:02] <twb> EVER
[03:02] <twb> And yes it would be a bottleneck if the green is not configured as a write-behind / read-mostly node in the array
[03:02] <twb> Still be a write bottleneck if you flooded the I/O
[03:06] <_dinsdale> twb: just seeing it now - it's only 5900 RPM .... brrr
[03:07] <twb> Give it to your kisd to take apart
[03:08] <_dinsdale> :-)
[03:18] <chelz> everyone hates on greens but they're inexpensive dangit
[03:18] <chelz> just make sure you have a solid and good raid and/or backup system
[03:25]  * starlocke|soho appears.
[03:28] <starlocke|soho> there's this ubuntu 11.10 server guide that suggests to doing "sudo apt-get install mail-stack-deliver" to get a bunch of email systems installed... i get an "E: unable to locate package"... is it in a particularly special repository?
[03:30] <chelz> hmm
[03:31] <chelz> starlocke|soho: http://packages.ubuntu.com/search?keywords=mail-stack-delivery
[03:31] <starlocke|soho> i'm also wishing that /etc/issue would list the ubuntu codenames... XD
[03:32] <starlocke|soho> yeah, it exists... it's just not getting into my particular server. =\
[03:32] <starlocke|soho> ah... naturally...
[03:32] <chelz> starlocke|soho: that's not so good. if an apt-get update doesn't fix it
[03:33] <starlocke|soho> my copy-paste-fu needed some work.
[03:33] <starlocke|soho> nothing to see here, nothing at all. :)
[03:33] <chelz> starlocke|soho: working?
[03:34] <starlocke|soho> perfectly.
[03:34] <chelz> good :)
[03:55] <starlocke|soho> yeah... i've had this question for ages now... what does "smarthost" mean? the term comes up in the postfix config.
[04:04] <qman__> starlocke|soho, smarthost configuration means pointing your server to your 'internet site' mail server to send mail
[04:05] <qman__> something I typically do, set up the main mail server, then set all the other servers in a smarthost configuration pointing at it, so I can get email alerts for SMART data and such
[04:06] <starlocke|soho> mmm... as in hard drive SMART...? i'd like to clarify whether the smart in smarthost has anything to do w/ that capitalized SMART...
[04:06] <qman__> nothing whatsoever
[04:07] <qman__> smartd can send email alerts when drives have SMART events
[04:07] <qman__> I configure my severs in that email configuration so I can get said email alerts
[04:07] <qman__> among others
[04:08] <starlocke|soho> smartd = smarthost? it's such a weird term... "smarthost" XD
[04:08] <qman__> no
[04:08] <qman__> smartd is the hard drive monitoring service
[04:08] <starlocke|soho> k.
[04:08] <qman__> a smarthost configuration is very simple
[04:08] <starlocke|soho> so what exactly makes a smarthost a smarthost? lol
[04:08] <qman__> server A is your internet site mail server
[04:08] <qman__> server B is not an internet mail server, but you want to be able to send mail from it to the internet
[04:09] <qman__> so you configure server B in a smarthost configuration, where the specified smarthost is server A, which is already capable of sending mail
[04:09] <starlocke|soho> ah.
[04:09] <qman__> then server B just relays its mail through the smarthost, server A
[04:09] <starlocke|soho> that's a great clarification.
[04:09] <starlocke|soho> thanks.
[04:10] <starlocke|soho> i was suspecting something along those lines... it's just weird that they decided to use the word smarthost XD
[04:10] <qman__> that's the commonly accepted term
[04:10] <qman__> it's not postfix or even linux specific
[04:11] <starlocke|soho> hehe. all right then. memorize new term.
[04:11] <starlocke|soho> :D
[04:25] <starlocke|soho> qman__: by the way... where's a good read about that stuff on the web?
[04:31] <chelz> starlocke|soho: probably the documentation for your mailserver of choice
[04:32] <starlocke|soho> okay... let's see if grep can find "smarthost" in man pages... :3
[06:38] <twb> Urgh, just for yuks I tested precise's current d-i
[06:39] <twb> EVEN IN priority=low, it only has a single "autoconfigure network" -- you can't pick between DHCP, IPv6 RA, &
[06:40] <twb> I appear to have picked one of the NICs that isn't cabled, because currently it's just hung for ten minutes trying to do rdnssd
[06:42] <twb> If I kill rdnssd, it carries on assuming the network is fully configured, when clearly (from ip a) none of the NICs even have an IP
[06:42] <twb> What a crock
[06:43] <_ruben> nice
[06:43] <_ruben> or "nice" actually
[06:43] <twb> Strangely, picking the second (of four) nics, it first does a test for no carrier, that passes, and it proceeds to try dhclient, which fails
[06:44] <SpamapS> twb: There's a lot of work going on to make IPv6 solid in precise..
[06:44] <twb> Now I have two of the four NICs cabled, both to a network that hands out DHCP addresses
[06:44] <SpamapS> twb: stgraber would be very interested to hear about the problems you're seeing
[06:44] <twb> SpamapS: it looks like that work isn't finished yet :P
[06:44] <SpamapS> twb: I know that a new ifupdown just landed last week
[06:44] <SpamapS> not sure about netcfg tho
[06:44] <twb> SpamapS: ifupdown isn't even used in d-i
[06:44] <twb> (right?)
[06:44] <SpamapS> right
[06:45] <twb> OK interesting, on the *third* nic, it gives me actual output while rdnssd runs, and it fails quickly, and appears to fall back to dhcp, which succeeds.  2 rdnssd and 1 dhclient process are still running
[06:46] <twb> so the first nic, which had the strange results, was one that was physically cabled.  I can tell because it has the same chipset as the third one (it's a two-port pcie nic)
[06:57] <twb> It has hung again
[06:58] <twb> Ah, OK, it seems the first time was hanging in the same place as well -- AFTER giving hostname and domain.
[06:58] <twb> Last times in log are writing DHCP stanza for eth2 / Success!
[06:59] <SpamapS> twb: would you perhaps have time to file these as bugs in d-i ? I will make sure they're looked at ASAP
[06:59] <twb> On tty1 all you see is the bogl black background with a black-on-white last line where stuff you type (e.g. ^[[A) appears
[06:59] <twb> Well, you can dump my irc log... I need to get back to actually doing the lucid install about now
[07:01] <twb> If I kill either rdnssd proc, then I get back to the menu and I can actually continue on with picking the mirror and the udebs to install
[07:01] <twb> Hm, it doesn't prompt me to use sha256 passwords -- is that just implied now?
[07:02] <twb> It also ignores the DHCP-supplied NTP servers and defaults to ntp.ubuntu.com
[07:02] <twb> Which I guess doesn't matter too much, since post-install the DHCP client will override that
[07:07] <twb> Doing mkfs.ext4 /dev/sda (via partman) seems to have hung at 33%, here sda is a 60GB OCZ Vertex.  Note: sda, not sda1
[07:08] <twb> Ah, running it by hand, I can see it chugging away doing a TRIM first, which seems to be O(n)
[07:09] <twb> But that did finish after the TRIM in O(1), so I guess it defaults to lazy allocation these days?
[07:10] <twb> WTF, after "mkfs.ext4 /dev/sda" by hand finishes, partman thinks it's an *ext2* filesystem, and won't offer the "don't reformat" option
[07:11] <twb> That is, it won't offer it when picking "Use as: ext4"; it does when picking "Use as: ext2"
[07:12] <twb> blkid says /dev/sda is ext4
[07:14] <twb> Telling it to treat it was ext2 and not format it works, until it tries to mount it (which, unsurprisingly, fails)
[07:15] <twb> OK, I give up, doing the proper lucid install now
[07:20] <twb> Re ntp, *lucid's* d-i *does* prefer the DHCP-supplied one, so that is a clear regression
[07:20] <twb> Also lucid successfully recognizes /dev/sda as being ext4 (not 2) formatted
[07:21] <iToast> hey
[07:21] <iToast> Can someone help me
[07:21] <iToast> Im stuck and i caznt continue unless someone helps :/
[07:22] <SpamapS> iToast: best to just ask the question.. it may be a while until somebody can answer
[07:22] <iToast> Im installing ubuntu server 10.4.3
[07:22] <iToast> Its acting for my cdrom driver to continue, im installing from a usb
[07:22] <iToast> what would I put?
[07:22] <starlocke|soho> ... have you tried turning it off and on again?
[07:22] <iToast> starlocke|soho, why -_-"
[07:22] <iToast> It changes NOTHING
[07:23] <starlocke|soho> standard knee-jerk IT response? sorry XD
[07:23] <iToast> starlocke|soho, BOO!
[07:23] <iToast> Your a IT not the sys admin ;)
[07:23] <starlocke|soho> lol
[07:24] <iToast> sysadmin <3
[07:24] <iToast> http://xkcd.com/705/
[07:24] <starlocke|soho> lol
[07:25] <iToast> sysadmin in a nutshell:
[07:25] <iToast> Servers got hit with atom bomb?
[07:25] <SpamapS> iToast: perhaps your usb key is corrupted...
[07:25] <iToast> 10 seconds of downtime?!
[07:25] <iToast> UNACCEPTABLE!
[07:25] <iToast> SpamapS, yes, thats why it boots...
[07:26] <iToast> Im using unetbootin
[07:30] <iToast> does anyone have any idea
[07:30]  * smb wonders whether iToast has the usb key in the blue or black usb socket...
[07:30] <iToast> smb, that made no sense...
[07:30] <twb> smb: good question
[07:30] <iToast> This machine has no usb 3.0 ports, it was released WAY before usb 3.0 really went out there..
[07:31] <smb> iToast, It does when knowing one is usb2 and the other usb3 (usually)
[07:31] <iToast> smb, Also, usb3 ports are compatible with usb2 devices...
[07:31] <iToast> The extra pins are in the back of the port, they dont interfear
[07:32] <smb> iToast, But in fact when in an usb3 port a usb3 driver module is requested and that is not in the installer. We just had a case like that
[07:32] <iToast> O.o
[07:32] <smb> Hence the question
[07:32] <iToast> Thats, insane..
[07:32] <iToast> Anyways, what do i type?
[07:32] <smb> iToast, If you say you got no usb3 then this does not matter
[07:32] <iToast> :P
[07:32] <twb> The blue ports include high-speed failure
[07:33] <iToast> :P
[07:33] <iToast> Usb 3.0 sounds good, but its slow crap compared to thunderbolt :/
[07:33] <iToast> It doesn't do the same things thunderbolt does
[07:33]  * iToast is waiting for the pci-e thunderbolt card for pcs
[07:33] <twb> Thunderbolt is crap, it still uses electrons
[07:33] <smb> I'd check (alt-f2 or so) whether the usb stick seems to be visible (/proc/partitions)
[07:34] <iToast> Thunterbolt is better then usb3.0
[07:35] <iToast> Its not cheap or as widly supported, or universal...
[07:35] <SpamapS> thunderbolt is better than twinkies
[07:35] <twb> Is thunderbolt peer to peer or master/slave?
[07:35] <iToast> I think both
[07:35] <twb> Does thunderbolt require stupid polling protocols like USB?
[07:35] <iToast> it lets you daisychain everything
[07:35] <twb> Firewire got both of those correct
[07:35] <iToast> usb has one fatal flaw.
[07:36] <iToast> If the usb controller is busy, well, it wont be responsive..
[07:36] <iToast> Usb transfers will be slow / stop suddently.. mice and keyboards will lag / stop working
[07:36] <twb> iToast: guess what, that happens when any controller is busy
[07:37] <iToast> twb, get that flaw on a ps/2 port ;)
[07:37] <twb> ps/2 can't be flooded; it has fixed data rate guarantees
[07:37] <iToast> :p
[07:37] <iToast> Doesnt thunderbolt fix that thaw?
[07:37] <iToast> flaw*
[07:37] <twb> You could get the same effect on USB by simply banning all full-speed and higher devices and limiting the controller to a single device
[07:38] <iToast> Does the thunderbolt controller ever get that problem?
[07:38] <twb> Who cares
[07:39] <iToast> Seriosly, its daisy channed so that would be a problem..
[07:39] <twb> It's still using electrons
[07:39] <iToast> So?
[07:39] <iToast> its cheaper that way
[07:39] <iToast> It has a big advantage too, the devices can be powered over thunderbolt...
[07:39] <twb> electrons are fundamentally slower than c
[07:40] <iToast> I thought of something, fiber optics are flaw'd in a  way
[07:40]  * SpamapS ponders why they'd be slower than c ...
[07:40] <iToast> Turning on and off fast won't be good eneugh. switch colors for different packets?
[07:40] <twb> SpamapS: because they have mass?
[07:40] <iToast> So lets say we have 4 users downloading 8 tb of data over 1 line.
[07:40] <SpamapS> I suppose just being attracted by all the conductors, sucked into those "holes" ... makes them take a non-direct route
[07:40]  * NeoNetNinja lmfao!
[07:41] <iToast> ITs basicly computer to computer, it would be limited by a computer taking to long, thats fixable by skipping that pc until its ready
[07:41] <iToast> But why not transmit with different colors per machine
[07:41] <SpamapS> twb: photons have mass
[07:41] <iToast> SpamapS, twb Lets not get into physics...
[07:41] <Japje> iToast: http://bit.ly/wKux1a
[07:41] <twb> Pfft, that is only a recent theory, it will prove to be a fad in time :P
[07:43] <iToast> Japje, Basically lights turning on and off fast..
[07:43] <iToast> Thicker wires transmit with different spectrums of light..?
[07:43] <iToast> Allowing more comunication per line
[07:43] <iToast> You guys, if you wan't to talk about physics, i got something for you.
[07:44] <SpamapS> iToast: full spectrum fiber optics is already the norm
[07:44] <iToast> Lets take a coper wire, a thicker one would have lower resistance correct?
[07:44] <iToast> SpamapS, correct?
[07:44] <Japje> no
[07:45] <iToast> Assuming its free of impurities
[07:45] <SpamapS> iToast: no, I don't believe thats true. The thickness is not as important as the number of copper atoms between the two points of differential that make up a voltage.
[07:45] <iToast> SpamapS, I was simplyfying it cuz its late here.
[07:45] <iToast> Assuming theres more room for a electrom to move from end to end, theres lest resistance right?
[07:45] <iToast> electron*
[07:45] <Japje> if you do that on these subjects you create wrong information
[07:46] <SpamapS> yes but its a complicated formula to figure out how much less resistance. :p
[07:46] <iToast> Lets just assume its lowered resistance...
[07:46] <SpamapS> lets just assume its cats instead of copper wire
[07:46] <iToast> A thinner wire would have higher resistance..
[07:47] <Japje> assume makes an ASS of U and ME
[07:47] <iToast> ...
[07:47] <iToast> Ok lets think of it like this
[07:47] <iToast> A thicker wire has more paths for electrons to move and a tinner one doesn't
[07:47] <SpamapS> Its true, yes. :)
[07:47] <twb> !u
[07:47] <iToast> By making things like cpu's smaller and thinning hte paths for electrons to go across, wouldn't that negate efficiency?
[07:48] <twb> Flippancy is not an excuse
[07:48] <iToast> the*
[07:48] <iToast> By making it smaller there is a shorter path to go across and it allows for higher speeds easyer, but comes with highered resistance and less surface to dissipate heat.
[07:49] <SpamapS> iToast: the resistance has most certainly gone up, but the amount of heat dissapation has gone down.
[07:49] <twb> SpamapS: btw have you noticed IP over Avian Carriers latest RFC shows a substantially higher increase in throughput compared to conventional carriers?
[07:49] <iToast> So by design a cpu is flawed
[07:49] <SpamapS> twb: I think its due to global warming
[07:49] <iToast> Cpu's by design are f*cked.
[07:49] <iToast> Smaller = Higher heat lower efficieny.
[07:49] <SpamapS> iToast: thats completely backwards
[07:49] <SpamapS> you're assuming the voltage and current required to activate the transistors is constant
[07:50] <Japje> iToast: "The larger the cross-sectional area of the conductor, the more electrons are available to carry the current, so the lower the resistance. "
[07:50] <iToast> Ok
[07:50] <iToast> We have the problem though of how thin we can go before electrons wont pass through.
[07:50] <SpamapS> Probably a few atoms.
[07:50] <iToast> Sure, we have nano tubes being tested by ibm...
[07:50] <Japje> SpamapS: id say 2
[07:51] <iToast> SpamapS, By design a cpu is flawed basically
[07:51] <iToast> You have a smaller area every generation to dissipate heat
[07:51] <SpamapS> flaw/flô/
[07:51] <SpamapS> Noun:	
[07:51] <SpamapS> A mark, fault, or other imperfection that mars a substance or object.
[07:51] <iToast> Im not sure if im correct on this, but in things like silicon and copper, heat = highered resistance?
[07:51] <SpamapS> Nothing is perfect
[07:52] <iToast> SpamapS, never said anything is
[07:52] <twb> SpamapS: math is
[07:53] <Japje> are you sure? :P
[07:53] <iToast> Can't we bypass the flaws in cpu's today easly?
[07:54] <iToast> Just lock the size of a cpu for high thermal transfer and lower resistance
[07:54] <iToast> Instead of getting smaller and smaller every revision, get a tiny bit bigger, by nm allowing more thermal transfer area
[07:54] <SpamapS> twb: can you please put math in an envelope and mail it to me? I want to gaze upon its wonder ;)
[07:55] <twb> Yes
[07:55] <SpamapS> iToast: bzzt, sorry, you've reached your crazy idea quota for the day. ;)
[07:55] <iToast> How is it crazy?
[07:55] <twb> The nice thing about math is you can express it in meatspace but it stays, itself, unsullied
[07:56] <Japje> i like meatspace
[07:56] <iToast> Im curios if this is a good idea, its already implemented in macs
[07:57] <iToast> Keep all ram constantly reserved by the system so its always ready for a application to launch
[07:57] <SpamapS> macs use intel CPU's
[07:57] <iToast> Im using a amd...
[07:57] <iToast> I still use intel, but for performance and price, amd just seems better
[07:58] <iToast> Ok, so no one hates on me
[07:58] <iToast> Lets say core i3 and athlon II are on level grounds, or should i choose a diff amd.
[07:59] <Japje> diff architecture
[07:59] <iToast> Lets compare ghz/dollar
[07:59] <iToast> http://www.newegg.com/Product/Product.aspx?Item=N82E16819115077
[07:59] <Japje> couldnt we compare ghz/pound
[07:59] <Japje> dollar isnt worth anything
[08:00] <iToast> 39.390 ghz/dollar for a i3.
[08:00] <iToast> Now lets choose a amd cpu
[08:00] <Japje> lets not
[08:00] <iToast> Ok off topic to this
[08:00] <iToast> http://www.newegg.com/Product/Product.aspx?Item=N82E16819103961
[08:01] <iToast> WHO NEEDS 8 CORES?!
[08:01] <twb> Japje: itym /watt
[08:01] <iToast> MY GOD! WHY?!
[08:01] <iToast> WHY 8 CORES?!
[08:01] <Japje> i do
[08:01] <iToast> why?
[08:01] <iToast> You better be rendering 3d hd video in blender...
[08:01] <Japje> how about virtualisation
[08:02] <iToast> What os
[08:02] <iToast> (guest and host)
[08:02] <Japje> esxi
[08:02] <Japje> or kvm
[08:02] <Japje> or any other hypervisor
[08:02] <SpamapS> iToast: I've made use of a box w/ 40 cores before.. :)
[08:02] <iToast> SpamapS, Thats a server mobo.
[08:02] <SpamapS> hadoop ftw. :)
[08:02] <iToast> That doesn't count!
[08:02] <iToast> Servers do not count, they need the cores.
[08:02] <Japje> ofcourse it does
[08:03] <iToast> My webserver is 1core... at 1.5ghz...
[08:03] <Japje> lets assume we want a server
[08:03] <iToast> Japje, you want 80 cores :D
[08:03] <iToast> 128tb of ram
[08:03] <Japje> only 80?
[08:03] <iToast> yea why?
[08:03] <iToast> were trying not to make it burst into flames...
[08:03] <Japje> i currently use more
[08:03] <iToast> 800tb of storage
[08:03] <Japje> also not enough
[08:03] <iToast> and 10 1k /watt psus for redundancy
[08:03] <iToast> Japje, ?!
[08:04] <iToast> Are you rendering uncompresed video from a red epic?!
[08:04] <Japje> i currently have about 2.2 Pb worth of data
[08:04] <Japje> about 400 cores
[08:04] <iToast> cluster?
[08:04] <Japje> you could call it that
[08:04] <Japje> lets call it... cloud
[08:04] <Japje> that way i can use buzzwords
[08:05] <iToast> I wan't to cluster my ubuntu machine to work for my windows machine...
[08:05] <iToast> So when vegas is rendering hd video, instead of up to 2 hours for 30mins of video, 1 / half a hour...
[08:05] <iToast> (has about 14 computers)
[08:06] <Japje> i want to cluster my cat and my dog
[08:06] <SpamapS> Japje: give the cat catnip and the dog an ubuntu CD... they'll cluster all day
[08:06] <Japje> thats a good idea
[08:07] <Japje> but its a Dog Air
[08:07] <Japje> it doesnt have a cd slot
[08:07] <iToast> Plug a external dvd rom into the usb port under the tail
[08:08] <twb> The more space you give users, the more content you have to check for badthinks and steganography
[08:08] <iToast> twb, Easy idea...
[08:08] <iToast> Don't let them make it public
[08:10] <Japje> tell that to the megaupload users
[08:10] <iToast> :D!
[08:10] <iToast> vim e_e
[08:10] <iToast> Nano <3
[08:11] <SpamapS> pipemeter!
[08:11] <Japje> sed!
[08:11] <iToast> How to secure you're server:
[08:11] <iToast> Step 1: Turn it off
[08:11] <iToast> Step 2: lock it away in a closet
[08:11] <iToast> Step 3: tell no one it exists
[08:12] <iToast> Step 4: hope no one finds it...
[08:12] <twb> Japje: I babysit prison servers, what I said is an actual issue there
[08:12]  * RoyK slaps iToast 
[08:12]  * iToast slaps RoyK with a smelly fish
[08:12] <Japje> i know :)
[08:12] <Japje> wasnt doubting it
[08:12] <twb> iToast reminds me of the electronic toaster from Red Dwarf
[08:12] <iToast> twb, what do you mean?
[08:12] <Japje> my users use space for database storage
[08:12] <Japje> most anyway
[08:13]  * RoyK sends iToast to http://www.youtube.com/watch?v=IhJQp-q1Y1s
[08:13] <iToast> No, i ment babysiting prison servers
[08:13] <Japje> how can you not understand that?
[08:13] <iToast> What's put on those servers
[08:13] <twb> 128MiB is enough for anyone
[08:14] <Japje> i kindly refuse
[08:14] <_ruben> hrm .. apparently ubuntu(-server)/linux/whatever doesn't like traffic originating from the network address .. bugger that
[08:14] <iToast> I agressivly refuse.
[08:15] <Japje> _ruben: ubuntu is not a network aware operating system
[08:15] <twb> _ruben: which address?
[08:15] <iToast> Japje, Letme get this straight
[08:15] <twb> _ruben: maybe you forgot to purge NM with extreme prejudice
[08:16] <iToast> By a network aware os, does that mean computers discover eachother and their open services
[08:16] <_ruben> twb: .0 in a /24
[08:16] <Japje> no, its an OS that has given network access by chuck norris
[08:16] <twb> ha
[08:16] <Japje> .0 is not chuck norris approved
[08:16] <iToast> Japje, out of all seriousness was that right?
[08:17] <twb> This channel has turned into a clown house
[08:17] <_ruben> i can use /31s just fine for ptp, but arp fails when it's sourced from .0/24
[08:17] <Japje> iToast: as for the last hour, your wrong
[08:17] <twb> It's not even september
[08:17] <iToast> Japje, Explain network aware os ;)
[08:17] <twb> _ruben: "doesn't like" as in it drops it on the floor?
[08:18] <Japje> its the time difference that makes me aggitated
[08:18] <Japje> that or the snow here in bucharest
[08:18] <_ruben> twb: seems to ignore it yes, and arping to .0 gets permission denied ?!
[08:19] <iToast> Oh, ugm, this is the stupidiest question you will hear.
[08:20] <iToast> Was the reason my samba install not working because i forgot to set the domain to WORKGROUP
[08:20] <iToast> Out of all honesty, i've used windows network shares and freenas, freenas automated samba for me...
[08:23] <iToast> Il stop being stupid...
[08:23] <iToast> e_e
[08:24] <iToast> ...
[08:24] <iToast> Now for my original question
[08:24] <iToast> What do i do for cdrom
[08:24] <twb> _ruben: are you root?
[08:25] <iToast> ...
[08:25] <iToast> Does anyone know?
[08:31] <_ruben> twb: yes
[08:31] <_ruben> for now i just bumped the firewall from .0 to .1
[08:31] <_ruben> will postpone in-depth research to later date
[08:36] <twb> You know .0 is not a normal address right?
[08:36] <twb> It's not a host address it's a network address
[08:37] <_ruben> yes, but most network stacks are "liberal" enough to treat it as a "normal" address, with cidr involved and all
[08:38] <onre> twb, it can be a normal address in any network with "wider" than 24-bit netmask.
[08:39] <twb> onre: granted
[08:39] <twb> _ruben: that sounds like a dirty hack
[08:39] <twb> the same way NAT is a dirty hack :P
[08:44] <_ruben> the network address should be "less special" than for instance the broadcast address, which atleast has *some* functionality :)
[08:50] <twb> _ruben: HTFU princess
[08:55] <RoyK> twb: 192.168.1.0 is a host address on our 192.168.0.0/21 network...
[09:14] <_Techie_> is there any downside to using 32bit ubuntu server with 8gigs of RAM and an i3-2100?
[09:15] <linocisco> hello
[09:16] <linocisco> where to find apt.conf in android?
[09:16] <_Techie_> linocisco: i dont believe android uses apt
[09:17] <linocisco> _Techie_: it is a big problem for me as I could not authenticate to use office network
[09:18] <linocisco> which rooting software is best for all android tablets and phones
[09:18] <linocisco> ?
[09:18] <_Techie_> linocisco: have you tried asking in #android
[09:18] <linocisco> i can't connect or join it
[09:19] <_Techie_> linocisco: i have a feeling that its restricted to registered users, you will need to register with nickserv
[09:19] <SpamapS> _Techie_: yes, memory addressing will incurr a performance penalty
[09:19] <linocisco> yes. I keep trying
[09:20] <_Techie_> SpamapS: how much of a performance penalty?, i will be using it mainly as a personal game server. team fortress 2, garry's mod, some teamspeak... and yes... minecraft
[09:22] <SpamapS> _Techie_: varies by CPU
[09:23] <SpamapS> _Techie_: http://www.phoronix.com/scan.php?page=article&item=ubuntu_32_pae&num=1
[09:24] <_Techie_> SpamapS: thankyou, i was just abotu to ask if you had any links showing the performace difference in different situations
[09:25] <SpamapS> _Techie_: notice how much the 64-bit system kills the 32-bit systems... if you have 8GB of RAM and a capable system.. there's very little reason not to do it.
[09:28] <_Techie_> SpamapS: looks like my choice is clear, ill start prepping a hard drive image to implement this weekend
[09:28] <_Techie_> SpamapS: thanks alot for your help
[09:30] <lambda_x> does anyone know how to remove faulty paths from multipath?
[09:55] <SpamapS> jamespage: you wouldn't happen to be around would you?
[09:58] <iToast> ...
[10:01]  * SpamapS fights the urge to go run across the street to 7-11 and buy a redbull + twinkie and keep working through till sun-up
[10:04] <jamespage> SpamapS, I am now
[10:07] <iToast> ops >_>
[10:08] <Tm_T> hi iToast, welcome to the Ubuntu Server discussion and support channel #ubuntu-server
[10:08] <SpamapS> jamespage: trying to discern how the documents in couchdb are populated from the server-tests-precise thing..
[10:09] <jamespage> SpamapS, well run-test.py puts most stuff into couchdb
[10:09] <jamespage> thats in the ubuntu-server-iso-testing codebase
[10:10] <SpamapS> right I think I lost that bit :-P
[10:10] <SpamapS> somewhere in the hangover/jet lag induced haze
[10:10] <jamespage> I drew a picture once lemme see if I can find it
[10:10] <SpamapS> jamespage: I've got run_tests in a state where I can make it reboot and run post-reboot tests..
[10:11] <iToast> Tm_T, hi
[10:11] <jamespage> SpamapS, nice; want me to review and test as well?
[10:12] <SpamapS> jamespage: I thinK I understand how it works, but I don't understand how to make sure it will work... I can make a manually booted kvm vm reboot and run tests and stuff.
[10:13] <SpamapS> jamespage: right now the way I'm doing it is storing the service state of before/after reboot in couchdb, which works quite nicely, but means that run_test itself has to push the result of that into the "results" ...
[10:13] <SpamapS> jamespage: I think I should probably change that to have it serialize those results back onto disk and let a post-reboot test pick it up
[10:14] <SpamapS> jamespage: will perhaps ask for a review late today or early tomorrow.
[10:17] <Daviey> jamespage: for the ec2 tests, do we save the euca-get-console-ouput log ?
[10:17] <jamespage> Daviey, yes
[10:17] <m_tadeu> hi...I'm trying to use mysqldbexport, but I'm getting the following error "ImportError: No module named mysql.utilities". How can I solve this?
[10:18] <Daviey> jamespage: i think i am being daft, where is it stored?
[10:18] <Daviey> jamespage: looking into https://jenkins.qa.ubuntu.com/view/Precise/job/precise-server-ec2/ARCH=i386,REGION=eu-west-1,STORAGE=ebs,TEST=cloud-config,label=ubuntu-server-ec2-testing/2/
[10:19] <jamespage> Daviey: well it tried to get a console output - but was not able to
[10:19] <jamespage> the instance never actually booted
[10:19] <jamespage> ec2 error
[10:20] <Daviey> jamespage: Did the reservation go awol?
[10:20] <jamespage> Daviey: nope it just crapped out
[10:20] <Daviey> jamespage: Hmm, the console-log suggests it couldn't be terminated..
[10:20] <Daviey> Does thta mean it went directly to terminated by AWS?
[10:21] <jamespage> Daviey: yes - I'm just going to see if its still running
[10:21] <Daviey> cool
[10:21] <jamespage> Daviey: possible - it not present in the instances listing for that region on that account
[10:22] <Daviey> ok, thanks
[10:23] <Daviey> jamespage: Do you know why cloud-config is unstable?
[10:23] <jamespage> Daviey: yes
[10:23] <jamespage> the test needs updating - its nothing to worry about for the a1 images
[10:24] <jamespage> someone did half a job just after alpha1
[10:24] <Daviey> jamespage: great, thanks
[10:24] <Daviey> heh
[10:24] <Daviey> jamespage: Does the account need to be able to run more concurrent instances?
[10:24] <Daviey> >InstanceLimitExceeded</Code><Message>Your quota allows for 8 more running instance(s). You requested at least 10</Message>
[10:25] <jamespage> Daviey: yes - I thought that had been done but its possible I got confused about which region
[10:25] <jamespage> you will notice that we are not running tests in the new sa region
[10:25] <jamespage> that def has been done
[10:28] <Daviey> groovy
[10:29] <jamespage> Daviey: I just requested that increase and asked them to validate its good in all regions.
[10:29] <jamespage> do you want me to fix the test; add the new sa region and run again once they confirm?
[10:31] <Daviey> jamespage: that would seem like a dandy idea.
[11:45] <jamespage> Daviey: OK - just waiting on amazon to confirm the instance limit increases....
[11:46] <Daviey> rocking
[12:24] <rbasak> jamespage: I have one test for a precise panda netinst on armhf itself, and another to test juju in a local environment. This is still in progress somewhat but ready to start getting integrated now I think
[12:25] <jamespage> rbasak, whats your central point of control?
[12:25] <rbasak> what I'd like to do is run a jenkins slave for these, so we can start automatically detecting breakages for these
[12:25] <rbasak> I have a "server" here that runs cobbler and my scripts, and I imagine will be running a jenkins slave
[12:26] <jamespage> rbasak, you will need to run jenkins itself; until this is in the QA lab and it can actually access a master instance you can just run everything from jenkins locally
[12:26] <rbasak> In time we'll be adding more tests to this mechanism. There's also talk of testing in qemu
[12:26] <jamespage> rbasak, so sudo apt-get install jenkins is a good start :-)
[12:26] <rbasak> I see - so we can't really hook it up to the existing master instance without it being in the QA lab?
[12:26] <jamespage> assume you are running oneiric or precise on this install
[12:26] <rbasak> I was hoping to give QA visibility of this
[12:27] <rbasak> I'm running oneiric but could run precise
[12:27] <jamespage> rbasak, we can still push results to jenkins.qa.ubuntu.com
[12:27] <rbasak> Cool
[12:27] <jamespage> thats just a read only copy of stuff from another 3 jenkins instances
[12:27] <jamespage> NOT slaves
[12:27] <rbasak> I see, OK
[12:27] <jamespage> rbasak, of course we need to get the nod from QA before starting to push results
[12:27] <rbasak> sure
[12:28] <jamespage> oneiric is fine
[12:28] <jamespage> its a slightly older version but is compatible with that in the lab
[12:28] <jamespage> I will be backporting the latest LTS to oneiric next week
[12:28] <rbasak> I've never touched jenkins before. Is there anything I need to know? Any particular way I should be setting it up?
[12:29] <jamespage> so if you need a particular feature
[12:29] <jamespage> ...
[12:29] <jamespage> so a couple of things you need to know
[12:29] <jamespage> 1) everything runs under the jenkins account by default; this is a good place to install tools (we normally user ${HOME}/tools/XXX-tool
[12:29] <jamespage> which jenkins jobs are going to use
[12:30] <jamespage> 2) get the naming right; if you look at jenkins.qa.ubuntu.com for guidance
[12:30] <jamespage> e.g. precise-arm-XXX-test of something similar
[12:30] <jamespage> precise-server-armel-XXX maybe
[12:30] <jamespage> anyway have a think
[12:31] <jamespage> its worth getting right - makes setting up dashboards using regex easier
[12:31] <jamespage> OH - and jenkins starts on port 8080 by default BTW and is not secured in any way to start with
[12:32] <rbasak> OK, thanks!
[12:32] <jamespage> I try to limit whats actually in a Jenkins job definition to a bare minimum - that way the tools can be run by anyone without jenkins
[12:32] <jamespage> just enough to execute the test and determine the results
[12:32] <jamespage> non zero exit code = job failure = RED ball
[12:33] <jamespage> rbasak, how are you running tests?
[12:33] <rbasak> Right now, just a script with an exit status
[12:33] <jamespage> i.e. can the test results be parsed easily into something Jenkins can understand?
[12:33] <rbasak> I presume I'll be needing to change script output to be more useful to jenkins
[12:33] <jamespage> it is possible to search for specific strings in the log
[12:34] <jamespage> but outputting something into JUnit  xml format means Jenkins can do more...
[12:35] <rbasak> OK, so to start with I can just use what I have, and then change the format later?
[12:43] <jamespage> rbasak, yes
[12:44] <rbasak> great!
[13:22] <roaksoax> smoser: howdy! So I was checking youtr homeserver and really don;t see any issues
[13:22] <roaksoax> smoser: as according to /proc/cpuinfo you have only 1 core, with 1.0Ghz
[13:51] <mdeslaur> SpamapS: FYI, please don't merge php5 from debian
[14:17] <jamespage> Daviey: I just kicked off the precise-server-ec2 testing again - limits should now be OK on all regions.
[14:17] <jamespage> utlemming ^^ sa-east-1 also being tested now on all test types for ec2
[14:19] <roaksoax> jamespage: btw.. are you still using the lab with automatic power managent on the sentry switch?
[14:23] <Daviey> jamespage: Is it difficult to do selective tests?
[14:23] <Daviey> ie, test those that failed due to that error?
[14:25] <smoser> roaksoax, well, its definitely not just 1.0Ghz
[14:25] <smoser> it does scale
[14:56] <autif> I have used xubuntu desktop for a long while and for the first time set up an ubuntu server machine for long term use - my question is how do I apply updates - just apt-get update && apt-get upgrade? If yes, then I have never seen the kernel change on the server - is that intentional or there is an apt command that I am missing - I am on 11.10
[15:04] <smw> autif, you need to install them manually
[15:05] <smw> it tells you that the kernel packages are "held back"?
[15:08] <eikke> did anyone ever use ubuntu orchestra/cobbler/koan to provision virtual machines on a server other than the cobbler host?
[15:10] <mtaylor> eikke: if you want to do that, wouldn't something like nova make more sense?
[15:11] <eikke> mtaylor: not sure, feels heavyweight for a jenkins/ci environment using VMs as work horses
[15:13] <mtaylor> eikke: ah. yeah
[15:15] <jamespage> eikke, I've not done it myself but its definitely possible - I think jibel may be  looking at this for further automation of Ubunt
[15:15] <jamespage> testing
[15:16] <eikke> I got orchestra/cobbler running for physical machine provisioning, yet running koan on another mahcine does find the cobbler install and can retrieve info about the VM I want to create (it's a System in orchestra/cobbler)
[15:16] <jamespage> Daviey: well it is possible - there is an adhoc job we can use for that
[15:17] <eikke> but then starts to retrieve some files from the cobbler host, and gets 404, and the files it attempts to retrieve look like RedHat-bases systems to me
[15:18] <Daviey> jamespage: just wondering if it's worth doing a full retest on things we know work?
[15:18] <jamespage> eikke: have you seen this - https://help.ubuntu.com/community/Cobbler/Deployment
[15:18] <jamespage> ?
[15:19] <jamespage> Daviey: if there was a nice easy way todo that yes I agree
[15:19] <jamespage> its a bit of a hack at the moment which makes it hard to gain a full set of results in one place
[15:20] <eikke> jamespage: that's what I'm attempting to do, but fails (the last command)
[15:20] <jamespage> I did toy with seperating all of the jobs into individual jobs which would mean you could just re-run one of them more easily
[15:20] <Daviey> jamespage: hehe
[15:20] <Daviey> jamespage: makes sense.
[15:20] <jamespage> Daviey - there might be a plugin - I will take a look
[15:21] <jamespage> eikke: please can you pastebin the error code/stack trace you are getting
[15:21] <Daviey> jamespage: meh, just retrigger it right now - but useful for the future i think
[15:22] <jamespage> Daviey: ack - I already did
[15:22] <jamespage> results should publish in the next few minutes
[15:23] <Daviey> heh
[15:26] <eikke> jamespage: https://gist.github.com/215475eba8817899212b
[15:26] <eikke> (thats using --profile, using --system I get the same result)
[15:26] <dax_roc> Afternoon all
[15:27] <dkn> howdy
[15:27] <dax_roc> Is the correct way to remount a fs rw after an error "mount -o remount,rw / " ?
[15:28] <dkn> google says mount -remount,rw /mnt
[15:29] <dkn> the -oremount looks like read only
[15:30] <eikke> mount -o remount,rw / should do the job
[15:31] <dax_roc> no doesn't seem to, Fails with "mount: cannot remount block device /dev/mapper/ubuntu--base-root read-write, is write-protected"
[15:33] <eikke> jamespage: any clue? the repo_mirror folder was empty initially, so I enabled it to reposync, ran reposync (and pulled in several GB of data), yet no reveal
[15:34] <autif> smw - thanks for the info, can you please also point me to how to go about installing the latest kernel manually? (apologies for the late reply, had to run to a meeting)
[15:34] <jamespage> eikke, just trying it out :-)
[15:35] <smw> autif, I just install manually
[15:35] <smw> autif, sudo apt-get install
[15:35] <smw> autif, there is also aptitude safe-upgrade
[15:35] <autif> ah
[15:35] <eikke> jamespage: ah, thanks!
[15:36] <autif> in that case, I will need to search for the latest kernel - presumable using apt-cache search
[15:36] <autif> right?
[15:37] <dkn> dax_roc, did you try specifying the paths? mount -o remount,rw /dev/foo /dir ?
[15:38] <dax_roc> dkn: just the mount point not the device
[15:39] <autif> smw - also - how are you notified when a new kernel is available? short of running apt-cache search linux-image.+server everyday
[15:39] <smw> autif, I don't know
[15:40] <dkn> the man page for mount remount shows both, you might try /dev/ and /dir if just /dir isn't working.... not an expert on it though
[15:40] <autif> thanks for the help!
[15:40] <autif> will research
[15:41] <patdk-wk> autif, personally, I just subscribe to the security alerts mail list
[15:41] <dax_roc> dkn: it should work with the mount point, I'll try force
[15:47] <eikke> jamespage: I never ran 'cobbler import' though...
[15:47] <jamespage> eikke: ah - well you need to get some basic information into cobbler otherwise it won't do much
[15:48] <jamespage> I did get a little further - appeared to try and create a kvm instance - however not sure it worked.
[15:48] <eikke> jamespage: well... I installed the ubuntu-orchestrator-server package, configured according to my needs
[15:49] <eikke> then created an 'oneiric-x86_64' system using cobbler_web providing a mac address, and was able to provision a physical server through pxe this way
[15:49] <eikke> there's other distros available in the (default) list as well, I didnt change anything over there
[15:50] <jamespage> hmm - well it should be OK then
[15:53] <jamespage> eikke, I think its your system - can you do sudo cobbler system dumpvars --name oneiric-x86_64
[15:53] <jamespage> pastebinit
[15:56] <eikke> https://gist.github.com/0b8a68f11fe6e70614b0
[16:03] <jamespage> eikke: sorry not being much help am I :-(
[16:03] <eikke> jamespage: all help is welcome ;-)
[16:09] <jamespage> eikke: using the standard oneiric-x86_64 profile I get a Cannot find install source in kickstart file, aborting. message
[16:10] <jamespage> using a profile I use on hardware everyday I get http://paste.ubuntu.com/826540/
[16:11] <jamespage> domains are running
[16:12] <roaksoax> jamespage: could ytou also pastebin your "cannot find install source in kickstart file"
[16:13] <jamespage> roaksoax: I think its the standard one from oneiric - I'll check
[16:13] <roaksoax> jamespage: jamespage ah your running cobbler/koan in oneiric?
[16:14] <jamespage> roaksoax, ah - my bad - neither server or profile as and associated kickstart!
[16:14] <roaksoax> jamespage: cause in precise this should have been fixed
[16:14] <jamespage> confusingly I'm using koan on precise and cobbler on oneiric
[16:14] <roaksoax> jamespage: ah :)! but yeah, there were koan/virtinst issues in precise related to not being able to detect an ubuntu mirror, which are already fixed
[16:15] <jamespage> gah eikke has gone - I suspect thats his issue.
[16:15] <roaksoax> jamespage: most likely
[16:15] <roaksoax> jamespage: btw.. are you still using the fence_cdu script for the lab?
[16:16] <jamespage> yep
[16:16] <roaksoax> jamespage: i'm gonna integrate it later today in cobbler/fence-agents so if you see any breakeage you know why it might be xD
[16:16] <eikke> jamespage: sorry, laptop died on me, hope I didnt miss any useful info :)
[16:18] <jamespage> eikke: roaksoax said that there where know koan/virtinstall issues detecting ubuntu mirrors
[16:18] <eikke> right
[16:18] <eikke> the problem I have is I don't know whether koan is looking for files it shouldnt look for, *or* cobbler isnt serving files it should
[16:19] <roaksoax> eikke: could you pastebin your koan output
[16:19] <roaksoax> please?
[16:20] <eikke> roaksoax: https://gist.github.com/215475eba8817899212b
[16:21] <roaksoax> eikke: is this in precise or oneiric?
[16:21] <eikke> both master and slave are oneiric x86_64
[16:21] <roaksoax> eikke: is it possible you could try installing virtinst for precise?
[16:21] <roaksoax> err precise's virtinst into oneiric?
[16:21] <eikke> on the slave? sure
[16:22] <eikke> if I can figure out how to mix'n'match those things :$
[16:22] <roaksoax> eikke: yeah on the machine you are running koan
[16:25]  * roaksoax bbl
[16:25] <eikke> roaksoax: should I do so using download + dpkg?
[16:29] <eikke> roaksoax: using 0.600.0-1ubuntu4, same error
[16:30] <roaksoax> eikke: uhmm interesting
[16:30] <eikke> didnt update koan, only virtinst
[16:31] <roaksoax> eikke: you shouldn't really need to update koan
[16:31] <eikke> I'll try to "cobbler import" an oneiric server ISO
[16:31] <roaksoax> eikke: maybe becuaes installing with dpkg didn't really upgrade the python modules
[16:33] <eikke> it should though, they're part of the package (http://packages.ubuntu.com/precise/all/virtinst/filelist)
[16:34] <koolhead17> hi all
[16:35] <roaksoax> eikke: ok found the issue. you need to update koan from precise too
[16:35] <eikke> will do
[16:36] <roaksoax> eikke: or
[16:36] <roaksoax> eikke: or
[16:37] <eikke> ? about to install python-koan and koan from precise :)
[16:38] <roaksoax> eikke: or sudo cobbler profile edit --name precise-x86_64 --ksmeta tree=http://@@http_server@@/cblr/ks_mirror/precise-x86_64
[16:39] <roaksoax> bug #807324
[16:41] <jamespage> Daviey: https://jenkins.qa.ubuntu.com/view/Precise/job/precise-server-ec2/
[16:42] <eikke> roaksoax: changing the oneiric ksmeta like that works, w00t
[16:43] <roaksoax> eikke: cool then
[16:43] <roaksoax> eikke: in the precise's koan that change is done automatically
[16:43] <eikke> interesting
[16:44] <eikke> thanks for the help, really!
[16:44] <roaksoax> eikke: no worries ;)
[16:44] <Daviey> jamespage: \o/
[16:44] <smoser> jamespage, we really need tomake console.txt wait before collecting
[16:44] <smoser> its pointless to collect an almost guaranteteeed empty file
[16:44] <mjt> smoser: you around?  Do you remember why ipxe package were split into kvm-ipxe and ipxe?  And maybe, do you know how the roms in ipxe package are being actually used, besides by qemu and the like?
[16:44] <mjt> heh
[16:45] <jamespage> Daviey: https://wiki.jenkins-ci.org/display/JENKINS/Matrix+Reloaded+Plugin
[16:45] <jamespage> I'll get it installed
[16:45] <smoser> hallyn, would know. mjt
[16:45] <mjt> smoser: he sent me to you :)
[16:46] <Daviey> jamespage: ooo, that is nice
[16:47] <mjt> i'm trying to decide what to do with ipxe in debian.  To me it looks like we should drop all the boot roms except of very few ones which we actually use - by qemu, by virtualbox maybe, etc, and compile them without boot delay.  The rest - IMHO - has no use at all, since if someone wants to embed it into their actual NIC they'll compile their own anyway.
[16:47] <jamespage> smoser: if so I really need to re-implement the test framework todo more multithreading
[16:49] <smoser> jamespage, yeah. i looked at it, and decided that too
[16:49] <mjt> hallyn: do you not agree ^^ ?
[16:50] <jamespage> hmm
[16:50] <hallyn> mjt: I wouldn't object.
[16:50] <jamespage> maybe use twisted :-)
[16:50] <smoser> mjt, i would agree.
[16:50] <smoser> that boot-delay is annoying.
[16:50] <smoser> and i'm not aware of any use for this other than kvm (or as you say, compiling their own).
[16:51] <mjt> and in that case, just one package "ipxe" should be enough (which provide ipxe-grub too)
[16:52] <smoser> jamespage, https://jenkins.qa.ubuntu.com/view/Precise/job/precise-server-ec2/ARCH=i386,REGION=us-west-2,STORAGE=instance-store,TEST=cloud-config,label=ubuntu-server-ec2-testing/lastBuild/artifact/None/i386/m1.small/instance-store/i-e688ccd6/uec2-20120202-1453-d28473559d2145-terminated.console.txt
[16:53] <hallyn> lynxman: was it by chance you who was interested in ipxe roms for non-kvm?
[16:53] <hallyn> or is that person a figment of my imagination?
[16:53] <hallyn> zul: can xen use any ipxe roms?
[16:53] <jamespage> smoser: looking now
[16:53] <zul> hallyn: no idea i havent been able to try
[16:53] <jamespage> smoser: we have seen that before....
[16:53] <lynxman> hallyn: I was indeed, and I solved the issue concerning that on my new build, I now have to kill some more packaging issues before the package is life :)
[16:53] <smoser> jamespage, that failure (from https://jenkins.qa.ubuntu.com/view/Precise/job/precise-server-ec2/ARCH=i386,REGION=us-west-2,STORAGE=instance-store,TEST=cloud-config,label=ubuntu-server-ec2-testing/lastBuild/)
[16:53] <smoser> i suspect is related to https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/898373
[16:53] <smb> hallyn, I do boot xen guests with pxe if that is what you want to know
[16:54] <smoser> maybe not.
[16:54] <smoser> where else had you thought we'd seen it?
[16:54] <hallyn> smb: what i want to know is whether any of the roms in ipxe but not in ipxe-qemu are needed
[16:54] <jamespage> smoser: I think that was the bug I was thinking about
[16:54] <hallyn> lynxman: ^
[16:54] <smb> hallyn, Oh ok... more detail I used to care about then
[16:55] <lynxman> hallyn: that should be trivial to add now, so no worries about it
[16:55] <jamespage> smoser: yes - its the same thing
[16:55] <lynxman> hallyn: lp:~lynxman/ubuntu/precise/ipxe/newsnapshot
[16:55] <smoser> why do you think its the same, jamespage ?
[16:55] <smoser> ah.
[16:55] <smoser> it is.
[16:55] <smoser> duh
[16:55] <utlemming> smoser: is that a launch blocker or just a release note?
[16:55] <smoser> i was looking too far down.
[16:56] <jamespage> so looking at https://jenkins.qa.ubuntu.com/job/precise-server-ec2/ARCH=i386,REGION=us-west-1,STORAGE=instance-store,TEST=cloud-config,label=ubuntu-server-ec2-testing/1/artifact/None/i386/m1.small/instance-store/i-835a3dc4/uec2-20111130-1920-5d6ee3bc489541-terminated.console.txt
[16:56] <jamespage> which is linked from the bug
[16:56] <jamespage> the instance fails to reboot with the same errors as in todays test
[16:56] <stgraber> hallyn: working on some changes to lxc-ubuntu to default to armhf and deal with building for a non-native arch
[16:56] <hallyn> lyxman: you did an update, but i can't tell from that - do you need any of the roms which are in ipxe but not in kvm-ipxe?
[16:56] <eikke> roaksoax: virt-manager etc now work beautifully as well, thank you, really. spent all day on this :P
[16:57] <hallyn> smb: so the answer is you don't need those?
[16:57] <lynxman> hallyn: nope, it's fine as it is in the update, got all the roms you copied over to kvm-ipxe as well
[16:57] <hallyn> stgraber: ok, pls remember there are staged changes in ubuntu:lxc bzr tree
[16:57] <smb> hallyn, No, the answer is I don't know whether I need those
[16:57] <lynxman> hallyn: so your choice of roms doesn't affect me :)
[16:57] <smoser> thats freaking annoying
[16:57] <jamespage> smoser: yes
[16:57] <smoser> i click on "#3", and the link i get is https://jenkins.qa.ubuntu.com/view/Precise/job/precise-server-ec2/ARCH=i386,REGION=us-west-2,STORAGE=instance-store,TEST=cloud-config,label=ubuntu-server-ec2-testing/lastBuild/
[16:57] <hallyn> mjt: ^ so i think the answer is JFDI and we'll hear about it if any deleted roms are needed :)
[16:57] <smoser> "latestBuild"
[16:58] <smoser> i want a perminent link
[16:58] <mjt> JFDI? Just F.. Do It? :)
[16:58] <mjt> yes that's probably the best way to know
[16:59] <stgraber> hallyn: hmm, latest entry I see in the branch is 0.7.5-3ubuntu16, can't see any staged changes. Did you push?
[16:59] <smoser> its sad, mjt, but sometimes that is the best way.
[16:59]  * smoser is happy for ditching 3 seconds of boot!
[17:00] <smoser> so jamespage utlemming ...
[17:01] <smoser> if, for some reason the filesystem is dirty on /mnt (or some other filesystem), we just dont boot.
[17:01] <smoser> is that the case?
[17:01] <smoser> cause that sucks
[17:01] <utlemming> smoser: yuck
[17:02] <utlemming> smoser: can you document that in the bug and we'll put it on the release notes
[17:04] <hallyn> stgraber: feh lemme check
[17:04] <stgraber> hallyn: my changes are fairly minimal, so I can easily rebase on whatever you have staged
[17:05] <hallyn> stgraber: this wasn't major either, it was just there so i wouldn't forget.  now i forgot.
[17:06] <hallyn> stgraber: nm, that was libvirt :)
[17:06] <hallyn> fire away
[17:07] <stgraber> hallyn: good. I'll also make a separate commit cleaning up all the tabs and using 4 spaces consistently in the file :)
[17:07] <hallyn> stgraber: that'd be great.  but be careful!  :)
[17:07] <hallyn> stgraber: btw do you have an idea about why the net-device-up for lo is not being fired in lucid containers?
[17:09] <stgraber> hallyn: not really, is udev working properly (udevd, the --add and udevtrigger)?
[17:09] <hallyn> stgraber: pretty sure it is, though remember we were going to drop those anyway
[17:10] <hallyn> ok, then i guess i'll look into it more - just thought you might know offhand what changed.  (I'm sure it was working until very recnetly)
[17:13] <jamespage> smoser: it would appear that that is the case
[17:16] <stgraber> root@castiana:/# cat /etc/apt/sources.list
[17:16] <stgraber> deb http://ports.ubuntu.com/ubuntu-ports precise-updates main universe
[17:16] <stgraber> deb http://ports.ubuntu.com/ubuntu-ports precise-security main universe
[17:16] <stgraber> hallyn: ^ that looks wrong ;)
[17:16] <mjt> anyone know what NICs virtualbox emulates?
[17:16] <stgraber> hallyn: where's the regular source? :)
[17:17] <hallyn> stgraber: is that on arm?
[17:17] <stgraber> hallyn: armhf, yes, well armhf running on x86, but code path should be the same
[17:17] <hallyn> rbasak: ^
[17:17] <smoser> jamespage, discussion in -devel with slangasek is that it is working as designed and he in all his knowledge does'nt have a solution for me other than "nobootwait"
[17:18] <hallyn> stgraber: what do you mean by 'the regular source'?  you mean without -updates?  or you mean not ports.ubuntu.com?
[17:19] <stgraber> hallyn: "deb http://ports.ubuntu.com/ubuntu-ports precise main universe" is missing
[17:19] <hallyn> stgraber: oh.  that sounds like a thinko
[17:19] <stgraber> hallyn: in my case, I'll also need some i386 sources in there but that's specific to my hack and I know I need to implement that bit
[17:20] <allenap> bigjools: You were saying it's a pi... going to be very easy.
[17:20] <Daviey> stgraber / smb: Is resolveconf behaving now?
[17:20] <bigjools> allenap: cake
[17:20] <hallyn> oh.  so no containers are getting regular 'source'?  d'oh.
[17:20] <stgraber> Daviey: no, we have one more change staged in the branch to be pushed post-freeze
[17:21] <allenap> bigjools: How does maas tell other components of the whereabouts of this blob service?
[17:21] <hallyn> no, mine has it
[17:21] <stgraber> Daviey: and then will need to deal with potential bugs that may appear because of that change
[17:21] <smb> Daviey, I have not re-tested (which needs a install of oneiric and upgrade after manual reconfig) but I have not noticed a note saying it should work now
[17:21] <stgraber> hallyn: my amd64 ones seem to be fine
[17:21] <bigjools> allenap: it's only for juju
[17:21] <Daviey> stgraber: do you have concerns?
[17:21] <mrevell> allenap, Do you have a moment to talk hosting the MaaS docs?
[17:21] <jamespage> smoser: so I see
[17:22] <Daviey> smb: thanks
[17:22] <hallyn> stgraber: but is that a debootstrap bug then?
[17:22] <stgraber> Daviey: well, our current trick to make resolvconf work in chroots was a bad idea (creating the symlink at boot time), so we'll revert that and replace by a relative symlink which should work for chroots but it's hard to think of all possible use cases
[17:22] <Daviey> stgraber: This seems to be a good candidate for a QA upgrade test?
[17:22] <allenap> mrevell: Sure.
[17:22] <Daviey> stgraber: right!
[17:23] <allenap> mrevell: Hangout.
[17:23] <stgraber> Daviey: so we didn't want to risk a rebuild for it and will upload today. I don't expect issues on upgrade (possibly less issues actually), but I have daily upgrade testing running here for all the flavours so that's easy enough to check
[17:23] <mrevell> allenap, Yo betcha, as they say.
[17:24] <mrevell> allenap, invite sent
[17:24] <stgraber> hallyn: could be, seems like qemu-debootstrap might be missing the initial sources.list line
[17:24] <stgraber> hallyn: I'll just update the template to wipe sources.list and write it from scratch instead
[17:24] <Daviey> stgraber: Something centralised would be good :).. and gema and jibel would love you.
[17:24] <hallyn> stgraber: ok
[17:25] <stgraber> hallyn: and add restricted and multiverse in the process, not sure why we don't have them at the moment
[17:25] <hallyn> stgraber: speed?
[17:25] <hallyn> adding those makes apt-get update forever on my laptop anyway
[17:25] <stgraber> Daviey: right, I have my own setup here because "Canonical no longer wants to spend resources on doing QA for flavours" ...
[17:26] <stgraber> hallyn: hmm, restricted and multiverse are pretty small compared to universe
[17:26] <hallyn> oh.
[17:26] <hallyn> yeah i was thinking universe.  i would admit not knowing what multiverse is, but i'm too manly
[17:27] <hallyn> so why would you want restricted and multiverse in a container?
[17:27] <Daviey> stgraber: Been drinking lemon juice? :)
[17:27] <stgraber> hallyn: main => 7.5M, multiverse => 592K, restricted => 129K and universe => 25M
[17:28] <stgraber> Daviey: ;)
[17:28] <hallyn> :)  ok.  so no problem, but still not sure what we'd want from there
[17:28] <hallyn> zul: smb: is "apt-get install openvswitch-switch" supposed to just work?
[17:28] <stgraber> hallyn: well, it'd be consistent with desktop and maybe someone may need some codecs or similar stuff for some weird setup :)
[17:28] <zul> hallyn: reputadely
[17:29] <hallyn> stgraber: oh, right and with arkose esp ppl might want codecs.  ok.
[17:29] <hallyn> zul: "FATAL: Module oepnvswitch_mod not found"
[17:29] <Daviey> hallyn: where are you seeing that?
[17:30] <zul> hallyn: try installing the openvswitch-dkms
[17:31] <hallyn> zul: it tried dkms automatically.  oh, there, very neatly hidden away, i guess the dkms build failed.  why not show me those erors, dingbat?
[17:31] <zul> but but :)
[17:32] <hallyn> yeah, errrors in brcompat.c
[17:32] <smb> zul, maybe we need sound support for dkms, doing the red alert thing when build fails
[17:32] <stgraber> hallyn: I'm assuming the typo in the module name is just a manual copy/paste typo and not a typo in a script right? (oepnvswitch_mod vs openvswitch_mod)
[17:32] <stgraber> oh, ok, yeah, if the module failed to build, that probably was just an IRC typo then ;)
[17:32] <zul> smb: red alert thing?
[17:32] <hallyn> stgraber: yeah.  this was in a vm on another laptop.  no cut/paste
[17:32] <smb> zul, like in star trek
[17:32] <zul> smb: ah right
[17:33]  * zul loves majel barret :)
[17:34] <hallyn> maybe it needs to #include <linux/module.h>?  it says expected declaration specificers or '...' before string constant on the MODULE_DESCRIPTION("..."); line
[17:36] <ahs3> mmmm....majel barret.....
[17:36] <smb> hallyn, sounds reasonable and probably is quick to test...
[17:37] <hallyn> smb: what's the quickest way?  if i change the src under /var/lib/dkms and do dkms build -m openvswitch -v 1.2.2, it reextracts the src :)
[17:37] <hallyn> do i have to rebuild the whole package and install it?
[17:37] <Daviey> hallyn: the build.log should show you how to reproduce the build out of band.
[17:38] <smb> hallyn, oh crap. somehow I expected it not to extrat every time. You could rebuild the tarball I guess
[17:39]  * hallyn gets cranky typing on this tiny keyboard with jacked-up capslock key
[17:39] <hallyn> (resituating)
[17:41] <hallyn> lol!
[17:41] <hallyn> well just a 'make' worked.  and yes that #include fixed that error.  which brings upthe next one:
[17:41] <hallyn> kernels after 3.0 are not supported by this version of openvswitch
[17:41] <hallyn> glad to see this is so heavily used :)
[17:42]  * smb wonders who tested that one before... :)
[17:42] <hallyn> well, all i'm saying is, if noone has noticed, then maybe we don't care all that much
[17:42]  * hallyn checks open bugs
[17:43] <smb> I am not sure I remember that one right, but wsn't that the package that first was discussed to move into main and then (a while ago) decided its not really required or so...?
[17:44] <hallyn> there is an open MIR for it.  i don't know about it being decided it wasn't needed.  zul?
[17:45] <zul> reading the backlog
[17:45] <zul> yeah i care about it, ill make sure it doesnt fail can you open up a bug about it
[17:45] <hallyn> well, sid has 1.3.0-1.  perhaps we should sync
[17:45] <hallyn> zul: sure
[17:45] <hallyn> zul: or do you just want to sync
[17:46] <zul> i was going to find some time to sync it up
[17:49] <hallyn> zul: filed bug 925611, thanks.  (let me know if you want me to do it, but happy to leave it to you as you've obviously done the last few).  I was just trying it out to see about making recommendations in documetnation for use with qemu+libvirt...
[17:49] <zul> hallyn: gotcha
[17:55] <mrevell> bigjools, Hey, do we have a "Bad" state for a node that's got some kind of problem?
[18:02] <bigjools> mrevell: no - we should though
[18:02] <mrevell> I've added one to the new glossary, which I'll share shortly
[18:02] <bigjools> mrevell: see src/maasserver/models.py
[18:02] <mrevell> thanks bigjools
[18:08] <adam_g> zul: that iscsi patch got merged, no need to carry patch
[18:08] <zul> yep i saw
[18:15] <stgraber> hallyn: any reason why we don't have a proper locale installed in the template?
[18:15] <hallyn> stgraber: i think we didn't know how to decide which locale is proper
[18:16] <stgraber> hallyn: well, having "a" locale would be nice, I'd just go with language-pack-en and en_US.UTF-8 by default. Using C is likely to break quite a few things ...
[18:17] <hallyn> i like locale C :)
[18:17] <hallyn> i suppose if there's a clean way to get the host's locale we could copy that into container at setup
[18:18] <stgraber> C.UTF-8 would be reasonable I guess if that works now
[18:20] <hallyn> stgraber: I'm just being an old man.  locales are new to me :)  Pls do what you think best.
[18:20] <hallyn> plus i don't run things like ff in my containers (yet)
[18:21] <hallyn> btw i'm working on a patch to have lxc-start bail early if insufficient privs
[18:21] <stgraber> hallyn: well, the biggest annoyance to me with our current lack of locales is all these warnings from perl that you get in APT and some other tools. I also remember having some pretty serious issues with PostgreSQL and other DB where they would store everything as ASCII unless the locale supported utf-8 at install time.
[18:22] <stgraber> if C.UTF-8 works, that should take care of all of these issues and not require any extra package
[18:22] <hallyn> ascii ftw :)
[18:22] <hallyn> ok
[18:23] <RoyK> EBCDIC FTW!
[18:23] <stgraber> ;)
[18:27] <tgardner> SpamapS, once you've created a cobbler repo by importing an ISO, is there a way to update it?
[18:30] <hallyn> stgraber: lp:~serge-hallyn/ubuntu/precise/lxc/lxc-start-checkperms  if you wanted to pull that patch (0032) into your tree
[18:35] <smoser> tgardner, what do you want to update?
[18:36] <tgardner> smoser, well, I imported a precise desktop ISO and I'd like to update to the most recent daily (which has a newer kernel)
[18:36] <smoser> would remove and re-add be not the same?
[18:37] <smoser> we have a tool that kidn of does this, but it really is hard-coded to do mini-iso stuff (cobbler-ubuntu-import).
[18:37] <tgardner> smoser, yes, but thats a bit of a pain in the ass, what with having to choose the right pre-seed and all.
[18:38] <smoser> it basically imports a new one, then renames it.
[18:38] <tgardner> been using that
[18:38] <smoser> so yeah...  we dont really have anything.
[18:38] <tgardner> ok
[18:38] <smoser> but you could take the logic from cobbler-ubuntu-import
[18:38] <tgardner> I'll have a look
[18:40] <adam_g> smoser: i took a look at that nova-volume charm patch yesterday. it requires the loopback device is first created on the host and allowed into the container by someone before charm deploy?
[18:40] <smoser> adam_g, yeah.
[18:41] <smoser> i do that in one of the pastebins.
[18:41] <zul> hallyn: around?
[18:41] <smoser> it assumes you have access to the loopback. whihc is not unreasonable.
[18:41] <hallyn> zul: yes, though lunching soon.  what's up?
[18:41] <adam_g> smoser: why not just look for /dev/loopN like any other block device while making the same assumption?
[18:41] <zul> hallyn: this use to work on oneiric: http://pastebin.ubuntu.com/826733/
[18:42] <smoser> adam_g, and assume that something else has alrady set up the block device?
[18:42] <smoser> because thats not generally helpful.
[18:42] <zul> hallyn: basically it would give you acccess to look at the console for libvirt-lxc and now it doesnt
[18:42] <smoser> (and that would already besupported if you just gave it 'loop0' as the config)
[18:42] <hallyn> zul: not sure, but offhand that could be due to getty change
[18:42] <smoser> note, the patch would work on ec2 also.
[18:43] <zul> hallyn: http://paste.ubuntu.com/826731/
[18:43] <adam_g> smoser: thats up the provider to decide before granting the container access to the device, no? the same thing could potentially happen with an EBS volume on amazon or openstack.
[18:43] <zul> hallyn: also there is some changes to /dev/ptmx and /dev/pts in libvirt i think as well
[18:43] <zul> hallyn: where would i start looking
[18:43] <smoser> adam_g, the charm has no notion of container.
[18:44] <smoser> and it should not.
[18:44] <hallyn> does this happen also with an oneiric container on precise libvirt?
[18:44] <smoser> maybe somewhere down the road, juju has some concept of "this charm requires access to these devices and these kernel modules"
[18:44] <smoser> but thats just not reasonable right now.
[18:44] <zul> hallyn: i havent checked yet
[18:44] <zul> hallyn: ill let you know
[18:45] <hallyn> ok
[18:45] <adam_g> smoser: sec, taking another look at the charm with fresh eyes..
[18:48] <adam_g> smoser: which is the pastebin that prepares the loop device on the host?
[18:51] <smoser> holdon
[18:52] <smoser> adam_g, http://paste.ubuntu.com/826756/
[18:52] <smoser> all that does is  make juju grant all containers access to all /dev/loop*
[18:56] <zul> hallyn:  same with oneiric its probably a libvirt issue i think
[18:56] <hallyn> zul: hm
[18:58] <zul> hallyn: http://libvirt.org/git/?p=libvirt.git;a=commit;h=c30a78c398135577c3038199cb81bfaa19708cc5
[18:58] <hallyn> zul: i don't think that's it
[18:59] <zul> well that patch isnt in libvirt yet
[18:59] <hallyn> zul: that just changes the underlying file from chardev to empty file (which we mount /dev/pts/ptmx onto)
[18:59] <zul> i have a strace if you want one
[19:00] <hallyn> zul: I'm guessing it has to do with the new routine for opening the ptys at container init.
[19:01] <zul> hmm
[19:01] <hallyn> i.e. virFileOpenTty
[19:02] <hallyn> what the heck is set_nonblocking_flag()
[19:04] <hallyn> zul: if you drop 'iflag=nonblock', does it work?
[19:04] <zul> in the dd?
[19:05] <zul> just sits there
[19:06] <hallyn> zul: that strace was not '-f' ?
[19:07] <zul> it wasnt
[19:07] <hallyn> can you give me -f output?
[19:08] <adam_g> smoser: ah, nevermind. charm looks good and works elsewhere. basically, i had no idea that 'losetup --find --show "$fpath"' actually sets up the device
[19:09] <zul> hallyn: http://paste.ubuntu.com/826777/
[19:10] <adam_g> smoser: ill probably steal get_block_device() in the next iteration of the swift charms
[19:10] <smoser> adam_g, sure. so you're merging from mine?
[19:11] <adam_g> smoser: yeah, my branch is still pending a MP into the main lp:charms, but your change will go with it.
[19:12] <Daviey> adam_g: you could merge into yours, and the MP will refresh
[19:12] <hallyn> zul: kernel change i'm thinking
[19:13] <zul> hallyn: damn
[19:13] <zul> thats unfortunate
[19:13] <hallyn> zul: got an oneiric box up?
[19:13] <zul> hallyn: i can do it in a vm...ill let you know
[19:13] <hallyn> zul: well open up a terminal, do 'tty', and try the dd on that pty
[19:14] <hallyn> that fails the same way here on my precise laptop
[19:14] <stgraber> hallyn: http://paste.ubuntu.com/826789/ is basically what we need to support armel/armhf/powerpc/... on x86
[19:14] <adam_g> Daviey: thats what i meant
[19:14] <zul> hallyn: yeah it fails here
[19:14] <stgraber> hallyn: I just uploaded mountall and sysvinit with the multi-arch changes, so hopefully in an hour or so, the template will work with these changes
[19:15] <hallyn> stgraber: what about soft freeze?
[19:15] <stgraber> hallyn: we've been out of soft freeze for 40min now
[19:15] <hallyn> doh!
[19:16] <hallyn> stgraber: being pedantic, but could you add comments at top of that new fn documenting expected $1..$3?  :)
[19:16] <zul> hallyn: ditto on my oneiric box
[19:16] <stgraber> hallyn: probably a good idea indeed ;)
[19:17] <Daviey> adam_g: ah
[19:17] <hallyn> zul: so how did that ever work...  ok, it must be something libvirt used to do to the ptys it creates for consoles, but no longer does
[19:18] <zul> hallyn: im not sure it just did is there a way where we can grab a console output under precise?
[19:23] <Unode|Work> hi everyone
[19:23] <Unode|Work> Can someone tell me how can I disable daily fetches of packages?
[19:24] <Unode|Work> I only want to update package lists when I want to upgrade things and not on a regular basis.
[19:24] <Unode|Work> Thanks
[19:27] <hallyn> zul: haven't found one yet
[19:28] <zul> hallyn: this is how everything is mounted in the container: http://pastebin.ubuntu.com/826803/
[19:47] <hallyn> zul: oneiric gives me the same behavior
[19:48] <zul> hallyn: yeah so i dont remember how that could have worked
[19:49] <hallyn> zul: use socat?
[19:49] <mjt> ok, thank you everyone (re ipxe)
[19:51] <zul> hallyn: how do you use socat?
[19:51] <hallyn> zul: 'socat - /dev/pts/1' for instance.  (http://ubuntuforums.org/showthread.php?t=1159220)
[19:52] <hallyn> do you expect this to work nicely while someone does 'virsh console <container>' too?
[19:54] <zul> not really i just want to get like a "picture" of the console
[19:58] <hallyn> zul: you know, oneiric got a version of the big pty update backported, so it's possible dd was broken by that after all
[19:59] <zul> hallyn: probably, although im not upset about it, just have to figure out something
[19:59] <hallyn> can socat work for you?
[20:00] <zul> hallyn: i think so
[20:00] <zul> hallyn: thanks
[20:00] <hallyn> cool, np
[20:21] <stgraber> hallyn: ok, so I got an armhf to install and boot, well, boot until mountall explodes when tryin to talk to udev ;)
[20:21] <hallyn> why does that explode?
[20:21] <stgraber> mountall:mountall.c:3700: Assertion failed in main: udev_monitor = udev_monitor_new_from_netlink (udev, "udev")
[20:22] <_Neytiri_>  i have bind9 set to do recursion but when i attempt to look up a domain i dont host it refuses the request how do i fix this
[20:30] <cr3> zul: a while ago, I asked for a way for a kvm guest to access the host filesystem. I found a very simple solution: nc
[20:32] <smoser> cr3, lame.
[20:32] <smoser> you should have found a kvm exploit.
[20:32] <smoser> that would be more general purpose.
[20:33] <jjohansen> hallyn: would you have time to test a kernel, and if so would you like -generic, -server, i386/x86-64?
[20:34] <cr3> smoser: if only you had suggested that when I first asked in the channel, you could've saved me so much time
[20:34] <hallyn> jjohansen: -generic would be great
[20:34] <jjohansen> hallyn: 64bit?
[20:35] <smoser> cr3, and you'd be rich and famous
[20:35] <hallyn> jjohansen: yes pls
[20:36] <jjohansen> hallyn: okay, I'll kick it off and let it build while I have some lunch
[20:36] <hallyn> jjohansen: cool.  this is for the detach problem?
[20:36] <hallyn> i'll try it out later tonight then
[20:36] <jjohansen> hallyn: yeah, see your mail
[20:37] <hallyn> (fwiw i try to check mail 3-4x/day :)
[20:38] <hallyn> stgraber: i wonder if it'd be safe to make /etc/init.d/lxc set -e
[20:39] <stgraber> hallyn: sounds dangerous, why would you do that?
[20:39] <hallyn> stgraber: bc it should fail if the network isn't set up right
[20:42] <stgraber> hallyn: well, it should fail, not just stop there. So it should handle return codes where relevant and undo whatever it did, indeed, set -e would just leave the system half configured
[20:42] <hallyn> stgraber: well, set -e + trap EXIT
[20:42] <hallyn> i wouldn't be un-tidy
[20:43] <stgraber> right, trap EXIT with a proper handler would be good
[20:47] <hallyn> stgraber: actually i'm just going to do trap and set -e only for the net setup, then undo them
[20:47] <hallyn> (it's just that adding 'if [ $? -ne 0 ]' after each stmt is ugly
[20:57] <stgraber> hallyn: looks like if I multi-arch makedev and iproute, then I can get a container to boot (with upstart, mountall, plymouth and iproute being amd64 in the armhf container)
[20:57] <lifeless> stgraber: that sounds unholy
[20:57] <stgraber> hallyn: testing armel now as I'm getting weird qemu output that may be related to armhf support, maybe armel will complain a bit less
[20:58] <_Neytiri_> how do i enable recursion in bind?
[20:59] <stgraber> lifeless: well, if qemu-user-static was doing a better job it'd just work ;) sadly anything using ptrace() fails under it (that's why I need an amd64 upstart) and apparently netlink is a bit broken too (breaking most of the network stuff and udev)
[20:59] <lifeless> hallyn: you can do statement || thingwhenfail
[21:00] <lifeless> hallyn: as an alternative to if [ $? -ne 0 ]
[21:00] <hallyn> lifeless: yeah, but then i want to cleanly unroll
[21:00] <lifeless> ah, well thats different :)
[21:02] <hallyn> lifeless: I'm thinking something like undo_network in http://paste.ubuntu.com/826916/
[21:13] <smoser> hallyn, the thing i've done for your trap on exit, which stacks better
[21:14] <smoser> is to set a global in that block of code
[21:14] <smoser> and then always trap exit and just check it.
[21:14] <smoser> note, that sh doesn't handle EXIT as well as bash
[21:14] <smoser> if user ctrl-c's it
[21:15] <hallyn> smoser: thing is i'd like to keep changes against debian's version as simple as possible.  and they don't have the whole network bit.
[21:15] <hallyn> now, maybe i should switch to upstart...
[21:15] <smoser> ah.
[21:23] <hallyn> smoser: i like the trick though.  will keep it in mind.
[21:31] <hallyn> stgraber: so i added an updated lxc.init to lp:~serge-hallyn/ubuntu/precise/lxc/lxc-start-checkperms
[21:33] <stgraber> hallyn: ok. I'm finishing to deal with some multi-arch issues to at least upload something that kind of boot instead of crashing and dumping you in a root shell
[21:33] <hallyn> hurray for root shell
[21:34] <stgraber> my fallback plan is to put a job as "start on startup" explaining that qemu-user-static is a bit limited and that the rest wouldn't boot ;)
[21:34] <stgraber> then start getty and that's it
[21:38] <rremer> I don't have the /desktop directory key in my gconf schema.  What package do I have to install to get that?  (specifically, I'm looking to be able to edit menu shortcuts: /desktop/gnome/interface:can_change_accels)
[21:39] <rremer> I've installed GNOME, but the key is not there still.
[21:47] <melter> has there been a change in how networking is configured in pangolin?
[21:52] <melter> anyone know why dns resolution isn't working in a new pangolin alpha 2 install?
[21:52] <RoyK> melter: does "host google.com" work?
[21:53] <stgraber> hallyn: yeah! got the container to boot "cleanly", I just need "upstart:amd64 mountall:amd64 iproute:amd64 isc-dhcp-client:amd64", so that's pretty much all of what's using netlink
[21:54] <stgraber> melter: did you run "sudo start resolvconf && reboot" as per the release notes
[21:54]  * hallyn isn't quite sure whether th cheer or not :)
[21:54] <stgraber> hallyn: well, it's better than nothing (I guess), but having ptrace() and netlink support in qemu-user-static would be magic ;)
[21:56] <hallyn> :)
[22:06] <melter> stgraber: "start: Job is already running: resolvconf"
[22:07] <melter> RoyK: no
[22:07] <stgraber> melter: ok, is that the first or second boot of the machine?
[22:08] <melter> stgraber: i've already rebooted several times
[22:08] <melter> so it wasn't the first
[22:08] <stgraber> melter: ok, so that's probably not the resolvconf race condition then. Can you confirm /etc/resolv.conf is a symlink to /run/resolvconf/resolv.conf?
[22:09] <melter> stgraber: yes
[22:10] <stgraber> melter: can you post your /etc/network/interfaces to a pastebin?
[22:16] <melter> stgraber: http://pastebin.com/hsjQbwgb
[22:16] <melter> it's the same as a 11.10 production machine, except the static ip is different
[22:17] <stgraber> melter: I'm hoping you didn't get that as a result of the installer?
[22:17] <melter> stgraber: no
[22:17] <stgraber> good, because it's wrong (on 12.04)
[22:17] <stgraber> well, not really wrong, rather incomplete
[22:17] <melter> stgraber: ah, ok
[22:17] <stgraber>  /etc/resolv.conf is managed by resolvconf now
[22:17] <stgraber> so anything you write to /etc/resolv.conf will be lost after a reboot
[22:17] <stgraber> you need to use:
[22:17] <stgraber> dns-nameserver 8.8.8.8 8.8.4.4
[22:18] <stgraber> *dns-nameservers
[22:18] <stgraber> and dns-search my.domain.com
[22:18] <stgraber> in /etc/network/interfaces
[22:18] <melter> so it isn't done automatically anymore?
[22:18] <stgraber> well, we weren't doing anything automatically before, /etc/resolv.conf was a static file
[22:19] <stgraber> now it's a dynamic file, so your DNS configuration needs to be in /etc/network/interfaces for resolvconf to configure resolv.conf properly
[22:27] <melter> stgraber: thanks, that worked
[22:27] <melter> is there a better way to set a hostname and static ip than editing that file?
[22:29] <stgraber> melter: well, the hostname should be in /etc/hostname and /etc/hosts but all the network configuration should indeed be in /etc/network/interfaces
[22:29] <stgraber> melter: you can get that file generated for you if you install using the static configuration
[22:30] <melter> ok, thanks
[22:46] <stgraber> hallyn: uploaded the remaining bits, hopefully in an hour or so I can test, then upload LXC
[22:47] <jjohansen> hallyn: I hit a snag with the test kernel I'll get you a build tomorrow
[23:11] <melter> in pangolin alpha2, "apt-get autoremove --purge -y g++" fails with "g++ is already the newest version."
[23:29] <SpamapS> Hmm, I'm trying to get a VM to pxe boot.. never seeing the DHCP requests on virbr#
[23:34] <adam_g> SpamapS: bug #924446
[23:34] <adam_g> SpamapS: apparently a bug in libvirt thats been fixed upstream and will be cherrypicked back thanks mr. hallyn
[23:35] <SpamapS> adam_g: good to know!
[23:36] <SpamapS> so if I had dist-upgraded today, I wouldn't have this problem
[23:36] <adam_g> SpamapS: only one way to find out?
[23:37] <SpamapS> adam_g: a Mutha****in MONTAGE!
[23:38]  * twb quietly remove's SpamapS' caffeine drip bag
[23:38] <SpamapS> twb: I switched to guarana and horny goat weed 2 hours ago
[23:39]  * SpamapS goes off to find some Yerba Mate
[23:48] <hallyn> SpamapS: I only pushed the change a few hours ago
[23:49] <SpamapS> hallyn: oh ok cool, well glad that the fix won't bite anyone else... cost me about an hour and two small patches of hair on the top of my head ;)
[23:50] <_godhelpme> hi could i get some help setting up permissions
[23:50] <hallyn> SpamapS: it's been broken for awhile, odd that everyone notices at once :)
[23:51] <SpamapS> hallyn: I think only recently we've wanted to pxe boot our vms
[23:51] <_godhelpme> how do i make it so that if a user makes a file it will by default create it so that the owner would be user and the group would be newgroup
[23:54] <starlocke> there's that PGP key server concept... is there anything similarly centralized for SSH keys...?
[23:56] <SpamapS> DEBUG:root:Test b4d2d6e6-af86-4772-b645-9d835382130c failed to execute within 20 minutes
[23:56] <SpamapS> durn it.. my poor machine couldn't finish the install in time. :-P