[01:45] <Demosthenes> so i have a lucid server, i'm trying to use the BIOS serial over LAN support, but I must disable ALL the font/mode setting... grub's console setting isn't enough. suggestions? i'm trying to stay in text mode on the console, no fonts, no changes.
[01:45] <qman__> add 'nomodeset' to your boot line
[01:46] <Demosthenes> that didn't seem to help
[01:47] <Demosthenes> the initrd set it anyway
[01:47] <Demosthenes> i even removed console-setup, but console-setup-min seems to set these anyway
[01:53] <twb> Demosthenes: oh that one.  It's bloody annoying.
[01:54] <twb> Demosthenes: blacklist vga16fb in /etc/modprobe.d/thingo.conf
[01:54] <twb> Demosthenes: also note that due to a bug in the installer, there is absolutely no way to disable the framebuffer during install
[01:54] <twb> (in lucid that is)
[01:54] <twb> Don't forget to update-initramfs -u -k all after you edit the modprobe.d
[01:54] <Demosthenes> hey twb, i think we've met before.
[01:55] <Demosthenes> i just tried the nomodeset again, no dice
[01:55] <twb> I also routinely purge plymouth packages as much as possible, then try to manually disable the plymouth upstart jobs that the upstart package has hard dependencies on
[01:55] <twb> I don't know if that's necessary or if it just makes me feel better
[01:56] <twb> >hate plymouth hate hate hate<
[01:57] <Demosthenes> twb: i can tell
[01:57] <Demosthenes> i edited /etc/modules.d/blacklist-framebuffer.conf and added vga, and my init has been updated.. no dice
[01:57] <twb> vga16fb
[01:58] <twb> There is no driver "vga"
[01:58] <twb> # Make the machine-room KVM usable.  --twb, Jul 2011
[01:58] <twb> blacklist vga16fb
[02:00] <Demosthenes> god i miss aix :P
[02:00] <Demosthenes> all this just to get an intel box i can remotely manage
[02:03] <twb> Demosthenes: still got that bottle in the bottom drawer?
[02:03] <Demosthenes> i didn't update the initrd after changing modules.
[02:03] <Demosthenes> i need one. i hate pc hardware
[02:03] <Demosthenes> this "server" failed while i was travelling. couldn't walk a user through booting it on a kvm, they eventually had to plug a usb kbd into the MB...
[02:03] <Demosthenes> *sigh*
[02:04] <Demosthenes> so i bought a whole new MB, CPU, and ECC RAM...
[02:04] <Demosthenes> with a MB with IPMI support
[02:05] <Demosthenes> and now to get SOL access via IPMI i have to disable the fonts on the console
[02:06] <twb> It's unlikely to be the fonts, it's very likely to be something forcibly loading stupid non-80x25 console
[02:06] <Demosthenes> well, but font-setting does that
[02:06] <twb> It drove me insane when I was using kvm -curses, for the same reason
[02:06] <twb> Demosthenes: no it doesn't, at least not IME
[02:06] <Demosthenes> right. it did it again. nomodeset and i added the blacklist and updated initrds, it STILL did it
[02:07] <twb> So just to be clear, you have "blacklist vga16fb" in /etc/modprobe.d/blacklist-framebuffer.conf, and /boot/initrd.img-... is newer than that .conf file?
[02:08] <twb> And after editing /etc/default/grub you ran update-grub?
[02:08] <twb> http://paste.debian.net/173073/ is my /etc/default/grub
[02:08] <twb> Also, can you tell how far it gets before it trashes your display?
[02:10] <Demosthenes> yeah, the kernel starts, and it starts the initrd..
[02:11] <Demosthenes> i've also tried adding "text", "textonly", and a few others to the kernel line
[02:11] <twb> You can't just guess those things
[02:11] <twb> Is there a GUI installed?
[02:11] <Demosthenes> nope.
[02:11] <Demosthenes> this was intended to be a headless server
[02:12] <twb> IMO better to use serial for headless boxes
[02:12] <twb> But I'm stumped, I can't think what else coudl be happening
[02:12] <Demosthenes> the last line i see on serial is "Loading initial ramdisk ...", and the dmesg output begins after the mode sets
[02:13] <Demosthenes> so before the frist [0000000] Linux message it has changed
[02:13] <twb> Are you booting with "quiet" ?
[02:13] <twb> Also on serial, don't forget you will need to create /etc/init/ttyS0.conf
[02:14] <Demosthenes> no, this is serial bios.
[02:15] <twb> Ugh
[02:15] <Demosthenes> if i stay in text mode, the bios bumps the vga text out as serial over lan
[02:15] <Demosthenes> so for instance, i can get into the BIOS no problem.
[02:15] <twb> You should tell it to stop doing that bullshit, and to just hand the serial back to the OS as soon as it loads the bootloader
[02:16] <Demosthenes> if i just dont' change the video mode at boot, it'd work fine.
[02:16] <twb> IMO what should happen is the BIOS allows you to use serial to configure itself, but ignores the VGA head and you then simply tell grub, the kernel and the getty to talk to ttyS0
[02:17] <twb> Having IPMI reexport a vga console over serial instead of just talking over serial, is bloody stupid
[02:17] <twb> (I'm ranting at the vendor, not you)
[02:17] <Demosthenes> well, if i don't do it that way, i can't set bios settings remotely
[02:18] <twb> Yeah I realize that
[02:18] <Demosthenes> so no editing the bootlist, etc.
[02:20] <Demosthenes> wtf, mountall and cryptsetup require plymouth?
[02:20] <twb> *EXACTLY*
[02:20] <twb> It's bullshit
[02:21] <twb> I bitched to cjwatson about it and he said something "everyone stop bitching, plymouth is more than just a splash screen, you need it or else"
[02:22] <Demosthenes> *sigh*
[02:22] <Demosthenes> i just butchered the hooks for initrd
[02:22] <Demosthenes> no more kbd, framebuffer, console-setup.
[02:23] <qman__> I agree completely
[02:23] <qman__> I don't understand how it can be required when if I simply nuke the upstart scripts, everything still works and works the way I want it to
[02:23] <twb> qman__: it doesn't tho
[02:23] <Demosthenes> what about 12.04?
[02:23] <twb> qman__: specifically if fsck fails, mountall will just hang forever if plymouth isn't doing its  thing
[02:24] <twb> qman__: you can't recover unless you have a live CD
[02:24] <qman__> hmm, haven't run into that situation
[02:24] <qman__> but one of my servers where I didn't do that, I never get to see my fsck anyway
[02:24] <twb> Also this happens if you tell LVM to make a LV read-only, because the boot-time fsck *always writes to ext partitions* -- WTF
[02:24] <twb> qman__: well OK maybe that is not plymouth being broken, it's just ubuntu being broken
[02:25] <Demosthenes> ok, i give. it must be compiled into the kernel
[02:25] <qman__> yeah, I know it's not broken in general, just my case
[02:25] <twb> All you need to know about plymouth being stupid is it came from fedora, like pulseaudio and systemd
[02:25] <qman__> but it still pisses me off
[02:25] <twb> qman__: agreed
[02:26] <twb> It also pisses me off that nobody lets you just ^C init jobs that have hung anymore
[02:26] <twb> Long ago, before debian got startpar, if something like postfix hung during boot, yo could just hit ^C to skip it and plough blindly on
[02:26] <twb> e.g. if the DNS server was cactus that was useful
[02:28] <qman__> I've run into that postfix hanging during boot problem, but only on fedora
[02:29] <twb> Well I can't remember which daemon exactly
[02:29] <twb> I do remember I ran into a really AWESOME feature of centos4 in that it uses ls to list files in /etc/sysconfig/networking to find out which ifaces to raise
[02:29] <qman__> postfix wasn't configured out of the box, but it'd try to start it anyway and just sit for about ten minutes before moving on
[02:30] <twb> And because it uses ls, and because this happens before the network is up (obviously) and thus the LDAP server is not accessible, if you put an extra file in there, say eth0-up.bak, it will take an extra TWENTY MINUTES to boot while ls tries to list it
[02:30] <qman__> heh
[02:30] <twb> And this issue would be completely bypassed if RH engineers had a flipping clue about how to write sh scripts...
[02:31] <twb> i.e. you do not bloody do for i in $(ls *.conf)
[02:31] <twb> Incidentally you can work around that by telling libldap to bind softly.
[02:49] <Demosthenes> further hacking apart initfs
[02:50] <Demosthenes> did they fix this in 12.04?
[02:51] <twb> TBH I don't know, the only host I'm running precise on has no dsub port at all
[02:51] <twb> (yaaaay)
[03:01] <rockets> Is it just me, or is rsnapshot trying to do the same thing three times here:  https://gist.github.com/2879531
[03:05] <twb> I agree with your analysis.  pastebin your .conf
[03:05] <twb> Ah, no, /media/nas/admin &c are different
[03:05] <twb> It is more obvious if you unwrap the lines
[03:08] <rockets> twb, yeah I just noticed that
[03:08] <rockets> thanks
[03:09] <rockets> do you think it's safe for me to copy hourly.7, say, to hourly.{0-6}, just so the all the hourlies are populated and i can have the dailies actually run??
[03:09] <rockets> or will that break everything
[03:21] <twb> dailies should not know or care about hourlies...
[03:21] <twb> AFAIK they're totally independent runs as far as rsnapshot is concerned
[03:21] <twb> (I may be misremembering...)
[04:00] <captbaritone> Any idea why "sudo chmod -R 0777" would return "chmod: getting new attributes of `…': No such file or directory"
[04:00] <captbaritone> (file names omitted)
[04:01] <twb> Because the file doesn't exist.
[04:01] <twb> DO NOT EVER run that chmod command.  BAD USER!  No biscuit!
[04:02]  * LordOfTime agrees with twb
[04:02] <twb> If you just ran that over a directory like /, /etc or /usr, you have permanently broken your system and you should reinstall from scratch right now.
[04:02] <LordOfTime> indeed
[04:03] <twb> Also please thump the idiot that told you to run it
[04:03] <captbaritone> <thumps self>
[04:03] <captbaritone> It was run on a sub directory
[04:04] <LordOfTime> its still the last command you'll ever want to run
[04:04] <LordOfTime> so if you dont mind me being perfectly outright with my question asking:
[04:04] <LordOfTime> why the hell did you run that command
[04:04] <twb> At absolute worst you probably want to use something like a+rwX, which will only set x if x was already set
[04:04] <LordOfTime> what came over you to chmod recursively with 0777 :/
[04:04] <captbaritone> Sure.
[04:04] <LordOfTime> twb:  indeed
[04:05] <captbaritone> I have a HUGE collection of files on my personal machine which I need to be accessible by the user that runs my torrent deamon, the user that runs my web server, and myself
[04:05] <captbaritone> I had them chmodded 772
[04:05] <twb> chmod -Rv go-rwx /srv/tftp | sed -e '/mode of .* retained as/d'
[04:05] <LordOfTime> well doing anything more than 744 is just wrong
[04:05] <twb> ^^ that is a more normal use of chmod -R
[04:06] <captbaritone> with them owned by a custom user and group (to which all three belonged)
[04:06] <LordOfTime> shut up virtbot
[04:07] <twb> LordOfTime: 644 or 755 (ref. +X above), not 744
[04:07] <captbaritone> My problem was that new files were being owned by the user/group of the user that created them
[04:07] <LordOfTime> twb:  mistyped
[04:07] <LordOfTime> meant 755
[04:07] <captbaritone> rather than the user/group that would allow access to all three users.
[04:07] <LordOfTime> (had to divert to another channel)
[04:07] <twb> captbaritone: add all three users to a group (probably creating that group) then make the parent directory  setgid that group
[04:09] <captbaritone> I see. Have all my scripts run setgid before they touch the files?
[04:09] <twb> If you intend to continue uploading new files in a broken way (e.g. scp -p), you may also wish to write e.g. an @hourly cron job that will chgrp -R over the dir in question
[04:09] <twb> captbaritone: no, setgid is a property of a directory
[04:10] <captbaritone> okay, I guess I'm confused by the man page
[04:10] <twb> If a directory is setgid and grouped to, say, www-data, then all new files created in that dir will default to group www-data
[04:10] <twb> setgid means something completely different for files, so do not set it on files, only the dirs in question
[04:10] <captbaritone> that makes sense (and is what I wanted!)
[04:11] <captbaritone> okay
[04:11] <captbaritone> (thanks by the way)
[04:11] <twb> sudo install -dm2775 -unobody -gwww-data /srv/www
[04:11] <twb> Something like that
[04:11] <captbaritone> So every time a script creates a directory, it should setgid that, or will the new directories inherit it?
[04:14] <captbaritone> I suppose I could just test that myself.
[04:15] <twb> setgid isn't inherited by new subdirs
[04:16] <twb> Also you will probably need root privileges to add setgid to a dir
[04:16] <captbaritone> Hmm
[04:17] <captbaritone> so my cron job which moves files into this deeply nested directory structure will need to run as root?
[04:17] <captbaritone> Seems a bit silly
[04:17] <captbaritone> I really should not have to do that
[04:18] <captbaritone> (the cron job moves files AND creates these directories)
[04:22] <captbaritone> So, is the gid different than the first digit of the four digit "mode"
[04:22] <captbaritone> ?
[04:22] <captbaritone> err
[04:22] <captbaritone> poorly worded
[04:24] <captbaritone> Here is a better way of getting at what I want to know: Is the purpose of using "install" over "chown" in the suggestion you made to make sure it only acts on directories?
[04:27] <twb> captbaritone: once the directories exist, files created in them will have the appropriate group by default
[04:28] <captbaritone> What about directories created in them?
[04:28] <twb> It's not recursive
[04:29] <captbaritone> Should I simply chown them after I create them?
[04:29] <twb> If you can't create the directories in advance, there is little point using setgid dirs.
[04:29] <koolhead17> hi all
[04:29] <twb> Yes, just using chgrp -R is probably the least silly.
[04:30] <captbaritone> why -R ?
[04:30] <captbaritone> oh
[04:30] <captbaritone> to get the files too
[04:31] <captbaritone> So that would take the place of setgid altogether
[04:33] <twb> Yes
[04:37] <captbaritone> Or I could just chgrp each file and directory imediatly after I create it (which would be a bit cleaner)
[05:00] <ihashacks> captbaritone: incron is something you could try using to autotmatically chgrp any folder that is created
[05:04] <ihashacks> I only caught the tail end of the conversation though so something like that might be overkill for your goal.
[05:08] <twb> ihashacks: summary is: he wants to drop a directory tree onto a new box, such that his user, www-data and the torrent user can all access those files
[05:18] <oasisbob> hi all, I'm still seeing 403 forbidden errors when trying to update from us-east-1.ec2.archive.ubuntu.com
[05:18] <oasisbob> according to @utlemming, the issue was thought to have been resolved
[05:19] <twb> oasisbob: how about this one http://paste.debian.net/173091/
[05:23] <oasisbob> twb: not sure about that one, but if I fall back to us.archive.ubuntu.com it works as expected
[05:23] <twb> OK
[05:23] <twb> Wasn't sure if you wanted to fix the mirror or just have apt-get working
[05:25] <oasisbob> just trying to get apt-get functional, we don't run our own mirror (yet)
[05:25] <oasisbob> ...but when the EC2 mirror goes has difficulties, it sure wreaks havoc with cloud-init, auto-scaling and the like
[05:25] <twb> That paste ought to pick a nearby mirror
[05:26] <twb> Ah, OK.  I don't do cloud stuff.
[05:28] <oasisbob> heh, thanks though. that works -- wasn't aware of that trick.
[05:30] <twb> I'd prefer cdn.ubuntu.com but that hasn't happened yet
[05:30] <twb> That's how debian does it.
[05:31] <twb> e.g. apt-file doesn't understand the stupid mirror:// syntax
[05:31] <oasisbob> the wrinkle with the mirrors:// would be in additional bandwidth charges, unpredictable latency, &c
[05:31] <twb> Yeah granted
[05:32] <twb> I'm not pushing it as a solution, just an option
[05:32] <oasisbob> but i'm assuming one could use the same thing, but with the EC2 mirror, then fall back to whatever... i need to understand how apt-get handles failure scenarios better.
[05:33] <twb> apt uses the order inw hich they appear in sources.list
[05:33] <twb> So if the same content is in ec2 and mirror://, just put ec2 first
[05:33] <twb> It should fall back on 4xx
[05:33] <twb> In fact that's exactly what I do here: http://paste.debian.net/173092/
[05:34] <oasisbob> hmmm, will it still exit without errors? (eg, a 100 exit code will abort cloud-init)
[05:34] <twb> file:///srv/apt is a local debmirror, i.e. we are not billed for I/O to it
[05:34] <twb> oasisbob: not sure
[05:34] <twb> oasisbob: usually I run apt-get by hand so I don't care, or I use something like debootstrap which can only take one mirror
[05:35] <oasisbob> yeah, sadly, I have to care about such things. easy enough to test though.
[05:35] <twb> If I were doing that cloud crap I would probably be rolling images in the same way
[05:36] <twb> Oh, I see, cloud-init does broadly the same job as live-config
[05:36] <twb> Or say oem-config
[05:37] <twb> The actual initialization of the image is already assumed to be done (presumably pre-rolled images from Ubuntu)
[05:37] <oasisbob> yup -- exactly.
[05:37] <oasisbob> to each their own -- but I really enjoy not having to seal images &c -- use an official image, cloud-init, then hand off to puppet or chef
[05:37] <twb> Why bother with cloud-init in the first place then?
[05:38] <twb> Just prepare an image that has puppet in it from the get-go
[05:38] <twb> http://paste.debian.net/
[05:38] <oasisbob> it handles the really low-level stuff like node identity, etc. when you initialize a VM in EC2, you can pass off the user-data which cloud-init picks up on
[05:38] <twb> Gah
[05:39] <twb> oasisbob: why doesn't that stuff just arrive via DHCP?
[05:39] <oasisbob> don't run the DHCP server. :)
[05:39] <twb> Humph
[05:40] <oasisbob> yup -- the universe comes in w/ 16k (32k?) max of data, and the instance goes from there. it's an attitude adjustment, to say the least
[05:40] <twb> 16k of what?
[05:40] <twb> bytes ?
[05:41] <oasisbob> yeah. there's a limit on the amount of what AWS calls user data that's passed to their API when you initialize the instance.
[05:43] <twb> It sounds like some of it is just workarounds for amazon being stupid
[05:44] <oasisbob> vs PXE/TFTP? :) to be honest, this part has never really bothered me.
[05:45] <Syria> Hello there, OpenVpn AS is installed and configured on my ubuntu 10.04 LTS server, Could you please let me know how can I view the accounts and configure it?
[05:45] <oasisbob> there's plenty of entropy in the system after your instances come up that make the bootstrapping seem downright peaceful
[05:46] <oasisbob> ... as long as you have a reliable apt repo, that is.
[05:46] <oasisbob> twb: thanks for the pointers
[06:03] <twb> Well you need the entropy to generate SSH host keys and TLS keys
[06:04] <twb> Does amazon host ARMv7 images yet, or only x86 / x86-64?
[06:31] <Demosthenes> *sigh* ok, i give up on why the hell i can't use the SOL function with this MB. i'll just open WINDOWS to run the java applet to administer my linux box.
[06:31] <Demosthenes> 'cause though it's java, javaws crashes with it in linux.
[06:32] <twb> SOL?
[06:33] <twb> Is that yet another acronym for "getting to the BIOS without plugging a monitor and keyboard directly into it" ?
[06:33] <twb> And yeah, IME they are all RFB under the hood, but wrapped inside a broken java layer that needs windows and is non-trivial to simply spoof by hand.
[06:36] <Demosthenes> serial over lan
[06:36] <Demosthenes> great acronym though to use with intels
[06:36] <twb> Ah right
[06:36] <Demosthenes> just one more reason they should never be servers, even at home.
[06:37] <twb> Well fungibility is a pretty good sales pitch
[06:38] <twb> I have not dealt much with power or sparc, but IME even tier 1 enterprise x86 gear absolutely sucks at doing stuff like ipmi or hw raid in any remotely sane way
[06:55] <Zanzacar> I am trying to use a dyndns for the cname in enom. It keeps having issues with this has anyone been successful with this?
[06:56] <Zanzacar> nm for some reason it works now
[07:04] <twb> CNAME can only direct to another name
[07:04] <twb> IPs go in A RRs
[07:06] <Demosthenes> twb: i have. multiple types. thats why i can't recommend intel for any mission critical systems.
[07:06] <Demosthenes> with prejudice
[07:07] <twb> But the point is I cannot just ring up a vendor three doors down from me and buy a 3U server running sparc or power for $1200
[07:08] <twb> Obviously if my hardware budget was $12,000 that would be less of an issue
[07:08] <twb> I wish I could, even if only to avoid the onrushing EFI clusterfck
[07:22] <remix_tj> uhm, anyone using winbind on precise to authenticate a client to an AD domain? i need always to restart winbind before logging in
[07:42] <muhqu1> hey guys, just wanted to let you know that us-east-1.ec2.archive.ubuntu.com is throwing 403 Forbidden on one of its IPs
[07:43] <twb> muhqu1: file a bug on launchpad
[07:44] <ivoks> i've noticed that too just now
[07:45] <muhqu1> ivoks: this describes the problem… https://gist.github.com/2880501
[07:46] <ivoks> yes, i'm working on workaround right now
[07:53] <jibel> jamespage, good morning
[07:54] <jamespage> jibel, morning
[07:54] <jibel> jamespage, all quantal server ec2 tests failed
[07:54] <jamespage> jibel, yes - thats me
[07:54] <jamespage> jibel, I just ran the job with the image for a1 - but its not published yet so all the test bombed.
[07:54] <ivoks> ec2 archive is broken
[07:54] <jibel> jamespage, ok, it's under control then  :)
[07:55] <jamespage> ivoks, I see issues in one region from the smoke I did for quantal this morning - what are you seeing?
[07:55] <jamespage> ivoks, https://jenkins.qa.ubuntu.com/view/ec2%20AMI%20Testing/view/Overview/job/quantal-server-ec2-daily/
[07:55] <ivoks> jamespage: archive httpd is returning 403; it's being worked on
[07:55] <jamespage> ivoks, great - I'll not go an poke people then
[07:56] <jamespage> jibel: yep - under control - just waiting for the images to spin out of production
[07:56] <jamespage> jibel, hows everything else looking
[07:56] <jamespage> ?
[07:59] <jibel> jamespage, everything else is looking good, as good as it can be with an A1.
[07:59] <jamespage> jibel, marvellous
[08:34] <ivoks> jamespage: it looks like it's solved
[08:34] <ivoks> muhqu1: try now :)
[08:36] <muhqu1> thanks a bunch, works
[08:36] <jamespage> ivoks, thanks - just waiting for the alpha1 images to finish production now.....
[08:36] <Daviey> jamespage: that was a grub typo?
[08:37] <jamespage> Daviey, what was a grub typo?
[08:37] <jamespage> the requirement for a re-spin
[08:38] <Daviey> jamespage: ec2 rebuild?
[08:38] <Daviey> (question, not a statement :)
[08:38] <jamespage> Daviey: looking at the email trail it was a requirement for a new kernel
[08:39] <Daviey> jamespage: just noted, http://bazaar.launchpad.net/~ubuntu-on-ec2/vmbuilder/automated-ec2-builds/revision/482
[08:40] <Daviey> jamespage: email trail?
[08:40] <jamespage> Daviey, I don't think that is related
[08:40] <jamespage> Daviey, between smoser, utlemming and I re testing of a1 images....
[08:41] <Daviey> jamespage: Is there anything documented publicly ?
[08:42] <Daviey> jamespage: Do you know why vmbuilder and live-build processes are being maintained for Quantal?
[08:43] <jamespage> Daviey, I thought we had moved over the live-build so no
[09:26] <lynxman> morning o/
[09:55] <Adonis> does anyone no a command to see what services are running on my ubuntu server?
[12:11] <zul> good morning
[12:17] <LordOfTime> any of you ever heard of PHP needing root to run python scripts with shell_exec() or exec()?
[12:18] <LordOfTime> (in Ubuntu)
[12:27] <smoser> Daviey, where do you see that we are maintining both processes?
[12:27] <Daviey> smoser: seeing commits to both trees
[12:28] <smoser> (i would not argue against that by any means, but i would actually be surprised if the vmbuilder path works. that said, clearly we dont intentionally do things to break it.)
[12:28] <smoser> link?
[12:30] <Daviey> https://code.launchpad.net/~ubuntu-on-ec2/vmbuilder/automated-ec2-builds , smoser
[12:30] <smoser> yes, i know where the code is :)
[12:31] <Daviey> smoser: then why are you asking?
[12:31] <smoser> because i dont see what you are seeing.
[12:31] <smoser> you're probalby assuming that 'vmbuilder-cloudimg-fixes' is only relevant to vmbuilder.
[12:31] <smoser> whihc is a reasonable guess, but is incorrect
[12:32] <Daviey> i'm probably assuming that being under the vmbuilder project, it is vmbuilder based :)
[12:32] <Daviey> and lp:~ubuntu-on-ec2/live-build/cloud-images is live-build based
[12:33] <Daviey> I guess i'm just old fashioned thinking that code under one project, is for that project.
[12:33] <smoser> well, you know what happens when you assume.
[12:33] <smoser> automated-ec2-builds clearly started as a vmbuilder project.
[12:33] <smoser> now it does more than that.
[12:34] <Daviey> shouldn't it be switched to be under ubuntu-on-ec2 project, at least?
[12:34] <smoser> ubuntu-on-ec2 project is gone.
[12:34] <smoser> it is historic only. ubuntu on ec2, is "just ubuntu".
[13:56]  * ogra_ sighs ... so my mailserver gets 501 errors from some servers it talks to since a few days
[13:58] <ogra_> does anyone else experience that under lucid ?
[14:00] <Jeeves_> Did you change your reverse dns?
[14:00] <ogra_> i never had reverse DNS on this machine ...
[14:01] <ogra_> it worked for the last 4 years like that ...
[14:02] <maxmahem> A lot of mail servers will reject you w/our rDNS. Including AOL and Yahoo I believe.
[14:05] <ogra_> maxmahem, well, as i said, it worked for the last 4 yeras and its only the web.de server that suddenly refuses
[14:44] <smoser> jamespage, you spent time doing i386 tests?
[14:44] <jamespage> smoser: a little
[14:44] <smoser> you, sir, are a king.
[14:44] <smoser> (although it might have been more useful to spend that time on amd64)
[14:45]  * smoser starts to do some testing there.
[14:45] <jamespage> smoser: actually that ISO downloaded first - hence why I did some
[14:45] <jamespage> (iscsi is automated btw)
[14:45]  * Daviey is so glad that one is automated
[15:05] <smoser> ivoks, if i upload https://code.launchpad.net/~ivoks/ubuntu/lucid/glib2.0/887946/+merge/81703 will you test it?
[15:07] <kayakyakr> Network guys hard-powered my server this morning to replace a UPS battery and now it won't come back up. The error message I'm seeing is identical to this one: http://jhonjairoroa87.blogspot.com/2011/01/ubuntu-boot-error-no-init-found-try.html
[15:08] <ivoks> smoser: will do
[15:09] <kayakyakr> when i'm dumped into busybox, i can mount that volume, without change, as long as I provide arguments: mount -t ext4 -f /dev/mapper/ubuntu-root /root
[15:10] <kayakyakr> already booted into the live cd, ran fsck on /dev/sda1 and /dev/ubuntu/root (the LVM partition)
[15:11] <kayakyakr> haven't yet tried to rewrite initramfs with update-initramfs. would that be your next suggestion?
[15:23] <patdk-wk> kayakyakr, sounds like it's mounting the wrong filesystem as root
[15:30] <kayakyakr> patdk-wk i think that might be the case. any suggestions on where I might be able to modify that setting?
[15:30] <patdk-wk> in grub
[15:35] <kayakyakr> well, it is mounting the correct volume, I don't think grub really cares about the type?
[15:36] <kayakyakr> we get past grub, get to initramfs before it blows
[15:46] <patdk-wk> kayakyakr, who said grub cares at all?
[15:47] <patdk-wk> hmm, so it mounts the correct root?
[15:51] <kayakyakr> yes
[15:52] <kayakyakr> patdk-wk it's trying to mount /dev/ubuntu/root which is the correct root. it fails when it actually tries to mount that.
[15:53] <patdk-wk> then you need to figure out what went wrong. lvm failed?
[15:53] <patdk-wk> I don't do lvm much myself
[15:53] <RoyK> kayakyakr: /dev/ubuntu/root is probably the lvm device, not the mount point...
[15:54] <RoyK> s/lvm device/lvm logical disk device/
[16:06] <kayakyakr> royk: yes. the mount point is /root (in initramfs). once the system starts /dev/ubuntu/root is mounted to /
[16:25] <AnGrYfUrBy> hey guys is anyone having issue's with atftp not loading on ubuntu 12.04
[16:26] <ihashacks> AnGrYfUrBy: https://bugs.launchpad.net/ubuntu/+source/atftp/+bug/972834
[16:26] <ihashacks> Like that?
[16:27] <AnGrYfUrBy> exactly like that
[16:27] <AnGrYfUrBy> is der a fix for us n00bs
[16:28] <ihashacks> add "--port 69" to the OPTIONS section in /etc/default/atftpd
[16:28] <ihashacks> also, if you are in launchpad, please mark that the bug affects you too :)
[16:31] <kayakyakr> fixed it.
[16:32] <kayakyakr> note: If you have an LVM partition as your root and an active snapshot of said partition, your system will not boot.
[16:32] <kayakyakr> simple fix. lvremove /dev/ubuntu/rootsnapshot
[16:33] <kayakyakr> *headdesk*
[16:35] <AnGrYfUrBy> ihashacks, it works how did you do it !!!
[16:36] <patdk-wk> kayakyakr, heh, I never snapshotted root before
[16:36]  * patdk-wk isn't sure he ever put root on lvm before either
[16:37] <kayakyakr> it was a very nice way to make a backup. loving the hell out of LVM in general... but lordy lordy that was a huge gotcha.
[16:37] <patdk-wk> I used to, but my disk i/o is too high for lvm snapshots
[16:38] <kayakyakr> i could see that
[16:42] <ihashacks> AnGrYfUrBy: patience and diagnostic skills
[16:43] <AnGrYfUrBy> ihashacks, very wise and true words
[18:29] <pseudo> Hi guys - As I understand it, ubuntu can be setup to use SELinux as well as kvm/libvirt - My question is, how easy is it to use sVirt on ubuntu?
[18:30] <hallyn> stgraber: would it be more useful to you to have a lxc package made with the lxcapi, or just the modified upstream lxc tree?
[18:32] <hallyn> pseudo: there is an apparmor plugin for svirt which is enabled by default in ubuntu.  that obviously is easy to use.  To use it with selinux, there'd be much for you to do.
[18:35] <stgraber> hallyn: if the package has liblxc in a public path and has a -dev package with the header and .so, then I prefer a package as it's easier to have other people test. Otherwise upstream lxc tree is fine
[18:48] <hallyn> stgraber: I've not yet made a package, so it would have whatever you like, subject to my incompetence
[18:51] <stgraber> hallyn: ok, IIRC lxc-dev already exists and ships the .h, what needs changing is the location of the .so.* from /usr/lib/<multi arch>/lxc to /usr/lib/<multi arch>
[18:51] <hallyn> but the current -dev package depends on lxc, rather than vice versa
[18:52] <hallyn> stgraber: why move to /usr/lib/<multiarch>?  that's just the norm?
[18:55] <stgraber> hallyn: lxc-dev depending on lxc is fine. lxc ships the .so.x.x and lxc-dev ships the .so and .h
[18:55] <hallyn> ok but i assume nwo we'll want people to have liblxc.so without having all of lxc installed
[18:55] <stgraber> hallyn: yeah, I don't think we need to have a whole namespace for us when we only ship one library. That made sense when it was private, but if we start making it a public library, we probably should move it to the a path that ld actually knows
[18:56] <hallyn> ok
[18:56] <hallyn> i'll get a package into ppa
[18:56] <hallyn> (then i'd like to get the hooks done, then be done with lxc for a bit)
[18:56] <stgraber> hallyn: hmm, good point, I guess at some point people may want liblxc without lxc itself, though it'll be easy enough to split the packages to lxc, liblxc1 and liblxc-dev and make lxc depend on liblxc1 and liblxc-dev just depend on liblxc1
[18:58] <hallyn> ok, thx, ttyl
[19:02] <zapotah> hi anyone have any tips for compiling the e1000 driver on 12.04
[19:02] <zapotah> not too much experience compiling drivers
[19:03] <zapotah> propably missing some library but which one
[19:04] <smoser> zapotah, i'd suspect 'apt-get install build-essential linux-headers'
[19:04] <zapotah> nope
[19:05] <zapotah> already got those
[19:05] <smoser> well, what is going wrong then?
[19:05] <smoser> what are you doing?
[19:06] <zapotah> trying to compile the e1000 V8 driver from intel
[19:06] <zapotah> for lacp bonding
[19:07] <smoser> zapotah, that doesn't mean anything to me, sorry.
[19:08] <smoser> i'm guessing you downloaded a tar.gz file from somewhere (that i dont know where)
[19:08] <zapotah> from intel
[19:08] <smoser> extracted it, then did something, and got some errors.
[19:08] <zapotah> yes
[19:08] <zapotah> about to get to that :)
[19:08] <smoser> what errors
[19:08] <zapotah> http://pastebin.com/eDDffD96
[19:09] <zapotah> wrong one
[19:09] <zapotah> heres the complete output
[19:09] <zapotah> http://pastebin.com/7RtmTk1F
[19:12] <smoser> zapotah, it would seem to me that their kernel driver is looking for files that do not exist in more recent kernels.
[19:12] <zapotah> could be
[19:12] <smoser> ie, those files were in 3.0 headers, but do not exist in 3.2
[19:13] <smoser> "those files" == pm_qos_params.h
[19:13] <zapotah> i even tried symlinking the pm_qos.h to pm_qos_params.h
[19:14] <zapotah> which is found in /usr/src/longheaderdir/include/linux
[19:14] <smoser> where did you download this thing?
[19:14] <zapotah> but apparently something has changed noticeably since the drivers came out
[19:14] <zapotah> intel driver site
[19:14] <smoser> link?
[19:16] <zapotah> here you go: http://downloadmirror.intel.com/9180/eng/e1000-8.0.35.tar.gz
[19:19] <smoser> zapotah, http://downloadcenter.intel.com/Detail_Desc.aspx?DwnldID=9180
[19:19] <smoser> "NOTE: The e1000 driver is scheduled to change to a kernel only support model on/around Q2 of 2012.  If you have open issues you would like to see resolved, please submit them through http://sourceforge.net/projects/e1000/ as soon as possible."
[19:19] <smoser> it is around Q2 of 2012.
[19:19] <smoser> i'd suspect you might re-compile the driver from the kernel source instead.
[19:20] <zapotah> hmm
[19:26] <zapotah> wish i knew more about how to do this stuff... Frankly i have to ask, how do i compile the e1000 from the kernel source?
[19:26] <zapotah> i never suspected this would be so complicated :)
[19:27] <zapotah> im more of a networking guy and everything else is configured and tested working, and making the lacp bonding on the server side is the last thing that still needs to be done.
[19:28] <patdk-wk> heh?
[19:28] <zapotah> but apparently the old V7 e1000 driver that ships with 12.04 does not support lacp with the 82541PI nic
[19:30] <patdk-wk> yuk, that is a desktop chip, and limited to pci bus speeds
[19:30] <zapotah> i know :P
[19:30] <zapotah> but it should serve its purpose
[19:32] <patdk-wk> oh, I just found it odd lacp wasn't supported, but then I never tried it with a pci based nic
[19:32] <zapotah> probly with a newer card with e1000e it would work
[19:32] <zapotah> however...
[19:32] <zapotah> :)
[19:41] <zapotah> this problem was stupid beyond belief
[19:42] <zapotah> ill blame it on the 14hr workday
[19:42] <zapotah> it works now
[19:42] <zapotah> as it should
[19:42] <zapotah> as it did from the beginning if i had paid attention
[19:43] <zapotah> now off to sleep, and sorry for the undue troubles :)
[20:22] <JonEdney> Are there any memory monitoring apps out there for servers?  I have a server monitoring app monitoring my system, and the memory increases everyday and nothing is on it but Apache, PHP, MySQL and is not being used.
[20:26] <hallyn> stgraber: lxc-init should technically be under libexec, right?
[20:27] <hallyn> hm, i see, that doesn't exist :)  nm
[20:28] <guntbert> JonEdney: for a quick look use top (or htop)
[20:28] <JonEdney> Thanks guntbert, not familair ill check it out
[20:29] <stgraber> hallyn: right ;) I actually have a libexec on my system but that was a workaround I used when packaging lttng (system tracing) where they need to run one daemon per architecture, spawned by their library (so wanted it out of the path and in a multi-arch path)
[20:29] <stgraber> but yeah, usually debian/ubuntu don't have libexec
[20:29] <guntbert> JonEdney: both show actually the same things, htop is "nicer" and better adjustable to your needs
[20:33] <JonEdney> Is it common to have multiple mysqld and apache processes?
[20:34] <bctrainers> how many is multiple to you? :p
[20:34] <bctrainers> Mine range from five upwards to 25+ depending on server :)
[20:34] <JonEdney> 10-15 of each?
[20:35] <JonEdney> Alright.
[20:35] <JonEdney> I installed a RevealCloud app on the server to monitor, it's running like 15-20 processes also, that may be eating my memory more and more.
[20:36] <bctrainers> just depends on your my.cnf and apache config settings for how many little workers it spawns in :p
[21:40] <erichammond> What's the (relatively new) technical term used for a one time use token that is provided to a user to allow them to perform an action?
[21:41] <erichammond> They used to be called "cookies" a couple decades ago (before the web took this over)
[22:43] <jkyle> I'd like to install the 3.4.x kernel on 12.04
[22:43] <jkyle> I found these debs: http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.4-precise/
[22:43] <jkyle> is that the recommended method? I could roll my own
[22:44] <Daviey> jkyle: Probably better to ask in ubuntu-kernel.. they are backporting 3.4 to Precise officially.
[22:47] <adam_g> w/in 5
[22:47] <Daviey> adam_g: l/ose 5 :)
[22:48]  * RoyK says hi from 79ºN
[22:48] <adam_g> :(
[22:49] <jkyle> Daviey:thanks! going to take that open vswitch module for a spin
[22:52] <Daviey> jkyle: It's in the Precise version.. What issues are you seeing?
[22:53] <Daviey> jkyle: BTW, i'd really like to hear your thoughts and plans on what you plan to do with openvswitch.