[02:38] <jMyles> Is there a quickstart to building a local cloud with oneiric?
[02:46] <jtran> hey all.  i'm having a problem after install nbd-server pkg, it says nbd doesn't exist in /proc/modules.  it's the exact same problem as described in this debian bug report:  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=628389
[02:47] <jtran> however, how do i get that equiv pkg for ubuntu?
[02:48] <jtran> i installed the debian pkg from above but i'm just checking to see if this is something that will make it upstream into oneirc
[02:49] <twb> Installing packages from Debian will explode your system
[02:49] <jtran> i really needed a fix for now :(
[02:49] <jtran> and it appeared to work
[02:50] <twb> You say it's the same problem -- does that mean you have a kernel you compiled nbd =y in?
[02:50] <twb> Or do you simply mean that it is the same symptoms
[02:51] <jtran> i mean nbd just works now
[03:12] <philipballew> Would it be possible to set up ssh on port 4444
[03:12] <twb> philipballew: I'm not stopping you
[03:13] <philipballew> but would something? twb
[03:13] <philipballew> its the only open port at the school i go to.
[05:30] <airtonix> i'm looking for a good article about reasons why it's a good idea to notify stakeholders a week in advance about server upgrades that would take the server offline for any length of time
[05:31] <airtonix> i'm trying to get it across to my boss that it's simply not a good idea to just perform the upgrade right now without warning clients
[05:40] <Myrtti> airtonix: when the upgrade goes south, it is more painful to inform
[05:40] <Myrtti> have personal experience as a client
[05:41] <airtonix> Myrtti: i try to perform the upgrade on a clone of the server first
[05:42] <Myrtti> and if you do warn beforehand, don't assume clients read your blog. send an email with bcc instead, or use mailchimp or similar
[05:44] <Myrtti> have experience on getting the email afterwards with every customer in to-field
[05:45] <Myrtti> when things didnt work for some, they hit Reply All-button and vented
[05:45] <Myrtti> nasty
[05:53] <airtonix> Myrtti: yep we send a notice out with mailchimp
[06:19] <eagles0513875_> hey guys i have a quick question about xen and 11.10
[06:19] <eagles0513875_> seeing as there is no xen kernel can someone still run xen paravirtualized guests?
[07:26] <RoyK> eagles0513875_: iirc all the needed Xen patches are in the default kernel, both host and guest
[07:30] <blinkiz> hello. I have a new Intel 204 chipset server here. First tried fake raid but ubuntu does not see any hard drives. why? Anyway, am struggling to get software raid to work. The machine will not boot! It is like grub is not found when UEFI tries to hand over the control.
[07:31] <eagles0513875_> ty RoyK :)
[07:36] <smb> eagles0513875_, Using the -virtual kernel has the advantage of having the pv net and disk driver built-in instead of being modules
[07:37] <eagles0513875_> virtual kernel???
[07:37] <eagles0513875_> ahh
[07:38] <smb> There is a linux-kernel.*-virtual. Its limited down to essential modules for virtual guests. And with oneiric there is now also an extra modules packages that ships the rest of them in case they are needed.
[07:41] <dhon_> hi all
[07:41] <eagles0513875_> smb: im doing some testing to see how i can interface orchestra into an existing debian squeeze xen setup with already existing xen guests
[07:42] <dhon_> I'm having trouble setting up a 10.04 install using software raid - specifically when it comes to installing grub
[07:43] <dhon_> I've created a raid-1 device with a partition table which has the following partitions: boot(512mb),/,/usr,/var,/home
[07:44] <dhon_> these appear at md0p1, md0p5, md0p6, md0p7 & md0p8 respectively
[07:44] <dhon_> (boot is the only "physical" partition)
[07:44] <smb> eagles0513875_, Ah. Well I am not much help with orchestra. But I would try to get the -virtual kernel packages installed. The generic-pae and or server image would work as well but the blkfront and netfront do not get autoloaded.
[07:45] <dhon_> grub setup failed during install of ubuntu, and now I'm not sure where I'm supposed to be installing it
[07:45] <eagles0513875_> ok
[07:45] <eagles0513875_> dhon_: the master boot record
[07:45] <dhon_> (sorry for talking over you guys)
[07:45] <eagles0513875_> dhon_:  your ok just brainstorming something im working on
[07:46] <eagles0513875_> dhon_: did you install using a pen drive
[07:46] <dhon_> yes, but using the desktop installer actually
[07:46] <eagles0513875_> cuz i have had the issue where for some reason i install ubuntu on my hard disk it installs grub to the MBR of the pen drive
[07:46] <dhon_> hmm strange
[07:46] <eagles0513875_> dhon_: try this put the pen drive back in and boot with it in i just want to confirm that it actually installed grub to the pen drive instead
[07:47] <dhon_> which MBR should it be installed to? /dev/sda & /dev/sdb, or /dev/md0 maybe?
[07:47] <eagles0513875_> dhon_: so you have software raid setup im guessin
[07:47] <dhon_> err, well I did have
[07:47] <eagles0513875_> O_O
[07:48] <dhon_> I'm currently reattempting from the start
[07:48] <dhon_> it's a new install
[07:48] <eagles0513875_> i would reformat everythign blowing away the raid setup etc
[07:48] <dhon_> yep
[07:48] <dhon_> is it okay to have boot on md0p1?
[07:48] <eagles0513875_> me grub using the net installation off a pen drive got installed to the right location
[07:48] <dhon_> grub knows about software raid devs?
[07:49] <eagles0513875_> when installing when it comes to the partitioning that is when you setup software raid if your using the server installer
[07:49] <dhon_> the strange thing is that I could chroot into the raid install, but grub-install would fail
[07:50] <dhon_> saying something about a missing mapping for md0
[07:50] <eagles0513875_> your best i think starting from scratch
[07:50] <dhon_> well, it spat an error about that - it may have succeeded afterwards, it was unclear
[07:51] <eagles0513875_> i dunno if i were in ur position i woudl do a clean install from scratch
[07:51] <dhon_> I appreciate what the grub devs have accomplished, but I've never had much luck with the software :/
[07:51] <dhon_> yeah maybe I should get the server installer
[07:52] <dhon_> desktop installer doesn't mention software raid
[07:52] <dhon_> I just manually set up the raid and selected those partitions during install
[07:53] <eagles0513875_> dhon_: server is where its at
[07:53] <dhon_> is it better to have each partition as a separate raid array?
[07:53] <dhon_> and format the raw devices (ie no partition table)
[07:56] <eagles0513875_> dhon_: no one raid array is best i believe
[07:56] <eagles0513875_> then again it depends on how many hard disks you have
[07:56] <eagles0513875_> my server has 2 so i have setup raid 1 mirroring
[07:56] <dhon_> 2 disks, raid1 - yep
[07:56] <eagles0513875_> dhon_: i have a single raid array
[07:57] <eagles0513875_> !raid
[07:57] <eagles0513875_> dhon_: ^ the first link would be of use how to set up raid :)
[07:58] <dhon_> fake raid is not the same thing, correct?
[08:01] <eagles0513875_> no
[08:01] <eagles0513875_> you said you have software raid the first link is the onlyone you need
[08:01] <dhon_> yep
[08:01] <eagles0513875_> dhon_: ill be back gonna have some breakfast really fast
[08:01] <dhon_> okay, thanks
[08:04] <jamespag`> morning all
[08:06] <dhon_> evening :)
[08:33] <koolhead17> hi all
[08:36] <lynxman> hey koolhead17, morning jamespage
[08:40] <eagles0513875_> dhon_: you managed?
[08:43] <dhon_> eagles0513875_: haven't progressed much, just been reading
[08:43] <eagles0513875_> ok
[08:43] <eagles0513875_> dhon_: try get started i have an hr before i leave so i can help u while im still here
[08:46] <dhon_> cheers, I don't have the server installer, but I'll give it another try with the desktop
[08:47] <dhon_> should I use the MBR, or GUID partion table on the raid array?
[08:48] <eagles0513875_> mbr is the location on the hard disk
[08:49] <eagles0513875_> !guid
[08:49] <eagles0513875_> dhon_: that has something to do with grub
[08:49] <eagles0513875_> grub 2 i believe uses GUID's for the partitions
[08:49] <dhon_> I'm using Disk Utility to create the partitions on the raid array
[08:50] <dhon_> it says I can have the following parition table types: MBR, GUID Partition Table, None, Apple Partition Map
[08:51] <eagles0513875_> never used the disk utility i usually let it setup the partitions on its own
[08:51] <eagles0513875_> you can always just setup first with no raid then seutp a raid array via the command line
[08:52] <dhon_> as in, do a normal install on a single disk and then use those partitions to create a raid array?
[08:53] <eagles0513875_> correct
[08:54] <dhon_> well that sounds like a good option
[08:59] <eagles0513875_> dhon_: one thing i dunno how to do is set up raid then via command line
[08:59] <eagles0513875_> i set it up during the ubuntu server install
[08:59] <eagles0513875_> dhon_: question though why are you setting up ur server with a gui
[09:00] <dhon_> 'cause I have the install-thumb-drive already
[09:00] <dhon_> I was just going to disable gdm
[09:00] <eagles0513875_> dhon_: ok
[09:00] <eagles0513875_> just wondering is all you will also need to install the server kernel
[09:01] <dhon_> hmm, okay
[09:01] <dhon_> what are the differences? scheduler?
[09:01] <dhon_> power management maybe?
[09:02] <dhon_> it's really sounding like I should just get the server installer
[09:02] <dhon_> might save me another day of googling
[09:02] <dhon_> :)
[09:02] <dhon_> it's frustrating when things are so close, but they keep eluding you
[09:04] <dhon_> will 12.04 be LTS?
[09:06] <ersi> dhon_: Yes.
[09:06] <koolhead17> lynxman: hellos
[09:07] <dhon_> ersi: cool, thanks
[09:07] <ersi> dhon_: Basically, every two years there's an LTS. 8.04 10.04 12.04 :)
[09:07] <koolhead17> lynxman: i managed keystone/dashboard/nova without any effort working on virtualbox. :P
[09:08]  * dhon_ slaps forehead
[09:08] <lynxman> koolhead17: good ;)
[09:09] <eagles0513875_> dhon_: i can tell u my server is 10.04 and its rock solid. im only trying out 11.10 on a 2tb hard disk i have in my desktop to test out some cloud solutions
[09:09] <koolhead17> lynxman: how have you been? how was party :D
[09:09] <koolhead17> hellos Daviey jamespage
[09:09] <lynxman> koolhead17: quite busy, party was good :)
[09:10] <lynxman> Daviey: hey there o/
[09:10] <koolhead17> nice :)
[09:10] <jamespage> morning koolhead17
[09:10] <Daviey> hey lynxman
[09:10] <Daviey> hey koolhead17
[09:10]  * jamespage just waves at everyone
[09:10] <koolhead17> jamespage: let me know if you have time will ping you for the dbconfig-common stuff
[09:19]  * lynxman feels waved
[09:21] <Daviey> jamespage: I think fop is begging for a merge :)
[09:21] <jamespage> Daviey: oh yes
[09:22] <jamespage> might even get away with a sync TBH
[09:22]  * jamespage is looking at merges and syncs this afternoon
[09:22] <memoryleak> hi, I get  PHP Startup: Unable to load dynamic library '/usr/lib/php5/20090626/sqlite.so' - /usr/lib/php5/20090626/sqlite.so
[09:22] <Daviey> jamespage: awesome
[09:22] <\sh> precise already open?
[09:22] <memoryleak> I used to look for it with find / -name sqlite.so but nothing found on the system
[09:23] <dhon_> okay, I need dinner, I'm going to attempt this raid thing again tomorrow - thanks for the help eagles0513875_
[09:23] <eagles0513875_> no problem dhon_
[09:23] <Daviey> \sh: oh yes
[09:23] <\sh> Daviey, very good...thx :)
[09:23] <Daviey> \sh: What is on your roadmap for this cycle?
[09:25] <\sh> Daviey, FAI 4.x + Puppet + DC² + DC² Node Classifier eventually getting rid of NFS for FAI
[09:25] <Daviey> \sh: ooooo, nice
[09:26] <Daviey> lynxman: How is mcollecitve looking? :)
[09:26] <\sh> Daviey, do you know if anyone has freeipa on the roadmap for debian/ubuntu?
[09:27] <Daviey> \sh: ISTR there was some discussion a year ago, but i don't think it went anywhere :(
[09:29] <\sh> Daviey, eventually someone can put it back on the table during upcoming UDS...it looks like a good thing to have it adjusted for debian/ubuntu
[09:31] <koolhead17> Daviey: https://bugs.launchpad.net/keystone/+bug/855467  needs to be closed now!!
[09:32]  * koolhead17 waves to kim0 
[09:33] <soren> Daviey, jamespage: Do we use fop for anything in server land?
[09:33] <jamespage> soren: Document generating I think - erlang
[09:33] <soren> jamespage: Ah.
[09:33] <soren> jamespage: Wow, so it's in main now? It used to be multiverse or something.
[09:33] <lynxman> Daviey: mcollective is looking like a long and windy road :) I need to finish the whitepaper first then make sure the activemq merge is working then start on mcollective
[09:34] <jamespage> soren: yep - I pushed it through MIR during oneiric
[09:34] <soren> jamespage: Moved to universe during Intrepid. How about that.
[09:35] <lynxman> Daviey: so I would say expect this to take me the best part of this week
[09:38] <Daviey> wow
[09:46]  * kim0 waves back to koolhead17 
[10:38] <zul> jamespage: fop?
[10:39] <jamespage> zul: yes
[10:39] <jamespage> I'll do it in a bit
[10:39] <zul> jamespage: which is....
[11:13] <B0bby> hello
[11:15] <B0bby> I have some issues in my openvpn client server connection since I have tried to tuning the performance... before it works but since I added some a parameters now nothing works !
[11:16] <B0bby> I got this error messages in the server side :
[11:16] <B0bby> Mon Oct 17 10:08:36 2011 VPN-Client/196.217.199.157:51446 Bad LZO decompression header byte: 0
[11:16] <B0bby> Mon Oct 17 10:08:36 2011 VPN-Client/196.217.199.157:51446 FRAG_IN error flags=0xffffffff: FRAG_TEST not implemented
[11:16] <B0bby> and  Mon Oct 17 10:12:54 2011 ROUTE: route addition failed using CreateIpForwardEntry: Accès refusé.   [status=5 if_index=20]
[11:16] <B0bby>  Mon Oct 17 10:12:54 2011 Route addition via IPAPI failed [adaptive] Mon Oct 17 10:12:54 2011 Route addition fallback to route.exe
[11:16] <B0bby> in the client side
[11:17] <pmatulis> B0bby: go back to your original config and add one parameter at a time
[11:18] <B0bby> pmatulis, I have added just two parameters : Lzo compression and fragment size
[11:18] <B0bby> nothing else
[11:20] <ersi> B0bby: Remove them and try again, then.
[11:21] <B0bby> it' seems that it's a psychological solutions rather than technical ones :P
[11:28] <jamespage> zul: formatted output protocol == document generation
[11:29] <zul> jamespage: ah
[11:29] <jamespage> zul: I think erlang uses it - hence in main
[11:35] <jamespage> hmmm: new depends - yuck
[12:07] <jamespage> Daviey: how badly do we want erlang docs?  http://tinyurl.com/6l4c53s
[12:07] <jamespage> bug 876413
[12:08] <soren> erlang is self-explanatory. No need for docs. :)
[12:17] <lynxman> soren: that phrase... hurts in unknown ways :)
[12:48] <zul> jamespage: not at all :)
[12:52] <zul> bah
[12:53] <jamespage> humbug
[12:59] <hallyn> jamespage, could you sponsor http://people.canonical.com/~serge/bacula-gawk.debdiff for bug 645082 ?
[12:59] <jamespage> morning hallyn
[13:03] <jamespage> hallyn: looking now
[13:07] <hallyn> jamespage, thanks :)
[13:07] <hallyn> happy monday morning to you! :)
[13:08] <zul> jamespage: erlang is probably not the first thing you want to merge btw ;)
[13:08] <hallyn> no you want to start with libvirt
[13:08] <jamespage> zul: not touching it for the moment
[13:08] <hallyn> it's an easy one
[13:11] <zul> hallyn: i was going to poke it with a stick
[13:12] <hallyn> zul,  you were? cool
[13:12] <hallyn> I was considering doing a few more pushes of fixes which are NOT yet in 0.9.6 first,
[13:12] <hallyn> so that when doing the merge we don't have to think about it
[13:12] <hallyn> (like the fix for lvm.conf having lvm_commands=1)
[13:18] <eagls0513875> !upgrade
[13:24] <jamespage> is vmbuilder still the preferred tool for creating vm's without building?
[13:24] <jamespage> /building/running
[13:27] <hallyn> jamespage, I think that's supposed to become live-build
[13:27] <jamespage> hallyn: so is that a re-name or a different tool?
[13:27] <hallyn> different tool
[13:28] <jamespage> ah - I see
[13:28] <hallyn> actually I need to do some testing of those - I haven't used them in something like a year
[13:29] <hallyn> I wonder what smoser uses to build AMIs
[13:29] <smoser> different tool.
[13:30] <smoser> cloud-images now uses live-build.
[13:30] <lynxman> hallyn: the power of his mind :)
[13:30] <smoser> but jamespage i would really suggest using cloud-images to bootstrap you.
[13:30] <smoser> download image, use it
[13:31] <hallyn> smoser, good suggestion, although i do think it'd be good for everyone to play with live-build a bit, both to shake out bugs and to know how to use it off top of their heads when they need to in a pinch
[13:32] <smoser> hallyn, i largely disagree.
[13:32] <smoser> i think building images is stupid
[13:32] <smoser> a waste of time that will result in re-inventing something that is already done.
[13:32] <smoser> and if the images don't fit your needs, then we should improve them.
[13:32] <hallyn> smoser, and i disagree, bc i've been part of a project where we needed to build a minimal, *minimal* image
[13:33] <smoser> i find effort on minimal to be a waste of time.
[13:33] <hallyn> smoser, i think it's wrong to discourage people from becoming competent with tools.
[13:33] <hallyn> smoser, sometimes you don't get to decide
[13:33] <smoser> oh, i decide!
[13:33] <hallyn> what's the smallest image you have?
[13:33] <hallyn> :)
[13:34] <smoser> 650M root filesystem.
[13:34] <smoser> i realize its not small
[13:34] <hallyn> not just not small, completely unsuitable for the use-case we had
[13:34] <smoser> i just think that trading 400M for 1 hour of your time is useful.
[13:34] <smoser> and 1 hour is *very* conservative.
[13:34] <hallyn> oh, now i get it
[13:34] <hallyn> you're looking ofr job security :)
[13:35] <smoser> what is the use case ?
[13:35] <smoser> remember the second part of my suggestion to not use live-build , was "we should make the images better"
[13:37] <smoser> generally, my feeling is that live-build is an installer. if you build atop live-build, you're going to then add some little fixes for your little environment.  and then you're going to tell people to use your tools, and you're going to get your tools into universe and or main, and then we're going to one other tool that fixes quirks found after installing a bunch of packages.
[13:37] <smoser> i dont like installers.
[13:38] <smoser> i think the cobbler-devenv avoids being an installer nicely, by just letting the server install iso do the install and preseed it.
[13:38] <hallyn> livebuild is pretty darned flexible
[13:38] <zul> soren: did you do the drawings yourself?
[13:38] <hallyn> anyway, shock, smoser and i disagree on something :)
[13:39] <smoser> livebuild is very flexible.
[13:39] <soren> zul: No, I had a professional art company do them. It only cost a million dollars. I thought it was a steal.
[13:39] <smoser> it will allow you to stuff your little "fix up this quirks" perfectly well outside of it.
[13:39] <smoser> and your quirks wont work in other cases
[13:39] <zul> soren: good to see the startup capital go to use
[13:39] <soren> zul: Oh, those graphs? YEah, I did those myself.
[13:41] <smoser> what drawings is this ?
[13:41] <smoser> so i'm willing to accep that the cloud-images dont work for *everything*, but we really want them to work for lots of things. (just as we want the ISO to work for lots of things).
[13:42] <hallyn> smoser, what is supposed to be the most discoverable way to get the latest cloud img?
[13:42] <smoser> if you're going to poke around with live-buid, then please, when you find a little quirk in a package install that you ahve to work around,  *please* get your changes upstream.
[13:42] <smoser> i would need to know what "discoverable" means.
[13:43] <hallyn> easiest url to remember, or where to get a link
[13:43] <hallyn> i.e. cloud.ubuntu.com -> click top link
[13:43] <hallyn> 100% agreed on get your changes upstream, of course
[13:44] <smoser> cloud-images.ubuntu.com
[13:44] <smoser> you want ami's ?
[13:44] <smoser> or downloads.
[13:45] <hallyn> downloads
[13:46] <hallyn> ami's are efine for us, but some ppl want images without needing the euca tools
[13:47] <smoser> hallyn, browsing cloud-images.ubuntu.com is probably the best way for that.
[13:48] <smoser> but we publish data that can easily be queiried
[13:48] <smoser> $ echo $(ubuntu-cloudimg-query p --format "%{url}" daily)
[13:48] <smoser> https://cloud-images.ubuntu.com/server/precise/20111017/precise-server-cloudimg-amd64.tar.gz
[13:49] <hallyn> ok.  for some reason i'm averse to - in a url i guess :)  but maybe i'll try using that more
[13:51] <dnmons> Postfix is started before Dovecot in 11.10. Something which causes it to crash when Dovecot is providing SASL authentication to it. http://ubuntuforums.org/showthread.php?p=11354218 Any idea on how to fix it?
[13:53] <jamespage> smoser: so if someone wanted to automate producing a virtual machine image for use on a desktop/laptop under KVM does that advice still hold true
[13:53] <jamespage> i.e. start with the cloud image?
[13:53] <smoser> i would suggest 2 paths:
[13:54] <smoser> or... 3
[13:54] <smoser> or... 2
[13:54] <smoser> a.) cobberl-devenv like "seed the installer" (i think security team has a tool that does this too... hallyn uses it, mdeslaur recently updated it)
[13:55] <smoser> b.) use cloud-image full disk image as a starting point
[13:55] <smoser> what kind of image is this that you're wanting to use ?
[13:55] <jamespage> someone == gnuoy who I noticed joined a while back
[13:56] <gnuoy> smoser, I'm just after vanilla natty to test installing openerp
[13:56] <zul> mdeslaur: can i take cifs-utils off your hands
[13:57] <mdeslaur> zul: what, for the dev release?
[13:57] <mdeslaur> zul: yes, please do
[13:57] <gnuoy> the serverguide for natty suggests using vmbuilder
[13:57] <zul> mdeslaur: yes k
[13:58] <jamespage> dnmons, do you happen to know if that is a regression from natty?
[13:58] <smoser> gnuoy, i'd be honored if you would take the time to try to follow https://help.ubuntu.com/community/UEC/Images#Ubuntu_Cloud_Guest_images_on_Local_Hypervisor_Natty_onward
[13:58] <hallyn> Daviey, so are you going to push spice-gtk now for p?
[13:58] <smoser> and tell me what did not work.
[13:59] <gnuoy> smoser, sure thing :-) (although it'll have to be after I've finished the day job)
[14:07] <smoser> mdeslaur, i think that vm-tools needs to be its own thing. and i need to improve it so it doesn't have to remaster an ISO to give a seed file.
[14:07] <mdeslaur> smoser: feel free to fork it
[14:08] <dnmons> jamespage: yes, it is.
[14:09] <mdeslaur> smoser: it's very oriented to the security team, it would be great to have a tools that's more generic
[14:09] <mdeslaur> smoser: (and isn't a 800 line bash script 90% composed of bad hacks)
[14:09] <dnmons> jamespage: same system on natty did not have this problem. dovecot started about 8 seconds before postfix. now dovecot is started 2–4 seconds after postfix. which gives postfix a headache.
[14:10] <hallyn> all right i think i'll sync qemu 0.15.0 today (maybe into tomorrow).
[14:10] <mdeslaur> smoser: so, I tried to use your external initrd idea so we didn't need to remaster the iso, and wasn't successful with virt-install
[14:10] <smoser> then you did something wrong :)
[14:10] <smoser> somehow i was successful.
[14:11] <smoser> i think its a matter of getting the kernel params and kernel and intiramfs right.
[14:11] <smoser> otherwise i dont know how it could go wrong.
[14:12] <mdeslaur> smoser: weren't you doing it with kvm directly?
[14:12] <smoser> yeah
[14:13] <smoser> but you can pass kernel and ramdisk and kernel params through libvirt
[14:13] <jamespage> dnmons: not ignoring you - just looking at what changes landed in oneiric for upstart and dovecot
[14:14] <mdeslaur> smoser: I haven't found a way to do that with virt-install
[14:14] <smoser> oh... i didn't know you were using that.
[14:14] <smoser> what value does it provide you?
[14:14] <mdeslaur> smoser: if you specify a kernel directly, it won't let you specify a cd
[14:16] <mdeslaur> smoser: anyway, I spent some time on it, but didn't manage...it would be nice if we didn't need to rebuild the iso so I hope I got it wrong
[14:16] <smoser> why are you using virt-install ?
[14:17] <mdeslaur> smoser: I don't know...what should I be using?
[14:17] <smoser> i just assumed you were creating a disk image, creating a cd-rom, and laucnhing a libvirt instance with those attached.
[14:18] <smoser> (via specifying your own libvirt xml)
[14:19] <smoser> woot! precise EC2 image boots!
[14:19] <smoser> (i realize it is basically oneiric, but still...)
[14:19] <jamespage> dnmons: I think I can see what change caused the issue
[14:19] <jamespage> dovecot and postfix now start/stop on the same runlevels
[14:19] <jamespage> whereas in natty dovecot started very early
[14:21] <jamespage> jhunt: around? need a little upstart guidance
[14:21] <jhunt> jamespage: yup
[14:22] <jamespage> jhunt:great
[14:22] <jamespage> so this is the situation - dovecot+postfix install
[14:22] <jamespage> postfix authenticates through dovecot using SASL
[14:22] <hallyn> zul, jdstrand, does the last comment in bug 874350 make sense to you?
[14:22] <jamespage> so is dependent on dovecot running
[14:23] <mdeslaur> smoser: I suppose that would be the next step to get rid of the iso rebuilding
[14:23] <zul> hallyn: comment #2?
[14:23] <jamespage> in oneiric a change landed which switched dovecot's start on from rc-sysinit to runlevel [2345]
[14:23] <mdeslaur> smoser: let me know if you get that working...and what did you mean by "vm-tools should be it's own thing"?
[14:24] <hallyn> zul, yeah
[14:24] <zul> hallyn: umm...no :)
[14:24] <jamespage> jhunt: which I think is causing it to start later in the boot process alongside postfix which is a standard init script
[14:24] <smoser> mdeslaur, can i 'apt-add-repository some-thing && apt-get install vm-tools' &
[14:24] <smoser> s/&/?/
[14:25] <jamespage> jhunt: postfix borks as a result - is there any way to ensure that dovecot starts up before postfix bearing in mind they don't both use upstart?
[14:25] <smoser> basically i think what you, jdstrand and kees created is useful to a much larger audience than is going to find it or use it under a subdir of a bzr repo.
[14:25] <hallyn> zul, ok, i'll roll with it then :)
[14:25] <hallyn> thx
[14:27] <mdeslaur> smoser: I think a larger audience should be using something that was designed better, and doesn't depend on all the security team environment
[14:27] <mdeslaur> smoser: but yes, I do think we should have a tool to easily preseed test vms
[14:27] <mdeslaur> smoser: maybe testdrive should gain preseeding support or something
[14:27] <alex88> hi guys, i'm using csf firewall and i've setup some prerouting dnat rules to redirect some ports to vm, is possible to  use something different then prerouting and dnat to achieve the same result? so the firewall can filter/block also those connections?
[14:28] <kirkland> interesting
[14:28] <patdk-wk> alex88, I think that greatly depends on what you call your vm
[14:29] <alex88> patdk-wk: sorry, virtual machine
[14:29] <patdk-wk> I know that
[14:29] <patdk-wk> but what kind :)
[14:29] <alex88> you mean like kvm?
[14:29] <patdk-wk> xen, kvm ,virtualbox, or a vps, lxc, ...
[14:29] <alex88> kvm
[14:29] <patdk-wk> I would just use proxyarp
[14:30] <alex88> btw, command to dnat i used is /sbin/iptables -t nat -A PREROUTING -p tcp -i vmbr0 --dport 2223 -j DNAT --to 192.168.1.3:22 #SSH
[14:30] <alex88> you mean like setup proxyarp that redirects packets after firewall?
[14:30] <patdk-wk> no, that is not proxyarp then
[14:31] <patdk-wk> I would use proxy-arp not whatever it is your thinking of, or think I'm saying
[14:31] <alex88> sorry but i haven't heard about that
[14:31] <patdk-wk> proxyarp would turn your firewall into a transparent system, as far as your vm knows
[14:31] <alex88> well, i have multiple vm, and each port goes to different vm
[14:32] <patdk-wk> in other words, you don't have real ip's for your vm's?
[14:32] <van7hu> hi all
[14:32] <alex88> exactly
[14:32] <van7hu> how could I check if samba4 installed?
[14:32] <patdk-wk> then your stuck with dnat
[14:32] <van7hu> and running?
[14:32] <alex88> just 1 ip for the dedi
[14:32] <alex88> yup i think so
[14:32] <alex88> i just wanted to let the firewall intercept packets before going to vm
[14:33] <alex88> so if an ip get banned, it blocks also those packets
[14:33] <patdk-wk> it should, unless you really really screwed it up
[14:33] <jamespage> dnmons: any chance you could raise a bug for that dovecot/postfix issue - it looks like a regression
[14:34] <alex88> patdk-wk: would you do a test to check where the rule for your ip goes and check why packets are not blocked?
[14:35] <patdk-wk> I dunno what your talking about, I don't have root on your system, so how could I check?
[14:35] <alex88> patdk-wk: you just try to hit a blocked port 5 times and i'll do an iptables -L | grep yourip to check where it goes, because if i do from here i won't have access to ssh anymore
[14:36] <patdk-wk> still makes no sense at all to me
[14:36] <alex88> ok don't worry
[14:37] <patdk-wk> if it's blocked on the firewall, it won't make it to the vm
[14:37] <patdk-wk> unless you really really screwed it up
[14:37] <alex88> it does, maybe becuse packets are prerouted before passing to other rules
[14:37] <alex88> don't know, but i can access services into vm and not on ssh on main host
[14:37] <patdk-wk> prerouting doesn't bypass the normal tables
[14:38] <patdk-wk> and you only prerouted a single port
[14:38] <patdk-wk> but then, you only showed a SINGLE LINE out of the whole firewall config
[14:38] <alex88> well, 15 ports
[14:38] <patdk-wk> how should I know
[14:38] <alex88> i just have multiple lines like that
[14:38] <alex88> with different ports/ip, but it's the same
[14:38] <lynxman> patdk-wk: you can't make out the firewall config by just reading one line? Damn the Matrix :)
[14:38] <patdk-wk> hehe :)
[14:39] <patdk-wk> iptables -F, now that I can from a single line :)
[14:39] <jhunt> jamespage: do we know why that change to dovecot was made? I think now dovecot + postfix will be starting "around the same time" (ie "runlevel" - racy!)
[14:39] <alex88> ;)
[14:39] <patdk-wk> dovecot was changed from old init.d to upstart
[14:39] <patdk-wk> sounds like postfix is still upstart
[14:39] <patdk-wk> postfix is still init.d I mean :)
[14:39] <jhunt> jamespage: is there any reason postfix can't be upstartified? seems a little odd that a main pkg is using sysv and a dependent pkg in universe is using Upstart...?
[14:40] <jamespage> jhunt: bug 820685 was the reason it was changed
[14:41] <jamespage> jhunt: I've not looks at upstart for postfix
[14:41] <jamespage> both packages are in main I thinl
[14:42] <zul> jamespage: it hasnt been converted yet
[14:42] <zul> jamespage: feel free ;)
[14:43]  * jamespage takes a deep breath
[14:43] <patdk-wk> my dovecot/postfix on lucid are both upstart'd
[14:43] <patdk-wk> have an issue, but rarely, but I haven't made postfix depend on dovecot
[14:43] <zul> jamespage: the ebox people might have done it though
[14:45] <jhunt> jamespage: It'd need testing of course, but "start on starting rc" (for dovecot) should give you the behaviour you're after".
[14:45] <jamespage> jhunt: its difficult because start of [2345] is actually correct
[14:46] <jamespage> its only in the instance here postfix+dovecot are deployed together with SASL auth that we get the issue
[14:47] <jamespage> dnmons: please see jhunts comment above - that should sort you out (but please still raise a bug!)
[14:48] <jhunt> changing dovecot to be "start on starting rc" should work in both cases.
[14:50] <jamespage> jhunt: so that should start dovecot just before the runlevel itself starts?
[14:50] <jamespage> (just trying to get my head around this)
[14:51] <jhunt> jamespage: no - that will start dovecot "just before" the SysV scripts (including postfix) are started.
[14:52] <jhunt> jamespage: since 'rc' starts on runlevel anyway, you get what you want :)
[14:52] <jamespage> jhunt: so 'start on starting rc RUNLEVEL=[2345]'?
[14:53] <jhunt> jamespage: event sequence will thus effectively be: runlevel, dovecot, rc, postfix.
[14:53] <jhunt> jamespage: right.
[14:54] <jamespage> jhunt: OK - thanks for the advice
[14:54] <jhunt> jamespage: np.
[15:03] <madalin> Nagios related question. The following scenario: I have server1, connecting to server2. Server2 has an internal network. I'm trying to get server2 to ping an INTERNAL ip on the network and passing its results to server1. I'm reading about check_nrpe and check_ping, but im way too lost. Anyone can help me out ?
[15:05] <alex88> patdk-wk: after banning an ip address fw adds the ip in input chain but i think that it's processed after the prerouting table is processed, right?
[15:06] <patdk-wk> yep
[15:07] <patdk-wk> but INPUT is only processed for packets to that machine
[15:07] <patdk-wk> so you need to to go to FORWARD chain, for vm's
[15:08] <patdk-wk> technically, you probably want it to go to both input and forward
[15:09] <patdk-wk> but it would be so much easier to put them in a seperate block list, that both input/forward used
[15:09] <alex88> yeah, like if it pass the input table, it's forwarded
[15:09] <alex88> right?
[15:10] <zul> lynxman: ping
[15:11] <zul> lynxman: er unping
[15:11] <lynxman> zul: awww
[15:12] <zul> lynxman: i was going to take rabbitmq-server off your hands but thought differently ;)
[15:12] <zul> lynxman: but i can upload puppet for you if you want
[15:12] <van7hu> !samba
[15:13] <lynxman> zul: sounds good :)
[15:14] <zul> lynxman: if you can point it to me the stuff you did for it
[15:14] <Jeeves_> Hi!
[15:14] <Jeeves_> How do I enable ubuntu-vm-builder (in Lucid) to create Oneirik images?
[15:15] <Jeeves_> Should I just copy /usr/share/pyshared/VMBuilder/plugins/ubuntu/maverick.py to oneiric.py and change maverick in the code to oneiric?
[15:15] <van7hu> it's quite strange, why does ubuntu server use samba-4 as default?
[15:16] <soren> hallyn: You know the debdiff from https://bugs.launchpad.net/ubuntu/+source/qemu-kvm/+bug/828789 will still require the new build-depends to be in main, right?
[15:18] <geekbri> I've got to roll my own php .deb file, I'm having a hell of time with it.   I've managed to get debuild to use the proper configure command, and it seems to run a make and make install, but when i inspect the resulting .deb file it only contains some documentation and doesn't install the binaries.  I'm probably screwing this up severely.  Anybody willing to help me figure out what i borked?
[15:19] <alex88> patdk-wk: http://www.faqs.org/docs/iptables/traversingoftables.html as you can see it doesn't get into the input table as soon as destination changes on prerouting
[15:21] <lynxman> zul: hmm gimme 2 mins
[15:35] <robo> anyone have an idea why iostat -n would show 900 ops/s but only 82 rops/s and 20 wops/s? 20 + 82 does not equal 900
[15:40] <smoser> RoAkSoAx, is there a reason you did: $SNIPPET('orchestra_disable_pxe')
[15:40] <smoser> rather than 'cobbler_disable_pxe'
[15:40] <smoser> surely users of cobbler also would prefer that their installed ysstems do not remain in a reboot loop
[15:40] <robo> smoser, ....
[15:55] <zul> lynxman: er puppet?
[15:58] <Daviey> hallyn: spice for P.. hell yes!
[15:58] <Daviey> lets do it today :)
[15:58] <Daviey> jamespage: in truth, erlang-docs don't interest ME in the slightest.
[15:58] <Daviey> I'm tempted to say just strip them if it's easier, and re-introduce them if they turn out to be required
[15:59] <Daviey> but TBH, in the interests of keeping the delta small.. i'd favour dropping them.
[16:00] <zul> SpamapS: ping when you are around
[16:01] <SpamapS> zul: pong, but I'm preparing for my UOW session which starts in 1hr, so not going to be super responsive
[16:01] <zul> SpamapS: sure just wanted to know whats up with mysql 5.5...im going to start merging php 5.3.8
[16:02] <lynxman> zul: just one minute! :)
[16:02] <SpamapS> zul: Norbert is still lagging.. my DD app is progressing tho, so I should be able to upload it to experimental in the next couple of weeks.
[16:02] <SpamapS> zul: please *do not* do php 5.3.8 though, there is a serious regression
[16:02] <hallyn> Daviey: well should we push qemu-kvm-spice on 0.14.1 version just to get it pushed?  (Then I can sync 0.15 tomorrow)
[16:03] <zul> SpamapS: i thought 5.3.7 had the regression and 5.3.8 fixed it
[16:03] <SpamapS> nope
[16:03] <SpamapS> 5.3.7 introduced it
[16:03] <zul> SpamapS: ok then...
[16:03] <SpamapS> 5.3.9 will revert it
[16:03] <zul> SpamapS: k
[16:03] <SpamapS> The php team strikes again!
[16:03]  * zul grumbles
[16:04] <zul> lynxman: how about now?...now...now...now :)
[16:04] <lynxman> zul: your call is important to us...
[16:04] <zul> lynxman: lie
[16:05] <Daviey> hallyn: works for me.
[16:05] <Daviey> hallyn: point me to a branch? :)
[16:05] <Daviey> (or diff)
[16:05] <SpamapS> I wouldn't be surprised if the debian PHP has a patch to revert the is_a() behavior tho
[16:05] <hallyn> lemme make one :)  back in a few
[16:06] <lynxman> zul: bug is #854899
[16:06] <lynxman> zul: branch is lp:~lynxman/ubuntu/oneiric/puppet/lp_854899
[16:06] <lynxman> zul: also attached to the bug
[16:06] <zul> lynxman: cool...ill poke it after lunch
[16:07] <lynxman> zul: so how about now? now now now? :D
[16:07] <zul> lynxman: *grumble* :)
[16:08] <lynxman> zul: rofl
[16:20] <Daviey> How goes?
[16:23] <zul> Daviey: it went
[16:24] <Daviey> zul: it went?
[16:24] <zul> Daviey: you asked how goes....i replied it went
[16:24] <Daviey> ah!
[16:30] <lynxman> Daviey: so now the question that pops to my mind is, should we switch the mcollective package from rabbitmq to activemq to reduce the delta? This is highly debatable
[16:32] <Daviey> lynxman: I think it makes best sense to align with upstream as close as possible, if it's ossible to introduce an alternative dep then we could (and submit it to debian)
[16:32] <lynxman> Daviey: hmm yeah, maybe a new metapackage for whoever wants to keep rabbitmq?
[16:33] <Daviey> lynxman: hmm.. maybe.  Really, just try and align as close to possible as Debian, and submit changes there.
[16:33] <Daviey> :)
[16:34] <lynxman> Daviey: yessir *salutes*
[16:35] <Daviey> hah
[16:36] <zul> lynxman: so i have issue with the puppet merge..a whole bunch of security patches got dropped from the debian package,  i know why, but it isnt mentioned in the changelog when you did the merge...also it should be set to precise now as well
[16:37] <Daviey> (and it looks like it's now a merge from testing, not unstable)
[16:37] <lynxman> zul: it should be in the changelog, hmmm...
[16:38] <lynxman> zul: I think it would just be easier to drop that FFe and do a straight merge from the latest testing on 2.7.5
[16:38] <zul> lynxman: yeah also the security team might freak as well
[16:38] <lynxman> zul: those patches come straight from the security team :)
[16:38] <zul> lynxman: yeah want me to handle that?
[16:38] <lynxman> zul: if you want, I'm swamped with the whitepaper
[16:38] <lynxman> zul: otherwise I can handle it in 2 hours or such
[16:38] <zul> lynxman: ok...
[16:39] <zul> lynxman: ill take it...wanna me close that bug
[16:39] <lynxman> zul: depends on how much do you want to exploit your minions today
[16:39] <zul> lynxman: meh...its sitting right in front of me...ill invalidate that bug then
[16:40] <lynxman> zul: okay I'll submit a new one tonight then
[16:40] <zul> lynxman: k unless if i get to it first
[16:40] <lynxman> zul: race! \o/
[16:41] <Daviey> zul: When you merge python-xattr, can you submittodebian dh_python2 ?
[16:41] <Daviey> did -stompy transition get pushed back?
[16:42] <zul> Daviey: not yet
[16:42] <Daviey> k
[16:44] <zul> just done
[16:45] <Daviey> rockign!
[16:48] <Daviey> Ursinha: around?
[16:51] <Daviey> jamespage: asm2, is it still worth our while having a delta?
[16:51] <jamespage> Daviey: I was looking at that earlier
[16:52] <Daviey> it *looks* like just running the test suite at build time?
[16:52] <jamespage> I think that is just about it
[16:52] <Daviey> jamespage: I'd sync it, then re-introduce a delta if there is an issue :)
[16:52] <Daviey> <-- yeehaaaa!
[16:57] <Daviey> jamespage: how did fop turn out?
[16:58] <jamespage> Daviey: sitting on that one for the moment
[16:58] <jamespage> it pulls back into main a load of packages we demoted last release
[16:58] <Daviey> ah
[16:58] <jamespage> inc. one which has updated to use maven
[16:58] <jamespage> 70+ MIR
[16:58] <jamespage> do we really want erlang docs?
[16:58] <Daviey> jamespage: Fancy opening a bug to track it?
[16:58] <jamespage> already have
[16:58] <Daviey> ah
[16:59] <jamespage> bug 876413
[16:59] <Daviey> crikey o'reily.
[16:59] <jamespage> thats not what I said :-)
[17:00] <lynxman> jamespage: I can see you saying that James
[17:00] <Daviey> jamespage: "gosh" ?
[17:01] <jamespage> I think I said: "by heck thats alot of packages" - well something along those lines
[17:01] <lynxman> jamespage: that sounds realistic too
[17:01] <jamespage> lol
[17:02] <Daviey> heh
[17:02] <jamespage> actually looking at archive mismatching something already pulled in most of the know Java universe!
[17:05] <Daviey> jamespage: is it antlr?
[17:06] <jamespage> Daviey: I don't think so - its hard to read its so long
[17:08] <Daviey> jamespage: something introduced ~10:00am today
[17:08] <jamespage> yeah
[17:08] <Daviey> (UTC) +/- 1hr
[17:10] <Daviey> As we don't have a reliable -changes list, cannot confirm
[17:10] <Daviey> (sigh*
[17:10] <tzabar> letar efter php, mysql, js... programmerare... ngn intresserad?
[17:10] <ersi> !english
[17:10] <ersi> tzabar: ^
[17:12] <jamespage> Daviey: it might be ant
[17:12] <hallyn> Daviey: I ought to do some testing, but http://people.canonical.com/~serge/qemu-kvm-spice-oct17.2011.debdiff
[17:12] <jamespage> hmm - or it might not be
[17:12] <jamespage> anyway thats one for tomorrow
[17:13] <hallyn> Daviey: can you push the spice-gtk package (from the FFE bug) in the meantime?
[17:13] <Daviey> hallyn: sure
[17:13] <hallyn> Daviey: I assume I can't dput it the .changes file for qemu-kvm as there is the new packagre?
[17:13] <hallyn> package?
[17:14] <hallyn> maybe i should re-try that sentence :)
[17:15] <tzabar> ersi?
[17:15] <hallyn> Daviey: taking a walk, biab
[17:16] <Daviey> hallyn: heh, if what you are dputting has a higher version number (and obv. includes Ubuntu changes between when you did it and now), sounds good
[17:16] <Daviey> hallyn: have fun
[17:18] <adam_g> .wun 21
[17:18] <adam_g> doh
[17:18] <dkn> should i run lvm on a raid if i want hot swap support?
[17:19] <dkn> i can't get an encrypted lvm to open until i reboot if pull the drive and put it back in
[17:22] <Ursinha> Daviey: hey
[17:24] <patdk-wk> dkn, heh?
[17:24] <RoyK> dkn: iirc hotswap will work correctly regardless of lvm being there or not
[17:24] <patdk-wk> lvm defently doesn't support hotswap at all
[17:25] <patdk-wk> if it's missing, it's failed, and the lvm fails
[17:25] <RoyK> patdk-wk: won't md hide that if on a redundant setup?
[17:25] <patdk-wk> not sure what you mean by encrypted lvm, didn't think lvm supports encryption
[17:26] <patdk-wk> md would yes, but that is another layer
[17:26] <Daviey> hey Ursinha.. how are you doing?
[17:26] <RoyK> patdk-wk: the only really usable for raid on linux, though
[17:26] <dkn> that's the odd thing, hot swap works... the lvm shows up, it's seeing the drive as the correct device, when i go to unencrypt it keeps prompting me for the pass over and over, works fine again after i  reboot or if i don't put the drive out hot
[17:27] <RoyK> dkn: is this over md or directly on disks_
[17:27] <RoyK> ?
[17:27] <patdk-wk> dkn, what your saying doesn't make sense
[17:27] <patdk-wk> lvm doesn't support encryption, so what is asking for a password?
[17:27] <martyn> Are you sure it's not a sudo command asking you for a password?
[17:27] <patdk-wk> if it's encrypted, you would have to be using dmadm (cryptsetup)
[17:27] <dkn> right now i have the pv, gv, and lvm all on one disk so i could encrypt the lvm since it's a hot swap offsite backup
[17:28] <martyn> (I know, sounds silly, but it's bitten  me once or twice)
[17:28] <patdk-wk> in that case, the encryption is working, but the lvm is not
[17:28] <patdk-wk> cause you have to export the lvm, then reimport it
[17:28] <dkn> i use the disk utitility to unlock it actually :$
[17:28] <patdk-wk> just cause the /dev/mapper exists, doesn't mean lvm is using the new one, it's still using the old one that is gone
[17:28] <dkn> how do i force it to update?
[17:29] <patdk-wk> dkn, tell lvm to non-active, then reactivate it
[17:29] <RoyK> dkn: I beleive what patdk-wk is trying to say, is that LVM does not support hotswap
[17:29] <patdk-wk> that means unmounted all drives
[17:29] <dkn> hmm
[17:29] <patdk-wk> ya, it sounds like he has a, harddrive -> dmadm(encryption) -> lvm, setup
[17:30] <dkn> i don't mind doing that
[17:30] <Daviey> Ursinha: Did you see the changes i pushed to the reports?
[17:30] <dkn> i unmount, lock, just was missing that lvm non-active step
[17:30] <patdk-wk> dkn, well, lvchange and vgchange will be your friends :)
[17:30] <dkn> sudo lvchange backup -a y ???
[17:31] <patdk-wk> that activates it :)
[17:31] <patdk-wk> yo uhave to deactivate it, and probably the vg also
[17:31] <patdk-wk> I haven't personally done that for a few years, I normally stay away from lvm
[17:31] <dkn> ugggh...
[17:32] <dkn> doesn't the deactivate option??
[17:32] <dkn> -a n?
[17:32] <patdk-wk> yep
[17:32] <martyn> dkn : You -always- have to deactivate
[17:32] <martyn> lvchange -a n <path to lvm> vgexport -a (don't worry it will only export groups that are ready to be exported)
[17:33] <martyn> (two commands there)
[17:33] <martyn> then you can reactivate
[17:33] <martyn> the -drive- is hot swap, not LVM
[17:33] <dkn> ya...
[17:33] <Ursinha> Daviey: no, I haven't
[17:33] <martyn> This is why I use hot-spare solutions, and RAID
[17:33] <martyn> and just leave the failed drive in the array until I'm ready to do maintenance
[17:33] <dkn> what do you think roy?? just run them in raid1? instead of trying to swap out?
[17:33] <Daviey> Ursinha: http://bazaar.launchpad.net/~ubuntu-reports-dev/ubuntu-reports/trunk/revision/43
[17:34] <Daviey> Ursinha: So, rather than using a tag for -rs and -ro, using a launchpad id.. Does that work well with you?
[17:34] <Ursinha> Daviey: you should request a code review before pushing changes, you know? :)
[17:34] <Daviey> Ursinha: I followed the current workflow... :)
[17:35] <Daviey> Ursinha: i did check if people were using merge proposals first :P
[17:35] <Ursinha> Daviey: there's no current workflow defined, and code reviews are a good practice, I thought that was common sense :P
[17:35] <Daviey> Ursinha: seems to be working, http://status.qa.ubuntu.com/reports/ubuntu-server/release-bugs.html
[17:36] <Daviey> Ursinha: Aye, which was why i checked first.. :)
[17:36] <Ursinha> omg, it seems to be working? you're lucky I like you, you know that? :)
[17:36] <Daviey> Ursinha: So i know this is odd, but i check stuff before uploading :)
[17:36] <dkn> lvchange: option requires an argument -- 'a'
[17:36] <Ursinha> Daviey: lol
[17:37] <Ursinha> Daviey: I'll take a look when I have reasonable internet connection
[17:37] <Daviey> Ursinha: sure.. on another note, the wiki output is broken.. but that was broken before i went near it :P
[17:37]  * RoyK just upgraded to 60Mbps symmetric, which is somewhat reasonable
[17:38] <dkn> http://paste.ubuntu.com/711114/
[17:38] <Ursinha> haha
[17:38] <Ursinha> Daviey: mind filing a bug?
[17:38] <Daviey> Ursinha: all the unit tests passed, so it can't be an issue.
[17:38] <Daviey> it does have unit-tests, right, right?
[17:39] <Ursinha> HAHAHA
[17:39] <Ursinha> you got me
[17:39] <Daviey> :D
[17:39] <Ursinha> I won't be a bitch and say you could have written them as well
[17:39] <Ursinha> oh wait, I just did that
[17:39] <Ursinha> :P
[17:40] <Daviey> Ursinha: There is a bug... it needs to check if current development series has a milestone..
[17:40] <hallyn> Daviey, not sure i understand what you said.  The new qemu-kmv has higher version #, 0.14.1+noroms-0ubuntu7, but it has a new package, qemu-kvm-spice.  i should be able to dput that?
[17:40] <Daviey> hallyn: oh sure.. the new binary package will wait in the archive binNEW queue for ~ubuntu-archive approval.
[17:41] <Daviey> but that is standard.
[17:41] <hallyn> Daviey, cool, then i'll push after some more testing
[17:42] <Daviey> hallyn: did you see, bug 875001?
[17:43] <hallyn> Daviey, yes i did.  it was obviously my plan all along, but zul said he would try it
[17:43] <hallyn> zul, ^ wanna take that bug?
[17:43] <zul> hallyn: working on it
[17:44] <Daviey> hallyn: Yeah, someone else opened a bug already.. I knew it was on your radar, but thought you'd want to quote the bug number in the changelog
[17:45] <hallyn> given the amount of stuff we're changing in p, i'll need to set up a precise test system post haste
[17:45] <Daviey> sounds good
[17:45] <Daviey> adam_g: around?
[17:45] <Daviey> anyone seen Ben today?
[17:47] <hallyn> Daviey, so did you push spice-gtk?
[17:47] <adam_g> Daviey: i am
[17:48] <hallyn> Daviey, sorry, actually nm - spicec suffices for testing
[17:48] <Daviey> hallyn: you want http://people.canonical.com/~serge/qemu-kvm-spice-oct17.2011.debdiff reviewed and uploaded?
[17:49] <Daviey> adam_g: Having fun sir?
[17:49] <hallyn> Daviey, i'm testing it right now.  i can push that myself, but wouldn't mind a good review!
[17:49] <adam_g> Daviey: trying to get thru back log of email from last week and find out whats going on. its been a while. :P
[17:50] <Daviey> hallyn: Pre-Depends is sort of a last resort.. is that really required?
[17:51] <Daviey> adam_g: hah
[17:51] <hallyn> Daviey, I don't recall, but it's in qemu-kvm's right now.
[17:53] <hallyn> Daviey, yes, it's needed for qemu-kvm.  but not for qemu-kvm-spice.  i'll drop that
[17:53] <Daviey> ok, cool
[17:53] <hallyn> (debian suggests pre-depends for a package needed during .preinst;  qemu-kvm.preinst uses addgroup)
[17:54] <hallyn> Daviey, suppose id on't need that provides: line either?
[18:05] <ebachle> hi all, I threw this question out at one point not too long ago, but it seemed like few people were around so i'm going to try again.
[18:05] <ebachle> I'm looking for the best way to configure a server so that once I graduate and my college no longer has me to maintain it that it is really set up in accordance with some set of best practices.  Any advice on where to start?
[18:08] <dkn> hire the future IT firm now
[18:08] <dkn> let them set it up how they want
[18:08] <ebachle> well considering they difficulty getting the administration to ok my measly $9/hr wage part-time
[18:09] <ebachle> contracting out doesn't sound like it's one of our options
[18:09] <patdk-wk> who is going maintain the security updates? replace failed parts? ...
[18:10] <patdk-wk> no matter how secure it is, if it's connected to anything external, it needs security updates
[18:10] <ebachle> My boss is pretty good overall, she'd be able to keep it going if it didn't break entirely
[18:10] <dkn> even if it isn't something will break eventually
[18:10] <ebachle> like she's totally capable of updating it and things, as well as troubleshooting any issues that come up in terms of individual programs or scripts or whatnot
[18:11] <ebachle> she just doesn't have the background to set it up the right way the first time to make sure that happens the least, and I've got some idea as to how i'd do it like on something i would use individually
[18:11] <ebachle> but the requirements on me at this job mean i have to be a bit more precise
[18:13] <dkn> what kind of server?
[18:15] <ebachle> There's two at least somewhat pressing for me
[18:15] <ebachle> one is BigBlueButton
[18:15] <ebachle> and the other is Kaltura
[18:16] <ebachle> so not any of the canned solutions unfortunately
[18:20] <RoyK> lol - from the cpufreq-set(1) man page: FREQuencies can be passed in Hz, kHz (default), MHz, GHz, or THz
[18:21] <genii-around> Heh, THz
[18:22] <RoyK> perhaps when someone creates an optical CPU one day.....
[18:23] <patdk-wk> I thought they did
[18:23] <patdk-wk> two of them
[18:23] <RoyK> I thought they'd only made simple things, not CPUs
[18:24] <dkn> can't you get a 512 core atom server?
[18:25] <dkn> not exactly the first thing that comes to mind with THz
[18:26] <RoyK> dkn: I'd missed that, but then, I'd love this one - http://carpetbomberz.com/2011/06/23/calxeda-software-partners/ - 480 core ARM thing in 2U
[18:27] <RoyK> now _that's_ dense
[18:27] <genii-around> There was some company in 2005-2006 purporting to have "quantum optical" chips but in the end it seemed like some scam. At the time they were claiming a 6 or 7 GHz laptop. Atom Chip Corporation
[18:28] <patdk-wk> wonder if vmware runs on that arm server :)
[18:28] <RoyK> patdk-wk: :)
[18:28] <RoyK> patdk-wk: xen might work
[18:29] <patdk-wk> wonder if it's still stuck to the 2 or 4gig memory limit :)
[18:29] <patdk-wk> oh wait, that is arm, not atom :)
[18:29] <reya276> Does anyone know what is the CIL Permissions (7**) so that a folder is read only by root?
[18:29] <RoyK> any idea how to get rid of these messages? as in how to ignore them and stop the kernel from bugging me? [100390.030872] TCP: Peer 85.228.171.248:28519/57137 unexpectedly shrunk window 3010574087:3010584391 (repaired)
[18:29] <patdk-wk> royk, don't talk to outside servers?
[18:30] <RoyK> patdk-wk: I'd rather ignore those messages
[18:30] <RoyK> or stop them from being logged in the first place
[18:31] <ersi> RoyK: Make a syslog filter which matches that and logs to /dev/null
[18:31] <RoyK> ersi: it still spams my dmesg, though...
[18:33] <ersi> RoyK: oh yeah, I forgot about that. heh'
[18:36]  * RoyK considers writing a fail2ban rule to end it
[18:37] <RoyK> anyway - packing 120 quad core CPUs into 2U is rather extreme....
[18:38] <patdk-wk> I hope you can fit enough fc/ib cards into it to make it useful :)
[18:38] <patdk-wk> unless it's a pure compute node
[18:38] <RoyK> most compute nodes needs disk space somewhat
[18:39] <patdk-wk> my mythtv frontends are very happy with iscsi/nfs over 100mbit :)
[18:39] <RoyK> but then, at least for what we're doing, the compute nodes are 99,lots% CPU and NFS over gigE is sufficient for storage
[18:41] <RoyK> I wonder how they have setup those ARM CPUs - what sort of communication between them? local memory for all?
[18:47] <patdk-wk> so much for ff 7 using less memory
[18:47] <patdk-wk> mine is up to 2gigs and growing
[18:53] <hallyn> Daviey, qemu-kvm-spice uploaded.
[18:54] <hallyn> are you able to approve the new binary package then?
[18:56] <Daviey> hallyn: I am not, no.
[18:57] <ersi> patdk-wk: Do you have like a billion tabs?
[18:58] <patdk-wk> only 22
[18:58] <patdk-wk> damn, restart and it's using 600megs, firefox6 used to only use 200megs after a restart
[19:00] <hallyn> Daviey, ok
[19:06] <hallyn> Daviey, i'm going to be filing ITP for netcf as per bug 520386.  just heads-up
[19:12] <Daviey> hallyn: cool, do you have a strategy to get it in Debian? :)
[19:14] <hallyn> Daviey: strategy?  Not so much :)  I don't want to wait on the package to go into debian first;  but I do want to propose my package to debian
[19:15] <hallyn> Daviey, I'll repply to debian bug 573072
[19:15] <Daviey> hallyn: sounds good!
[19:17] <hallyn> email sent
[19:19] <Daviey> \o/
[19:20] <soren> smoser: sorry, missed your question. The drawings in question are the ones in http://alturl.com/wqqpn
[19:23] <jhunt> quit
[19:37] <blkperl> hi server team if you havn't already seen it #876130 blocks the use of puppet in oneric, can someone assign an importance level so this gets looked at, thanks
[19:51] <mdeslaur> ohm...does oneiric not support ip aliases?
[19:52] <guntbert> mdeslaur: why do you ask?
[19:52] <mdeslaur> guntbert: I tried to add eth0:0, and it's not working on oneiric, works fine on natty and hardy
[19:53] <guntbert> mdeslaur: (I'm trying to remember some changes in that regard, about a ear ago): how did you try to create it?
[19:54] <guntbert> *year
[19:54] <mdeslaur> guntbert: /etc/network/interfaces ...
[19:54] <mdeslaur> guntbert: let me paste the config
[19:54] <guntbert> !plastebin | mdeslaur
[19:54] <mdeslaur> guntbert: yes, I know
[19:54] <guntbert> :-/
[19:55] <mdeslaur> http://paste.ubuntu.com/711266/
[19:55] <mdeslaur> yep, someone else just tried it and confirmed
[19:55] <mdeslaur> darn
[19:56] <RoyK> mdeslaur: eth0:x is rather the old way - adding 'up ip addr add 10.100.0.200/24 dev eth0' is a bit easier these days ;)
[19:57] <mdeslaur> RoyK: adding that to what?
[19:57] <guntbert> RoyK: *that* was it, thx
[19:57] <RoyK> mdeslaur: /etc/network/interfaces
[19:57] <RoyK> instead of eth0:0 ......
[19:57] <RoyK> just that one line does all you need
[19:58] <RoyK> and perhaps add 'down ip addr remove ...'
[19:58] <RoyK> erm, make that ip addr del
[19:58] <RoyK> man ip
[19:58] <mdeslaur> RoyK: thanks, I'll give that a try
[19:59] <RoyK> mdeslaur: just 'ifconfig eth0:0 down' first
[19:59] <RoyK> or reboot :P
[19:59] <Patrickdk> up ip addr add 10.100.0.200/24 dev eth0 label eth0:0, :)
[19:59] <RoyK> the ip command can be used directly from the commandline
[19:59] <RoyK> Patrickdk: heh
[19:59] <Patrickdk> or if you want more fun
[19:59] <Patrickdk> up ip addr add 10.100.0.200/24 dev eth0 label eth1
[20:00] <RoyK> heh - didn't know the label
[20:00] <Patrickdk> the label makes it, ifconfig, compat :)
[20:00] <mdeslaur> RoyK: where is this documented?
[20:00] <RoyK> mdeslaur: no idea where I read it - google it :)
[20:01] <Patrickdk> man ip :)
[20:01] <Patrickdk> it's part of the iproute2 package
[20:01] <RoyK> mdeslaur: 'up/down' just says 'whenever this in goes up/down, run this command'
[20:01] <mdeslaur> RoyK: because the current debian documentation still says to do what I did
[20:01] <Patrickdk> ya, I use a mix of both methods
[20:01] <RoyK> so 'up echo WTF! I'm alive!' will probably work
[20:01] <Patrickdk> royk, I can't get that to work :(
[20:02] <RoyK> mdeslaur: has debian moved to linux 2.6 yet? ;)
[20:02] <mdeslaur> ok, I still think this is broken
[20:02] <mdeslaur> RoyK: you're just working around it
[20:02] <RoyK> (i know, i know, but debian is a BIT on the conservative side)
[20:05] <RoyK> mdeslaur: sorry, just trying to help here :)
[20:06] <mdeslaur> the examples in /usr/share/doc/ifupdown/examples also say to use iface eth0:1
[20:07] <SpamapS> I do actually think is time we did a bit of an overhaul on ifupdown best practices and docs
[20:07] <SpamapS> the manual method that RoyK is describing is far superior IMO
[20:07] <SpamapS> and ifupdown is a big 'ol mess that is hard to keep up with the times
[20:08] <mdeslaur> SpamapS: that doesn't help with upgrades
[20:08] <RoyK> SpamapS: and the iproute2 method is what, 10 years old?
[20:09] <sbeattie> SpamapS: the whole point of doing it via interfaces is to abstract away so that when iproute3 comes along, I don't have to care.
[20:09] <SpamapS> mdeslaur: great point. We should be able to effectively convert them.
[20:09] <SpamapS> sbeattie: god help us
[20:10] <Patrickdk> royk, the iproute2 method is still not complete
[20:10] <SpamapS> Seriously, choice between having explicit iproute2 commands and ifupdown managing my interfaces through a long server lifecycle.. I'm choosing iproute2 commands even if I have to fix them every time
[20:10] <RoyK> Patrickdk: what's missing there?
[20:10] <Patrickdk> you can't work with arp in iproute2 yet, it's support is incomplete
[20:10] <Patrickdk> still have to use the unmaintained arp package that is 7years old
[20:10] <SpamapS> Just go digging around in the source for a while
[20:10] <SpamapS> *scary*
[20:11] <RoyK> Patrickdk: I didn't know that - thanks for the update
[20:11] <Patrickdk> yep
[20:12] <Patrickdk> attempted to update some firewall scripts to use iproute2
[20:12] <Patrickdk> and hit that issue
[20:12] <SpamapS> Still choosing that over ifupdown's nowebm source mess. :)
[20:12] <sbeattie> SpamapS: why bother with ifupdown at all then and not just use your own rolled rc.local?
[20:13] <Patrickdk> sbeattie, going slackware method? :)
[20:13] <SpamapS> sbeattie: encapsulation is good
[20:13] <sbeattie> SpamapS: but not good enough for aliased interfaces, apparently.
[20:13] <SpamapS> and I acknowledge all the benefits of having a declarative place to contain interface configs
[20:14] <SpamapS> but ifupdown is a horrible implementation
[20:14]  * SpamapS goes OTP
[20:15] <RoyK> hm... does md have something analogous to zfs scrub? as in checking every block if it's readable/correct. I somewhat doubt it, but it'd be nice
[20:15] <Patrickdk> sure
[20:15] <Patrickdk> it's called scrub :)
[20:16] <RoyK> http://wiki.xdroop.com/space/Linux/Software+Raid+Scrub ?
[20:17] <Patrickdk> yep, just add to cron :)
[20:17] <Daviey> zul: if you are tackling the libvirt merge, can you assign the bug to yourself please? :)
[20:18] <zul> Daviey: yeah im just about to eod so ill do it tonight...build fails right now because of the testsuite
[20:19] <RoyK> Patrickdk: IMHO a thing like that should be added to mdadm
[20:19] <Daviey> zul: testsuites are overated :)
[20:19] <zul> indeed...stupid nc
[20:20] <Patrickdk> royk, maybe, but there are already many things mdadm don't do, that I do
[20:20] <Daviey> heh
[20:20] <RoyK> Patrickdk: and that's probably the most stupid answer I've heard in a while :P
[20:20] <Patrickdk> :)
[20:21] <RoyK> even if I or you can do things manually, that doesn't mean it shouldn't be possible in a more user- or adminfriendly way
[20:21]  * Patrickdk never thought mdadm was *friendly*
[20:21] <RoyK> well, it's adminfriendly and rather well documented
[20:22] <RoyK> hacking around in proc or sys is like mdb/zdb hacking on solaris, it's deep nerdy and cool, but very far from admin friendly
[20:22] <Patrickdk> no, proc and sys is more like /etc/system
[20:22] <Patrickdk> much safer
[20:23] <Patrickdk> mdb is evil :)
[20:23] <RoyK> well, yes
[20:23] <RoyK> evil can be used for good :P
[20:23] <RoyK> just tread _really_ carefully :P
[20:24]  * Patrickdk attempts to always reboot, after finalizing the mdb adjustment
[20:24] <Patrickdk> incase of random acts of corruption caused by the admin :)
[20:25] <RoyK> first hit on mdb on google is an article named 'Solaris Things: Don't be afraid of mdb'
[20:25] <Patrickdk> unless you use the -w switch
[21:00] <kirkland> SpamapS: SRU team favor, please?
[21:00] <kirkland> SpamapS: could you get https://bugs.launchpad.net/ubuntu/+source/byobu/+bug/796422 promoted to -updates asap?
[21:00] <SpamapS> kirkland: I actually just started reviewing SRU's 5 minutes ago. :) good timing...
[21:08] <SpamapS> kirkland: I'm going to go ahead and verify the update on maverick so that we have an unbroken chain lucid->natty->oneiric
[21:08] <kirkland> SpamapS: \o/
[21:08] <kirkland> SpamapS: cool, thanks.
[21:08] <SpamapS> kirkland: released to natty, lucid/maverick will release shortly
[21:08] <kirkland> SpamapS: ec2 instance should work just fine
[21:08] <SpamapS> kirkland: thats what I'm using. :-D
[21:08] <kirkland> SpamapS: :-)
[21:09] <SpamapS> heh.. 10.10 , with its double motd.. :-P
[21:09] <SpamapS> oh wait, thats not double.. its just *GIANT*
[21:19] <jMCg> Hey folks - trying to boot a VM, I'm hitting this issue here: http://ubuntuforums.org/showthread.php?t=1423305 -- is that.. seriously serious? I can't have a separate /var ?
[21:20] <lenios_> jMCg, you don't have this issue
[21:20] <lenios_> ureadahead message has nothing to do with it
[21:21] <lenios_> can you paste your fstab?
[21:21] <jMCg> I can.
[21:22] <jMCg> I first have to kill the VM.
[21:24] <jMCg> lenios_: http://sprunge.us/EVXX
[21:27] <lenios_> can you try without the last line?
[21:29] <SpamapS> kirkland: maverick and lucid done as well
[21:30] <kirkland> SpamapS: woot!
[21:30] <kirkland> SpamapS: thanks, you da man!
[21:31] <SpamapS> np
[21:35] <jMCg> lenios_: same thing
[21:36] <jMCg> 23:16:58 < jMCg> And this is how far virt console gets: http://dpaste.com/636142/
[21:36] <jMCg> 23:18:34 < jMCg> And this here is the xml of the VM: http://sprunge.us/BOAU
[21:39] <lenios_> i would suggest trying to mount with defaults options
[21:41] <lenios_> (both /var and /srv)
[21:46] <jMCg> lenios_: I have, on another machine http://sprunge.us/Ychb -- but that's run by 11.04 - haven't done the upgrade yet. Not sure I want to, anytime soon.
[21:49] <lenios_> i once had a similar non-verbose issue using a bad fstab
[21:52] <lenios_> jMCg, i would try to mount /var and /srv as ext4 defaults 0 2
[21:54] <jMCg> lenios_: any idea if doing this by UUID will help any?
[21:54] <lenios_> i don't think it would help
[21:55] <jMCg> lenios_: I'll do the same for /var - see how it does.
[21:57] <jMCg> Same thing.
[21:59] <daishadar> i can't seem to get cp to preserve nfs4 acls... does anyone mounting nfs4 mounts see the same thing?
[21:59] <lenios_> jMCg, can you paste the result of fdisk -l or sfdisk -l ?
[22:00] <jMCg> lenios_: from the host or from the not booting VM ;)
[22:00] <patdk-lap> daishadar, why would cp perserve them?
[22:00] <lenios_> from the not booting VM
[22:00] <jMCg> lenios_: that sounds difficult.
[22:01] <lenios_> don't you have any livecd?
[22:01] <daishadar> patdk-lap: if you have acl inheritance on i would think it should... this is the behavior in other oses like solaris
[22:01] <jMCg> lenios_: Well, that would be the Host system, sort of.
[22:01] <patdk-lap> man cp, says to use -/ for extended attributes
[22:01]  * patdk-lap notes linux != solaris
[22:01] <lenios_> what are you using to run the VM?
[22:02] <jMCg> lenios_: libvirt/kvm
[22:02] <jMCg> Gimme a sec to download a grml ISO.
[22:03] <jMCg> 2.6 seconds, to be exact.
[22:03] <lenios_> you really have a good connection
[22:04] <jMCg> It's the small one :)
[22:04] <patdk-lap> he just has a *close* connection to a grml mirror :)
[22:04] <jMCg> 110 MiB
[22:04] <lenios_> that's more than 40MiB/s
[22:05] <jMCg> Well I sure hope so. The severs are connected with 100Mbps - not sure how much the grml server has I connected to.
[22:05] <patdk-lap> heh, I'm connected at gigabit, but lucky to get 10mbit most of the time
[22:06] <patdk-lap> but when I download from vmware servers, I get a nice 200+ mbit
[22:20] <jMCg> lenios_: init=/bin/sh helped and was easier than trying to attach, and boot from a cdrom.
[22:20] <lenios_> true
[22:21] <jMCg> http://dpaste.com/636178/
[22:22] <jMCg> http://dpaste.com/636179/
[22:22] <lenios_> :/
[22:23] <jMCg> Iḿ not happy either.
[22:23] <jMCg> It mounts on the host.
[22:24] <lenios_> what about sfdisk -l ?
[22:24] <jMCg> lenios_: that's the second paste.
[22:29] <jMCg> Okay, so I thought, give it a shot, and this is where we part from sense and reasoning: http://dpaste.com/636182/
[22:30] <jMCg> # swapon -a
[22:30] <jMCg> [  632.178201] Adding 8388604k swap on /dev/vdb.  Priority:-1 extents:1 across:8388604k
[22:39] <jMCg> I just looove it when it doesn't make any sense at all
[23:08] <jMCg> If I get this right, then I cannot boot 11.04 with a 11.10 kernel.
[23:09] <jMCg> verifying that theory right now.
[23:24] <jMCg> Yup. Gets me one step closer: http://dpaste.com/636205/
[23:35] <jMCg> Nice, reproducibl every time. Even though the bug is supposedly fixed. https://bugs.launchpad.net/ubuntu/+source/linux/+bug/613273
[23:58] <jMCg> Man, sometimes I hate computers.
[23:58] <jMCg> I should've listen to what my momma told me.