[02:38] <jeeves_moss> how do I setup a rule to move all e-mail marked with "**SPAM**" to the user's junk box?
[02:39] <twb> jeeves_moss: in what?  procmail?
[02:40] <jeeves_moss> dovecot
[02:40] <jeeves_moss> brb
[02:40] <twb> jeeves_moss: so, you want to do this for existing mail, or every time new mail comes in?
[02:40] <greppy> jeeves_moss: look into sieve for dovecot.
[02:41] <twb> greppy: hmm, could sieves replace my users' use of procmail?
[02:41] <greppy> if using dovecot, yes.
[02:41] <twb> Cool
[02:42] <greppy> there's even squirrelmail and roundcube plugins to talk to sieve, so they can manage thier filters that way.
[02:43] <twb> Actually I hope to roll out prayer instead, simply because it's not PHP
[02:55] <jeeves_moss> twb, yes, existing (since once account has ~8K e-mails in it) and all existing
[02:55] <twb> jeeves_moss: OK, then you can just do that easily with mutt or python imaplib
[02:56] <jeeves_moss> ??
[02:59] <twb> In mutt for example, mutt -f imaps://user@imap.example.net
[02:59] <twb> Then T ~s**SPAM** to tag stuff with spam in the subject line
[03:00] <twb> Then ; C =/Spam RET to copy into the spam folder (and mark the local version for expunging)
[03:00] <twb> Then $ or q to expunge the spam from the local folder.
[03:01] <jeeves_moss> kk, thanks
[03:02] <twb> That won't be as efficient as writing a little 10-line python script
[03:21] <patdk-lap> or if it's in maildir format, even easier
[03:22] <patdk-lap> using dovecot 2.x, just use doveadm :)
[03:57] <twb> I don't think dovecot 2 is recommended yet
[03:58] <patdk-lap> I'm sure every single feature in it isn't 100% yet
[03:58] <patdk-lap> but I haven't had any issues with mailbox or mdbox imap/sieve/lmtp usage with it
[04:51] <osmosis> i want the kernel to cache less and to just leave the unused ram free.  I set swappiness to 10 (from the default of 60) and now it is making even more cache.  uhh...why?  isnt  a low swapiness value suppose  to avoid swapping processes out of physical memory?
[04:56] <ScottK> My advice is don't try to be smarter than the kernel.
[05:00] <qman__> there is no benefit to having RAM free instead of caching
[05:00] <qman__> as cache is automatically purged when RAM is needed
[05:02] <qman__> and that has nothing to do with swap, either
[05:17] <MTecknology> this is a bit frustrating....
[05:18] <MTecknology> i have a file named 't' with a really long list and I use whiptail --textbox "t" 40 80 but there's no ability to scroll the file
[05:18] <MTecknology> just the first 40 lines and the rest is cut off
[07:19] <kaushal> hi
[07:29] <kimmy> Any body know how to set up freeradius for authenticating mac addresses? plz help
[07:30] <kaushal> The server recognizes only 3Gigs out of 4 Gigs of RAM
[07:30] <kaushal> http://paste.ubuntu.com/592471/
[07:31] <kaushal> Its a 64 Bit machine
[07:31] <twb> kaushal: did you install ubuntu amd64?
[07:32] <kaushal> twb: yes
[07:32] <twb> kaushal: ddr2 or ddr3?
[07:32] <kaushal> x86_64
[07:33] <kaushal> twb: ddr1
[07:33] <kimmy> Any body know how to set up freeradius for authenticating mac addresses? plz help
[07:33] <twb> kaushal: I really doubt it's ddr1
[07:34] <twb> http://en.wikipedia.org/wiki/DDR_SDRAM
[07:34] <twb> DDR1 was pre-pentium iirc
[07:34] <kimmy> Any body know how to set up freeradius for authenticating mac addresses? plz help
[07:34] <kimmy> how to set mysql database for doing mac auth for freeradius?
[07:34] <twb> kimmy: be patient.  Repeating yourself will not help.
[07:35] <kaushal> ok
[07:38] <kaushal> twb: so is there a limitation ?
[07:38] <kaushal> I have seen it in BIOS
[07:39] <kaushal> so for setting up a gateway server, 3 Gigs would suffice ?
[07:41] <twb> kaushal: IIRC you have to populate DDR3 in triplets, not pairs
[07:41] <twb> kaushal: so if you have four sticks in there, that's why only three work
[07:43] <kaushal> ok
[07:52] <_ruben> the mainboard determines the use of dual/triple channel, not the dimms themselves
[07:53] <_ruben> 3g should be plenty for most gateways, kinda depends on what you'll be running on it obviously
[07:56] <kaushal> _ruben: ok
[07:57] <kaushal> _ruben: still not fully understood
[07:57] <kaushal> apologies for asking
[07:57] <twb> I take "gateway" to mean "it's a router", i.e. 16MB is plenty
[07:58] <kaushal> On the BIOS it shows DDR1
[07:58] <twb> But I expect he meant something else
[07:58] <kaushal> twb: shall i pastebin the dmidecode output ?
[08:00] <twb> kaushal: didn't you already do that?
[08:02] <kaushal> twb: it was just for memory only
[08:02] <kaushal> http://paste.ubuntu.com/592486/
[08:09] <kaushal> _ruben: you around ?
[08:13] <_ruben> twb: 16megs wouldn't suffice for a full ipv4 feed ;)
[08:13] <twb> "feed"?
[08:13] <twb> Last time I looked it was packet-switched
[08:13] <_ruben> +table
[08:14] <twb> Are you talking about that travesty known as NAT?
[08:14] <_ruben> no, about having 300k routes in your routing table
[08:14] <twb> Maybe if he's the backbone
[08:14] <twb> I doubt it
[08:15] <twb> ssh alpha.cyber.com.au ip route | wc -l ==> 15
[08:16] <_ruben> i run edge sites but do have full routing tables at several of 'em .. so "the backbone" is a flexible thing ;)
[08:16] <_ruben> either way, i doubt this is the case for kaushal
[08:18] <kaushal> ok
[08:20] <kaushal> I have 4 slots and chip of DDR1 400 Mhz
[08:20] <kaushal> still trying to understand
[08:22] <_ruben> you really sure its 64bits kernel? did you grant an insane ammount of memory to the (onboard) gfx card? etc..
[08:29] <jussi> how does one upgrade ubuntu server from lucid to natty (or maverick)?
[08:30] <kaushal> _ruben: http://paste.ubuntu.com/592497/
[08:31] <kaushal> http://paste.ubuntu.com/592499/
[08:33] <_ruben> jussi: edit /etc/update-manager/release-upgrades, then use do-release-upgrade to perform the actual upgrades
[08:33] <ScottK> jussi: sudo do-release-upgrade
[08:34] <ScottK> You will need to edit it as _ruben said to go to a non-lts.
[08:34] <_ruben> kaushal: looks sane
[08:34] <kaushal> ok
[08:35] <jussi> _ruben: ScottKthanks
[08:36] <kaushal> _ruben: so ?
[08:36] <jussi> also, is there an appropriate tool that is equivalent to webmin for ubuntu?
[08:36] <_ruben> kaushal: what does 'free -m' show?
[08:36] <ScottK> jussi: vim
[08:36] <_ruben> ebox
[08:36] <jussi> ScottK: hah!
[08:36] <jussi> ebox is rather strange though. :/
[08:37] <kaushal> http://paste.ubuntu.com/592502/
[08:38] <_ruben> kaushal: how about 'dmesg' ?
[08:38] <kaushal> sure
[08:38] <kaushal> please give me a moment
[08:39] <kaushal> http://paste.ubuntu.com/592504/
[08:44] <_ruben> my guess is that your motherboard somehow masks the area between 3G and 4G .. which isn't that uncommon for older hardware
[08:51] <twb> I thought only Windows did that
[08:55] <kaushal> ok
[09:01] <kaushal> _ruben: so i have to settle it for 3G ?
[09:02] <kaushal> anyways
[09:03] <kaushal> thanks
[09:03] <kaushal> 3Gigs should be sufficient
[09:03] <kaushal> _ruben: any wiki to setup gateway or router
[09:04] <_ruben> sudo sysctl -w net.ipv4.ip_forward=1
[09:04] <_ruben> done
[09:05] <kaushal> I mean some document or wiki
[09:07] <koolhead11> _ruben, that turns your machine into router isn`t it?
[09:08] <_ruben> koolhead11: indeed
[09:09] <_ruben> kaushal: google probably has a huge list of those, it all depends on what you want exactly, it's not like you're giving out a lot of info here
[09:09] <kaushal> sure
[09:10] <kaushal> Thanks
[09:10] <kaushal> much appreciated
[09:40] <jussi> I can ask also in the mozilla channel, however, does anyone know how to get bugzilla to send via gmail?
[09:41] <twb> Uh, surely bugzilla is like any other well-behaved unix service, and sends via /usr/sbin/sendmail?
[09:42] <twb> IOW: configure your damn MTA
[09:42] <jussi> twb: it can, yes. However, it also can use a smtp service - and Id like it to do so. the issue is with the tls authentication mail uses.
[09:43] <twb> Sanity check please: for a headless VM server, do I need virt-*manager* installed, or do I only need to install that on the desktop side?
[09:43] <twb> jussi: IME that is more hassle than it's worth
[09:48] <_ruben> jussi: you want to go through your local mta, just in case the remote smtp service isn't available for whatever reason, your local mta would queue the mail, unless that one's down as well
[09:53] <twb> Good point
[10:18] <twb> To answer myself: no, you only need libvirtd-bin on the server side
[10:19] <soren> twb: Indeed. libvirtd and an ssh server, and you're all set.
[10:19] <twb> Except that virt-manager disconnects from my pty, so it fails to ask for my password
[10:19] <soren> You're not using keys?
[10:19] <twb> (It's *stupid* to assume that just because I'm using X, I have a full desktop.)
[10:20] <twb> soren: it worked once I ran "ssh-add" on the remote server on which virt-manager was installed.
[10:20] <soren> It should be using ssh-askpass, too.
[10:20] <twb> ssh-askpass is GUI, right?
[10:20] <twb> There's no SSH equivalent of pinentry-curses
[10:20] <soren> You have virt-manager running on a remote server? Uh.. Why?
[10:20] <twb> soren: because I do not want to pollute my personal laptop with such frippery
[10:20] <soren> Sorry, I'm missing something. If you can run virt-manager, why can't you run ssh-askpass?
[10:20] <twb> soren: because it isn't installde
[10:21] <soren> Ah.
[10:21] <twb> It's not installed because virt-manager is installed on a VM, not on a full-blown GNOME desktop
[10:21] <soren> That seems a bit silly.
[10:21] <twb> And I'm using xinit /usr/bin/ssh -X twb-scratch virt-manager
[10:21] <twb> soren: it's a VM on a different VM system
[10:22] <twb> soren: I *would* have installed it directly on the user's system who will be using virt-manager, but that's Fedora Core 7 and I don't think that will be easy
[10:22] <soren> No, nothing is easy in that context, I imagine.
[10:22] <twb> But that doesn't detract from my assertion that it's *wrong* for virt-manager to assume it's installed on a full-blown desktop with ssh-askpass
[10:23] <soren> I think it's reasonable to assume what is true in 99% of the cases, and consider corner cases like yours corner cases, and hence sort out the general case first and leave the rest as wishlist items.
[10:23] <twb> Bah
[10:24] <twb> Aaaaaanyway, where can I find instructions on migrating vmware-server 1.x VMs to libvirt/kvm?  I asked #virt on OFTC but they're asleep.  I know how to use qemu-img for migrating to plain kvm, but not how to write the XML file for libvirt
[10:24] <soren> virt-goodies (and Ubuntu package) has a script that converts vmx files to libvirt xml.
[10:25] <soren> courtesy of jdstrand (*hugs*)
[10:25]  * twb looks
[10:25] <twb> It recommends munin-node?  How odd.
[10:26] <soren> That's a point of view.
[10:26] <twb> Ah, because it provides munin monitoring bits
[10:26] <soren> Right.
[10:27] <twb> Which is fair enough
[10:35] <twb> Looks like virtd also doesn't like me naming my ifaces meaningfully
[10:35] <twb> 1 theta libvirtd: 18:56:42.547: error : udevStrToLong_ui:73 : Failed to convert '009' to unsigned int#012
[10:46] <twb> soren: OK, vmware2libvirt is sensible.  But what's the standard locations for the disk images and the XML files?
[10:46] <twb> soren: also, can I use an LV instead of a disk-image-on-a-file ?
[10:47] <soren> twb: LV's are fine, yes.
[10:47] <soren> twb: The only somewhat-standard location is /var/lib/libvirt/something-that-I-don't-remember
[10:47] <soren> ...but I don't think anyone really uses it.
[10:47] <twb> OK
[10:48] <twb> I guess I dig out the virt docs and work out how to change <disk type='file' device='disk'> for LV
[10:48] <soren> ...since we, in Ubuntu, generally don't run as root, and thus can't easily create disk images there.
[10:48] <soren> Don't change anything (other than the path).
[10:48] <twb> Huh, OK
[10:48] <twb> Also, <on_poweroff>destroy</on_poweroff> is a bit worrying
[10:52] <soren> destroy isn't as destructive as it sounds.
[10:52] <twb> http://libvirt.org/formatdomain.html suggests it's OK
[10:52] <soren> It just means "shut down", really.
[10:52] <soren> Yeah.
[10:52] <twb> Yay for making up new terminology that fights old terms :-/
[10:52] <soren> Heritage from Xen.
[10:52] <twb> Ah
[10:53] <twb> That'd be why they're referring to "domains"
[10:53] <soren> Hence the talk of "domains" rather than e.g. "virtual machines".
[10:53] <soren> Right.
[10:53] <twb> I didn't connect because they didn't just say "dom" :P
[10:53] <soren> You're not alone :)
[10:59] <twb> But hang on, if there's no standard location for the XML files, how does libvirtd find them when it boots?
[11:00] <twb> Oh, I see, virsh is expected to do that part
[11:11] <soren> twb: Oh, for the xml files? Sorry, I thought we talked about disk iamges.
[11:12] <soren> twb: You "define" your xml files. That imports them into libvirt.
[11:12] <soren> twb: They land in /etc, but don't touch them. If you want to edit them, use "virsh edit" or "virsh dumpxml" followed by "virsh define".
[11:13] <twb> soren: sorry I meant both
[11:14] <soren> twb: You did say so, I just missed it.
[11:14] <twb> I shouldn't edit them why?  Because virtd will overwrite my changes when it exists?  or just so that "virsh edit" does proper locking?
[11:14] <soren> twb: Those are two of the reasons.
[11:14] <twb> okey dokey
[11:15] <soren> twb: libvirt expects to own those files. If you edit them, libvirt might overwrite the changes, and will certainly not read your changes at runtime. virsh dumpxml and define are the correct interfaces to edit the files. virsh edit is a clever wrapper for those two.
[11:15] <twb> Grmph
[11:15] <twb> I do not like systems that try to be clever
[11:16] <twb> Although I suppose then I shouldn't be runing virtd at all, just doing it by hand
[11:16] <soren> The idea is fine, the problem is just that the files end up in /etc (and look like they're supposed to be edited).
[11:16] <twb> Yeah, putting them in /var/lib would fix that
[11:16] <soren> Being in /etc make them managed by etckeeper, though, which is rather handy.
[11:16] <twb> Nod
[11:16] <twb> I am a biiiiiiig fan of etckeeper
[11:17] <soren> Anything that makes up for my lack of memory is a big win, so me too.
[11:17] <twb> (Not so much ubuntu changing the default backend to bzr, but whatever.)
[11:17]  * soren <3 bzr
[11:17] <twb> I will concede that bzr is probably tolerable if you have a lp login, you tell bzr about it, and you store everything inside lp
[11:18] <twb> The #bzr people basically told me that if you don't, you don't get the bzr smart server (over SSH), which is why my performance sucked so hard.
[11:19] <twb> That and it requires around 750MB of RAM to bzr branch Emacs' repo :-/
[11:28] <twb> Should I be worried if hardy's "qemu-img info foo.vmdk" doesn't realize the vmdk is split into a shitload of -s0001.vmdk bits?
[11:28] <twb> I thinks foo.vmdk is a 4kB raw file
[11:29] <soren> You could help it along a bit and tell it that it's a vmdk.
[11:30] <twb> Oh you're kidding
[11:30] <twb> my non-root user has read access to *some* of the -sNNN files
[11:31] <twb> Unless those -NNNNNN-MMM files are snapshots or something?  My brain hurts.
[11:32] <twb> http://paste.debian.net/113661/
[11:33] <twb> Oh, apparently those are a second HDD
[11:34] <twb> Aaaand the first disk isn't used anymore.
[11:40] <twb> OK, WTF.  As root I can run "file" on all of the vmdk files, but "qemu-img -f vmdk foo.vmdk" is complaining that it can't read them
[11:41] <twb> http://paste.debian.net/113662/
[11:44] <patdk-lap> twb, those look like the 2gig max, split vmdk format
[11:44] <twb> patdk-lap: they are
[11:45] <twb> patdk-lap: I want to convert them to raw format so I can put them on LVs
[11:45] <patdk-lap> why use qemu to do it?
[11:45] <twb> patdk-lap: what else is there?
[11:45] <patdk-lap> why not just use vmware to convert it
[11:45] <twb> Because I will not touch vmware unless I absolutely have to
[11:46] <twb> Call that "plan B" if you wil
[11:46] <patdk-lap> heh, I never use split mode
[11:46] <twb> I'm migrating from legacy crap that I didn't set up
[11:46] <soren> strace?
[11:47] <twb> soren: I tried that but I couldn't see what was wrong.
[11:47] <soren> twb: Can you share it?
[11:47] <twb> soren: now I'm trying to just copy the full directory and then try on the copy, in case it's getting confused about something.  (I'm not allowed to just chown and chmod)
[11:47] <twb> soren: share the VM?
[11:47] <soren> twb: The strace output.
[11:47] <twb> Oh sure, one moment
[11:50] <twb> open("fish (DO NOT USE).vmdk", O_RDWR|O_CLOEXEC) = -1 EACCES (Permission denied)
[11:51] <twb> WTF, qemu-img
[11:52] <twb> http://paste.debian.net/113664/
[11:53] <soren> Not sure why it wants RDWR.
[11:53] <twb> It's even doing it in the full cp -r I made
[11:53] <twb> Maybe it doesn't understand space and aprens and shit
[11:55] <soren> I have a hard time imagining a bug where space and parens would case it to change the mode with which it attempts to open a file :)
[11:57] <twb> Changing it to fish.vmdk (including the slice names in the index text file) didn't help
[11:57] <soren> You do have write access to the ones in the copy, don't you?
[11:58] <twb> Yes.
[11:58] <twb> I'm root, and I just chown root: and chmod 755 for good measure
[11:58] <soren> Can you run it with strace again, please?
[11:58] <twb> qemu-img *can* read the individual slices, and correctly identifies them as sparse vmdk 2G images
[11:58] <twb> soren: sure
[11:59] <twb> http://paste.debian.net/113665/
[12:00] <soren> Ah.
[12:00] <soren> Different problem, I guess.
[12:00] <soren> Can you share fish.vmd?
[12:00] <soren> vmdk, even.
[12:01] <twb> Just the index file, you mean?
[12:01] <soren> Yes.
[12:01] <twb> http://paste.debian.net/113668/
[12:02] <soren> Oh, you /did/ change that, too :)
[12:02] <twb> Now I'm just trying this: qemu-img convert -f vmdk -O raw fish-s*.vmdk /mnt/scratch/fish.img
[12:02] <twb> It seemed to work, dunno what it created though
[12:02] <soren> *chuckle*
[12:02] <soren> frankendisk.
[12:03] <twb> Blood users
[12:03] <twb> *blooody
[12:03] <twb> You know, the only reason I'm doing this is because we need Windows to talk to the Australia Tax Office -- who are windows out the wazoo
[12:05] <twb> Hmm, stupid question -- does libvirt need raw format just because it's LV?
[12:05] <twb> Probably using qcow2 format on top of LV would be better...
[12:12] <soren> twb: No, that won't work.
[12:12] <soren> and why would that be better?
[12:12] <soren> (ignoring the fact that it won't work)
[12:12] <twb> Because you can't have sparse LVs
[12:12] <twb> So I would have to allocate a couple hundred GB up front
[12:12] <soren> Yes...
[12:13] <twb> I would just prefer not to, that's all
[12:13] <soren> How do you expect they would grow if they held a qcow2 inside?
[12:13] <twb> soren: it explodes if it tries to grow  too big
[12:13] <soren> twb: And that's your desired outcome?
[12:13] <twb> Sure
[12:13]  * twb is not thinking too clearly at this point
[12:14] <twb> FWIW, qemu-img convert on the slices got enough for "chmod 400 fish-1.img; kvm -hda fish-1.img -vnc :0" to get as far as a window koops screen, so it seems to have worked at least partially.
[12:14] <soren> You don't have to convert to raw first. qemu-img can convert directly onto a raw device.
[12:14] <soren> like an lv.
[12:15] <twb> soren: like "qemu-convert -f vmdk -O raw fish-s*.vmdk /dev/theta/fish" ?
[12:15] <soren> There's a "host_device" "format".
[12:15] <soren> Or something to that effect.
[12:15] <twb> Hum
[12:15] <soren> it's the same thing, really.
[12:16] <twb> OK
[12:19] <twb> Ahaha
[12:20] <twb> http://paste.debian.net/113672/
[12:20] <twb> Apparently default units for LVM is the pitibyte
[12:20] <twb> Er, pibibyte
[12:20] <twb> pebibyte?  Whatever.
[12:22] <_ruben> no, apparantly its default is 1/millionth of a PiB
[12:22] <twb> Oops, yeah
[12:22] <_ruben> billionth
[12:23]  * soren always goas "-L 40G" or whatever.
[12:23] <soren> s/goas/goes/
[12:23] <twb> I would if it was a new disk
[12:25] <maswan> twb: "Default unit is megabytes" as per docummentation. And I always use explicit units too.
[12:25] <soren> twb: Whyh?
[12:27] <_ruben> conversion of existing disks, not that odd to specify the new location's size in bytes
[12:27] <soren> Oh.
[12:27] <soren> Right, right. Got it.
[12:27] <soren> Good point.
[12:27]  * twb switches back from "panic" to "grumbling"
[12:27] <_ruben> heh
[13:19] <TeTeT> jamespage1: hi there, been toying around with Jenkins - pretty nice. However, I stumble over installing the 'violations' plugin on testing from the PPA. The Web UI tells me it is installed, but I don't see it listed in the plugins nor is there anywhere the violations checkbox for any project
[13:27] <zul> morning
[13:32] <TeTeT> jamespage1: I wrote a minimal python script and stored it in lp and used Jenkins to test it, works really nicely
[13:32] <TeTeT> hi zul, any updated on nova / lxc? E.g. should I test again?
[13:33] <zul> TeTeT: sure if you want
[13:33] <TeTeT> zul: ok, I'll update and let you know if it works today :)
[13:33] <jamespage1> TeTeT: hey - glad its working for you - I'll take a look at the violations plugin and see
[13:33] <zul> TeTeT: cool
[13:39] <jamespage1> TeTeT: OK - I think that you will need to install the Maven Project plugin; looks like its a dependency
[13:39] <jamespage1> TeTeT: /var/log/jenkins/jenkins.log should be complaining on startup
[13:43] <TeTeT> jamespage1: thanks, I'll give it a try. Didn't know about the jenkins log file!
[13:44] <jamespage1> TeTeT:  so you have hit one difference between the Ubuntu packaging and upstream; I was not able to include the maven-plugin by default as its not buildable from source  - no Maven3 in archive
[13:45] <Daviey> hallyn, Hey... is bug 742770, just needing uploading?
[13:49] <hallyn> Daviey: that is fixed.  Doesn't lp automatically update status of any bug listed in changelog when package is pushed to archive?
[13:49] <ScottK> SpamapS: Your drizzle package seems to build better than the one Monty uploaded to Debian.  If you were perhaps interested in preparing an NMU, I'd sponsor it.
[13:50] <hallyn> Daviey: or does the person pushing have to do that by hand?
[13:51] <hallyn> of course it wasn't the latest anyway.
[13:51] <hallyn> I"ll change the bug status :)
[13:55] <Daviey> hallyn, hmm.. i guess there was an issue with syntax
[13:55] <soren> hallyn: That's generally the case, yes. Sort of.
[13:55] <hallyn> well the upload did fix 4 bugs at once
[13:56] <soren> hallyn: dpkg-gencontrol parses the changelog entry (or entries), and adds a x-launchpad-bugs-fixed field to the _source.changes files.
[13:56] <soren> s/files/file/
[13:56] <soren> This gets read by Launchpad, which then goes to close the bugs (if the bug has a bug task for the given source package).
[13:56] <hallyn> hm, actually the changelog is a bit messed up - it has two duplicate entries (with sequential version #s)
[13:56] <soren> So Launchpad doesn't per se read the changelog entries.
[13:57] <Daviey> hallyn, what upload version was it?
[13:57] <hallyn> right now it is at 0.7.4-0ubuntu6
[14:01] <hallyn> zul: so something went wrong with the lxc changelog.  It's different int he package from in my source tree.  See the entries for versions 0.7.4-0ubuntu3  and 0.7.4-0ubuntu2
[14:01] <soren> hallyn: It looks like ubuntu2 and ubuntu3 were never uploaded.
[14:01] <zul> hallyn: ok ill have a look
[14:02] <soren> hallyn: Judging by https://launchpad.net/ubuntu/+source/lxc/+publishinghistory
[14:02] <soren> dpkg-gencontrol by default only parses the top-most changelog entry.
[14:03] <hallyn> soren: ok, that explains that then, thanks :)
[14:03] <soren> It doesn't ask launchpad (or anything else) what the most recent version is, so you're supposed to pass -v<most recently uploaded version> to dpkg-buildpackage to make sure it includes older entries.
[14:04] <soren> hallyn: sure thing.
[14:04] <hallyn> soren: oh, so when you do that it would change the bug status for all changed entries?
[14:04] <hallyn> zul: I'm comparing to http://bazaar.launchpad.net/~serge-hallyn/ubuntu/natty/lxc/lxc-fix-3bugs/view/head:/debian/changelog
[14:04] <soren> hallyn: Yes.
[14:05] <zul> hallyn: grr....sorry about this ill fix it today
[14:05] <hallyn> zul: oh, i see there are actually 2 entries for ubuntu2  :)  so no information actually lost, no big deal :)
[14:05] <hallyn> thanks
[14:42] <RoAkSoAx> morning all
[14:58] <Daviey> smoser, Are you able to confirm bug 712026 easily?
[14:59] <smoser> it would be impossible to completely confirm
[14:59] <smoser> it happend ~ 4% of the time
[14:59] <Daviey> hallyn / zul: Is the debootstrap entry on bug 740167 invalid?
[15:00] <Daviey> smoser, Ah, so perhaps jamespage should see it in jenkins for b2?
[15:00] <zul> i think so it got worked around
[15:00] <smoser> well, we should see lack of it.
[15:00] <smoser> Daviey, i would suggest we should try running a bunch of instances to try to catch it.
[15:01] <jamespage> smoser, Daviey: do you want me to try running a subset of the tests to reproduce?
[15:01] <Daviey> smoser, Is euca enough, or do you think it's limited to aws?
[15:01] <jamespage> multi-instance might be a good one :-)
[15:01] <hallyn> Daviey: i blievei t should be valid, and i believe it is still a real bug in debootstrap
[15:01] <Daviey> jamespage, I'll defer to smoser's recommendation :)
[15:01] <smoser> i have never seen it on euca. its timing. so its either more likely, less likely, or as likely on eucalyptus
[15:02] <Daviey> smoser, "There are known unknowns that we know we don't know"
[15:03] <smoser> jamespage and i only ever saw it occur on i386.  so i would suggest that we launch 20 instances in each of 5 regions of i386 m1.small and attempt to connect and  collect logs.
[15:04] <smoser> that would seem like a reasonable effort
[15:04] <jamespage> Right-oh
[15:04] <smoser> however, we might not have -server images with that kernel at the moment.
[15:04] <jamespage> I'll hold off for the moment - I'll need to just check on the setup anyway
[15:12] <smoser> Daviey, i'm pretty sure, that ther eis no updated kernel published yet
[15:12] <smoser> right ?
[15:13] <jamespage> smoser - will that blow my account limits? 20 in each region....
[15:13] <smoser> it might. i dont know.
[15:14] <smoser> i can run them if you'd like
[15:14] <smoser> i dont recall if the default is 10 or 20
[15:14] <smoser> they are per-region, though
[15:14] <Daviey> smoser, not sure
[15:15] <ttx> hey guys, testing Natty current uec-image / LXC / Nova, have a few issues I discussed with zul already
[15:16] <ttx> The container starts, but there is a sleep 60 somewhere and I can't ping it until 70 seconds after start
[15:16] <ttx> then there are quite a few missing files affecting operation...
[15:16] <zul> also the ssh seems to be stopping for some reason
[15:16] <ttx> /etc/default/locale, /etc/ssh/ssh_host_{dsa,rsa}_key[.pub]
[15:17] <ttx> ~ubuntu/.ssh/authorized_keys
[15:17] <ttx> trying to make sure it's an image issue, not somethign we need to fix on our side
[15:18] <ttx> let me know if that rings a bell.
[15:24] <smoser> ttx, zul i think you're confused.
[15:24] <smoser> zul told me on friday that they worked perfectly ;-)
[15:24] <zul> smoser: wth kvm
[15:24] <zul> smoser: this is with lxc now...not so much
[15:24] <smoser> i suspect the sleep 60 is form cloud-init
[15:25] <smoser> that its waiting for a nework device to come up
[15:27] <zul> hallyn/smoser: im getting something weird in my log files ill pastebin it in a sec
[15:30] <zul> smoser/hallyn: http://pastebin.ubuntu.com/592646/
[15:31] <smoser> i would qualify that as a little more than "weird"
[15:31] <smoser> you said you had tested this, right zul ?
[15:32] <zul> smoser: i did test it...not with lxc recently though
[15:32] <smoser> ttx, zul, there is, explicitly a 60 second timeout in /etc/init/cloud-init-nonet.conf
[15:32] <smoser> well, even 70 seconds (10 + 60)
[15:33] <smoser> so i suspect that is not functioning well with lxc.
[15:33] <ttx> smoser: i could live with the 70second-timeout... if I could ssh in afterwards
[15:33] <ttx> but due to the missing files I can't
[15:33] <smoser> well, you probably have messages lke "gave up waiting for a network device."
[15:33] <smoser> in your ocnsole logs
[15:34] <jMCg> Hey folks, I'm having trouble booting a KVM (both host and guest are Ubuntu 10.10) http://dpaste.com/530912/ -- the vm just hangs there..
[15:34] <ttx> smoser: no... http://paste.ubuntu.com/592627/
[15:34] <hallyn> zul: what am i looking at?
[15:35] <ttx> smoser: I have network up after 70 seconds.
[15:35] <ttx> I can ping, but ssh fails because of missing /etc/default/locale, /etc/ssh/ssh_host_{dsa,rsa}_key[.pub]
[15:36] <hallyn> zul: what's weird about it?
[15:36] <zul> hallyn: the mknod failing
[15:36] <hallyn> zul: libvirt-lxc only creates the 1 tty, so tty2-4 won't exist.
[15:36] <zul> hallyn: ok
[15:36] <hallyn> which mknod is worrying you?
[15:37] <zul> hallyn: gimme a sec im on a call now
[15:37] <hallyn> me too
[15:40] <hallyn> kirkland: Daviey: re bug 700511, I'm about to try just syncing debian's vgabios verbatim.  They have a new version now which does the vmware+qxl bioses, so I think its the way to go.
[15:40] <hallyn> any complaints?
[15:41] <kirkland> hallyn: hrm, i always check with aliguori privately on vgabios syncs/merges, he knows which ones are good and bad, usually
[15:42] <kirkland> hallyn: but along those lines...
[15:42] <kirkland> hallyn: i'd think that vgabios would be one thing that we could have a checklist of things we need to verify before uploading
[15:42] <kirkland> hallyn: ie, a matrix of vga modes, resolutions, and OSes
[15:44] <hallyn> kirkland: sure, that makes sense.  but at the same time compared to other things that seems very low priority.
[15:44] <kirkland> hallyn: extremely low, yes, sorry, I didn't mean "let's do that now" :-)
[15:44] <hallyn> which is really why I'd be very happy to be back at a zero-delta debian sync
[15:45] <SpamapS> ScottK: re the drizzle package for debian, ACK, I will take a look at it later today.
[15:45] <ScottK> Thanks.
[15:46] <ScottK> I didn't compare the Ubuntu and Debian packaging.  No idea how different it is, but yours at least builds.
[15:47] <ppetraki> jMCg, get to the grub prompt, erase the quiet arg and add '--debug'
[15:47] <SpamapS> ScottK: I think in Debian its still a pre-release not the GA (2011.03.13)
[15:48] <ScottK> No.  It's the same upstream version.
[15:48] <SpamapS> Interesting
[15:58] <jMCg> ppetraki: that does quite a bit.. I can pastebinit.
[15:58] <ppetraki> jMCg, we really only need to see the last bit, where it hangs
[15:58] <ppetraki> jMCg, pb is good
[15:59] <jamespage> Daviey: merges proposed as we discussed earlier - thanks...
[16:00] <jMCg> ppetraki: http://pastebin.com/zQWFG3SW
[16:02] <Daviey> jamespage, cool
[16:14] <ppetraki> jMCg, looks like the filesystem is mounted, and network is up. If you press enter a couple times, does a prompt show up?
[16:18] <jMCg> ppetraki: nope.
[16:19] <ppetraki> jMCg, yeah, I couldn't tell you. The startup looks fine, I can't tell why you're not getting a prompt, unless it's something like the console is on the wrong tty
[16:26] <jMCg> ppetraki: the XML: http://pastebin.com/TeJ649ZA
[16:26] <ppetraki> jMCg, what do you need " serial=tty0" for? the console= should be enough
[16:27] <jMCg> Oh.. duh.
[16:27] <jMCg> I forgot to put a /etc/init/ttyS0.conf in there
[16:28] <ppetraki> jMCg, I was wondering why I didn't see that in the init logs ;)
[16:31] <zul> soren: im going to upload a newer version of glance to the archive today fui
[16:31] <zul> soren: s/fui/fyi/
[16:39] <jMCg> ppetraki: thanks for your help -- now to find out why the networking setup doesn't work.
[16:40] <ppetraki> jMCg, :), good luck
[16:41] <smoser> hallyn, zul it would appear to me that (per http://pastebin.ubuntu.com/592646/), /dev/ is read-only
[16:41] <smoser> dev should be mounted RW by either or ramdisk
[16:41] <zul> smoser: that occured to me as well
[16:41] <smoser> either kernel or ramdisk
[16:42] <smoser> what kernel is this ?
[16:42] <smoser> your'e trying to run this on a lucid host
[16:42] <smoser> i suspect
[16:42] <zul> smoser: lxc container on natty
[16:42] <smoser> well, something needs to mount /dev as a tmpfs
[16:43] <zul> smoser: right
[16:43] <smoser> err... devtmpfs.
[16:43] <smoser> ttx, did you see that mknod stuff in your logs ?
[16:44] <hallyn> the mknod stuff could just as well be due to devices cgroup
[16:44] <hallyn> zul: can you fire off a job to check /dev/?
[16:44] <hallyn> zul: it's possible that this was broken to my change to lxcguest.
[16:44] <zul> i have the image still mounted if that helps
[16:45] <hallyn> cools o what is under /dev?
[16:45] <hallyn> as in, output of df and 'mount'
[16:45] <zul> http://pastebin.ubuntu.com/592674/
[16:46] <zul> should i chroot into it to get you that?
[16:46] <hallyn> no that won't suffice
[16:46] <hallyn> can you log into the console?
[16:46] <ttx> smoser: no I didn't see that
[16:46] <zul> hallyn: lemme try
[16:47] <smoser> ttx, do you have access to look at the filesystem ?
[16:47] <ttx> I can restart an instance for that, yes
[16:47] <smoser> i suspect that /var/run/network/ifstate does not get updated
[16:48] <smoser> and so the "is it already up" check of cloud-init-nonet.conf (grep -qv '^lo' /var/run/network/ifstate) just fails, and then the network device doesn't come up after that (possibly because its already up)
[16:48] <hallyn> smoser: lxcguest can't make /var/run a tmpfs for now, yes
[16:48] <hallyn> so you just need to manually create /var/run/network
[16:49] <zul> hallyn: http://pastebin.ubuntu.com/592678/
[16:49] <smoser> well, something will need to populate it for this stuff to work.
[16:49] <smoser> (normally that is populated via ifup)
[16:50] <hallyn> right so why doesn't ifup do it here?
[16:50] <hallyn> lxc containers are coming up fine, so i'm confused as to why you're having troubles
[16:51] <ttx> smoser: anything in particular you want me to look at in the container fs ?
[16:51] <hallyn> zul: how do you create the basic rootfs which gets used?  can you trivially add another package?
[16:51] <hallyn> zul: first off, this will cease to be a problem once we get proper shutdown/reboot support for lxc
[16:51] <zul> hallyn: i just mount through qemu-nbd basically
[16:51] <hallyn> zul: the problem is that mounting tmpfs over /var/run stops lxc init's parent from seeing the guest's utmp and seein gif it is a reboot or shutdown
[16:52] <hallyn> zul: so basically you can either:
[16:52] <hallyn> 1. uncomment the lines in /lib/init/fstab.lxc
[16:52] <hallyn> 2. install your own /etc/init/libvirt-lxc.conf which works around this
[16:52] <hallyn> is (1) pretty easy to test?
[16:52] <zul> hallyn: i can install packages in it without a problem
[16:53] <ttx> smoser: no /var/run/network directory
[16:53] <hallyn> zul: ok, i guess create your own lxcguest real quick with a fstab.lxc with the lines uncommented
[16:54] <zul> hallyn: ok
[16:54] <hallyn> zul: so it's ok for this to be a one-off for now?
[16:55] <hallyn> zul: or do i need to think of a clean way to handle this generically?
[16:57] <zul> hallyn: i think we need a clean way to handle this generically since it might bite us in the future again
[16:57] <hallyn> well it'll be a shorterm thing in any case
[16:57] <hallyn> i'm still curious about the actual problem
[16:58] <hallyn> i.e., why does lxc seem to create /var/run/network just fine while libvirt-lxc does not?
[16:58] <smoser> ttx, yeah, so zul/hallyn are going to have to sort that out..
[16:58] <hallyn> are you changing anything in the upstart jobs?
[16:58] <hallyn> or is this a pristine natty guest plus lxcguest?
[16:58] <zul> its a pristine natty plus lxcguest
[16:59] <zul> hallyn: because it doesnt use the lxc-templates in lxc
[16:59] <ttx> smoser, zul, hallyn: ok, let me know when/what I can retest... tomorrow.
[16:59] <hallyn> but the lxc natty template doesn't play with those
[17:00] <zul> hallyn: right thats what im saying :)
[17:00] <hallyn> zul: so waht would be the diff between lxc and libvirt starting it
[17:01] <zul> hallyn: im not sure
[17:01] <hallyn> zul: you can trivially change the string you put in init's ENV right?
[17:01] <hallyn> (for 'container=lxc' or whatever?)
[17:02] <zul> hallyn: i think so
[17:02] <hallyn> if so, it's sort of admitting that we'll always have to be hacky, but we could make libvirt do 'container=libvirt'
[17:02] <hallyn> then lxcguest can DTRT
[17:02] <zul> hallyn: that might be what needs to happen
[17:02] <zul> and then get a proper fix after natty
[17:03] <hallyn> yeah like i say we're going to fix the proper reboot/shutdown hopefully before lxc sprint, but if need be then at lxc sprint
[17:03] <zul> hallyn: agreed
[17:03] <hallyn> zul: can you make the libvirt change?  I'll do the lxcguest change and get it to you to push?
[17:04] <zul> hallyn: sure if i need where to look
[17:04] <ScottK> python-psutils is in Main now.
[17:04] <hallyn> zul: say huh?
[17:04] <RoAkSoAx> ScottK: thanks!
[17:05] <hallyn> zul: ok, biab.
[17:07] <ppetraki> FYI ALL, if anyone is having problems with SANs on Ubuntu, I'd like to hear about it.
[17:09] <zul> ppetraki: you probably get some more traction if you send it to ubuntu-server ml
[17:11] <ppetraki> zul, true
[17:51] <kirkland> RoAkSoAx: ping
[17:51] <RoAkSoAx> kirkland: pong
[17:51] <kirkland> RoAkSoAx: https://bugs.launchpad.net/ubuntu/+bug/732759
[17:51] <kirkland> RoAkSoAx: i'm checking status on that
[17:52] <kirkland> RoAkSoAx: looks like pitti gave you the FFe on 3/15
[17:52] <kirkland> RoAkSoAx: any thing uploaded?
[17:52] <RoAkSoAx> kirkland: yeah.. https://launchpad.net/ubuntu/+source/python-ethtool/0.6-0ubuntu1
[17:52] <RoAkSoAx> kirkland: I think the upload never closed the bug report
[17:53] <kirkland> RoAkSoAx: k, i'm going to close it then
[17:53] <RoAkSoAx> kirkland: but as far as I can remember it was marked as Fix Released for Ubuntu and New for Debian, so maybe they messed something up when they were working on launchpad
[17:53] <kirkland> RoAkSoAx: thanks.
[17:55] <RoAkSoAx> kirkland: btw... you were the archive admin that approved the upload :)
[18:07] <SpamapS> hggdh: not sure why bug #735072 is assigned to canonical-server .. what are we going to do?
[18:12] <zul> SpamapS: rewrite netbios ;) samba already truncates the hostname
[18:14] <SpamapS> It seems like this needs to be handled as the user requests.. by warning/limitting in the installer.
[18:14] <zul> SpamapS: ditto ive already argued that point before
[18:21] <hggdh> SpamapS, zul: I am not sure why we have to limit the host name -- it goes against the RFC
[18:23] <zul> hggdh: right its a netbios limitation
[18:24] <hggdh> SpamapS, zul: I can see this being done on, say, the samba post-installer
[18:24] <hggdh> and the task is on samba(ubuntu)
[18:25] <zul> hggdh: what like a hostname | wc -m and warn install if its more than x number characters
[18:26] <hggdh> zul: or just truncate it?
[18:26] <zul> hggdh: no because its going to do that anyways
[18:26] <hggdh> adn warn it has been truncated
[18:27] <hggdh> zul: I am confused, then: what is truncated, the system hostname, or the netbios name?
[18:27] <zul> hggdh: my understanding is that it uses the hostname for the netbois name and it gets truncated when using netbios
[18:28] <hggdh> zul: and so, what is the bug about, then?
[18:28] <hggdh> or is it a SRU thingy?
[18:28] <zul> hggdh: that netbios is not respecting the rfc
[18:29] <hggdh> zul: well, this is how it is, nothing we can do without breaking it for everybody else
[18:29] <zul> hggdh: agreed
[18:30] <hggdh> zul: could you then please add your comments -- the hostname is already truncated, and netbios is happy, etc, etc?
[18:31] <zul> hggdh:
[18:31] <zul> yes
[18:37] <hggdh> zul: and, of course, that it does not make sense to limit the hostname unilaterally -- the user can select that, and so on
[18:37] <zul> hggdh: yeah im thinking of doing a warning in postinst
[18:37] <axisys> hi all.. any beneft in having ssds on ubuntu server? on solaris I can put the ZIL on ssd to improve write IOs
[19:20] <ScottK> netbios name doesn't have to be hostname.
[19:20] <ScottK> That's an implementation convenience.
[19:25] <hggdh> ScottK: agreed. The only thing left was to decide what approach to take.
[19:26] <ScottK> I vote not enforcing non-RFC limits on hostnames.
[19:26] <ScottK> Samba and netbios can do whatever they want.
[19:30] <hggdh> +1
[19:48] <ppetraki> axisys, beyond the obvious by way of random read/writes being about the same as seq rw?
[19:48] <Scunizi> How do I verify the validity of an email address given as part of the registration process on a web site?  assuming that I have a comma delimited file that I can use to feed the tool addresses?
[19:49] <Trainbird> hi everybody
[19:50] <Trainbird> I wanted to know if its possible to install ubuntu server on a RAID 1 environment
[19:50] <axisys> ppetraki: how do I implement it?
[19:51] <axisys> ppetraki: in linux
[19:51] <ppetraki> axisys, implement what exactly? what I described is a property of the media/firmware, it's OS independent
[19:51] <sss314> [noob] Is there a software for Ubuntu for streaming videos/music?
[19:53] <Scunizi> sss314: lots..
[19:53] <sss314> Scunizi, rephrase: What easy software wou.d you recommend fo making a streaming server?
[19:53] <axisys> ppetraki: is there way I can use SSDs on linux to allow async IOs to write on it?
[19:54] <ppetraki> axisys, well, yes, there is an async-io api
[19:56] <RoyK> Trainbird: it is - either software raid or hardware raid
[19:56] <ppetraki> axisys, I don't get what you're trying to deploy though, there's so many options with Linux, you've got to pick a fs or sw raid stack first, and then start tweaking
[19:56] <Trainbird> software raid
[19:57] <RoyK> Trainbird: just choose manual partitioning, create one partition on each drive for each mirror, choose software raid config, create RAID1 groups, put filesystems/swap on them, done
[19:57] <axisys> ppetraki: i have two ssds in this server .. it was supppose to be used for solaris .. but we decided to go with ubuntu server.. it is sun fire x4170 m2
[19:57] <Tom___> I know that the Ubuntu-Desktop have alternate cd images for advanced installations but is this true for ubuntu server as well as I am trying to see how to install ubuntu server with raid on HP ML115 G5
[19:57] <axisys> ppetraki: so we are wondering how can we benefot from the two SSDs ..
[19:58] <axisys> ppetraki: this server will mainly run splunk
[19:58] <Trainbird> RoyK: and then it will also be possible to create new partitions to raid them for normal data-use?
[19:58] <RoyK> sure
[19:58] <ppetraki> axisys, splunk?
[19:58] <axisys> ppetraki: splunk index storage does not need SSDs unless the indexing is throttling.. which is not in our case
[19:58] <RoyK> Trainbird: it's just partitions and raid - it doesn't matter what you'll use them for
[19:59] <axisys> splunk: http://splunk.com is to index your logs
[19:59] <Trainbird> RoyK: okay, I'll give it a try :)
[19:59] <Tom___> I know that the Ubuntu-Desktop have alternate cd images for advanced installations but is this true for ubuntu server as well as I am trying to see how to install ubuntu server with raid on HP ML115 G5
[20:00] <RoyK> Tom___: the alternate install, now called the DVD install, is both desktop and server
[20:00] <RoyK> Tom___: if that server has a RAID controller, use that
[20:00] <Scunizi> sss314: depends on what you want to do.. check this out. http://www.google.com/linux?hl=en&q=streaming+server+ubuntu&btnG=Search
[20:00] <Tom___> is this available from cdimage.ubuntu.com if so can you give me the link so I can see exactly which image to sownload
[20:01] <RoyK> Tom___: do you need X?
[20:01] <Tom___> no
[20:01] <Tom___> cli is fine
[20:01] <RoyK> then just install from the server image
[20:01] <RoyK> you can setup RAID etc from that quite easily
[20:02] <RoyK> Tom___: but then, an ML115 might have a RAID controller?
[20:02] <Tom___> It has a hardware raid controller
[20:02] <Tom___> I guess its a case of rtfm
[20:02] <RoyK> then use that
[20:02] <RoyK> boot on the smartstart cd and setup raid with that
[20:03] <RoyK> ubuntu will see one drive, which is fine
[20:03] <ppetraki> axisys, so regardless of application, you can use something like blktrace to start getting data on your application, and take it from there
[20:04] <ppetraki> axisys, then again SSDs are fairly inexpensive at this point so the "try it and find out" methodology is relevant  too
[20:04] <Tom___> also one more thing out of all the raid levels which one would you suggest for a media server style of host i.e. high availability, robust mirroring or striping?
[20:05] <smoser> kirkland, i think you did the uec-seeds commit wrong
[20:06] <smoser> we needed to add 2 binary packages: cloud-initramfs-growroot and cloud-initramfs-rescuevol . it seems you added source package cloud-initramfs-tools (http://paste.ubuntu.com/592770/)
[20:07] <RoyK> ppetraki: using SSDs for servers is a waste unless you need the extra IOPS - they wear out far quicker than spinning rust
[20:07] <ppetraki> RoyK, I wasn't exactly recommending them :-p
[20:07] <ppetraki> RoyK, you don't get much warning when they fail either
[20:08] <zim> Hi all. I am trying to setup a raid 1 with 2 x 2TB drives. I have formated the whole 2TB on both to EXT4 then used configure software raid added both sdb1 and sdc1 to the raid but it is showing as unusable.
[20:08] <RoyK> ppetraki: that's why I use zfs for critical data storage :P
[20:09] <zim> is 2TB to big or am I missing somthing?
[20:09] <RoyK> no, it's not
[20:09] <ppetraki> RoyK, yeah, btrfs has cost me data 1-2 times now, not too happy with it
[20:09] <zim> I did not thinkk so as that was the ext3 limit
[20:10] <zim> what am I doing wrong?
[20:10] <RoyK> ext4 supports 1EiB
[20:10] <RoyK> should suffice
[20:10] <ppetraki> zim, you're using mdadm, software raid?
[20:10] <RoyK> ppetraki: btrfs isn't flagged stable
[20:10] <RoyK> ppetraki: zfs is rock fucking stable :)
[20:10] <zim> yup but setting it up with the installer. (as I have done before and all worked great)
[20:11]  * RoyK has some 380TiB on ZFS
[20:11] <ppetraki> RoyK, I know, but I was expecting just a little more stability, good thing I do nightly backups
[20:11] <RoyK> ppetraki: you shouldn't expect stability on software flagged as experimental
[20:12] <ppetraki> RoyK, the source of all disappointments  :-)
[20:12] <zim> ppetraki: what is wrong with ext4?
[20:12] <ppetraki> zim, oh nothing
[20:12] <RoyK> zim: he was using btrfs
[20:12] <zim> ok
[20:12] <eichi> http://pastebin.com/Vh3Jv43e someone can tell my, why apt-get install postfix does REMOVE so much packages?
[20:12] <ppetraki> zim, so these partitions are marked as "linux raid" and you added them to the array?
[20:13] <RoyK> eichi: because postfix replaces exim
[20:13] <RoyK> you don't want two MTAs on the same system
[20:13] <zim> nope. Free space -> make ext4 partition on both -> add to raid
[20:13] <zim> is that wrong?
[20:13] <ppetraki> zim, so there's your problem
[20:13] <ppetraki> zim, yeah
[20:13] <RoyK> zim: yes, that's wrong
[20:14] <RoyK> zim: flag the partitions as "use for RAID" (don't remember the exact text)
[20:14] <zim> kewl problem half solved dam crappy howto :)
[20:14] <ppetraki> zim, assuming you're using one partition to span the entire disk
[20:14] <ppetraki> we'll, we should have better docs too
[20:15] <zim> yup just want a 2tb /Data partion
[20:15] <zim> the os is on another drive
[20:15] <RoyK> zim: just create a partition on each drive and flag it for RAID use
[20:15] <eichi> RoyK: shouldnd be a problem for the system?
[20:15] <ppetraki> zim, it's partition tag "fd"
[20:16] <RoyK> eichi: the system wants an MTA, any will do
[20:16] <zim> I am doing it from the ubuntu-server 10.04 installer
[20:16] <RoyK> zim: should work well
[20:17] <RoyK> zim: I've installed software RAID on a dozen or so servers the latest year - works for me (tm)
[20:17] <zim> I have just deleted all partions so now have 2 tb free space on both
[20:17] <RoyK> zim: create partition - instead of choosing ext4, choose "use as physical ... for RAID"
[20:18] <zim> thats where I went wrong will get back if I have any other problem Many Thanks to you both.
[20:18] <ppetraki> np
[20:19] <ppetraki> RoyK, so are you using zfs on linux or do you have some solaris instances kicking around?
[20:19] <RoyK> ppetraki: openindiana
[20:20] <RoyK> zfs fuse sucks rather hard on write performance
[20:21] <ppetraki> RoyK, that's pretty cool, I never heard of that project before
[20:21] <RoyK> opensolaris fork
[20:21] <ppetraki> yeah, I'm reading the faq now
[20:21] <RoyK> :)
[20:21] <ppetraki> it's amazing how long linux has been around and we still don't have the fs thing figured out
[20:22] <RoyK> well, ext4 works, but btrfs should have stabilised ages ago
[20:22] <RoyK> but seems there's a key person at Oracle working on certain parts
[20:23] <RoyK> fsck for one
[20:23] <RoyK> also, they don't have any raid[56] code working
[20:24] <RoyK> I guess btrfs may stabilise in a few years, but only if development resources are allocated to it from someone like RedHat
[20:25] <RoyK> or Canonical
[20:31] <kirkland> smoser: oh, doh
[20:31] <kirkland> smoser: did you fix, or do you want me to?
[20:32] <smoser> i cannot commit, kirkland so please do
[20:32] <smoser> (no coredev)
[20:32] <soren> zul: Cool.
[20:32] <kirkland> smoser: doing so now
[20:34] <kirkland> smoser: fixed
[20:34] <kirkland> smoser: that teaches me to commit on your behalf without a proper merge proposal :-)
[20:35] <zul> smoser: why was 60 seconds choosen anyways?
[20:36] <smoser> zul, no good reason
[20:36] <smoser> we basically depend/expect there to be an eth0
[20:37] <smoser> we assume there is eth0 and it is dhcp-able
[20:37] <smoser> and within cloud environment it should be pretty consistent that such a thing would come up within 70 seconds of boot.
[20:38] <zul> smoser: cool...just wondering
[20:46] <koolhead17> kim0: around
[20:46] <koolhead17> hey smoser
[20:47] <smoser> hi
[20:50]  * RoAkSoAx off to lunch
[20:51] <eichi> how to make postfix start automatically after installing ?
[20:52] <guntbert> eichi: you have to edit the config and change some value to tell it to start - if I remember correctly (but i forgot the line, it is easily spotted though)
[20:59] <zim> Hi guys back again how long should it talk to format a 2TB raid 1 partition?
[20:59] <zim> s/talk/take
[20:59] <RoyK> zim: ext4?
[21:00] <zim> yup
[21:00] <RoyK> dunno - perhaps 15 minutes to an hour
[21:00] <zim> wow that long
[21:00] <zim> ok
[21:02] <zim> it get to 33% in half a second then just hangs/is doing it but nothing changes
[21:02] <ppetraki> zim, that might be a cue for you to get a coffee or something
[21:02] <ppetraki> zim, and check on it in 15 mins or so :-p
[21:03] <zim> kewl ok bugger just quit it after about 30min :)
[21:03] <zim> will let it run
[21:03] <ppetraki> zim, so is the array rebuilding while you're formating the new md disk?
[21:04] <zim> its a pity the progress bar is not informative
[21:15] <qman__> zim, that progress bar is only indicative of the number of tasks that it has completed out of the number it's doing
[21:15] <qman__> so if you have one really small partition and one huge one, it will instantly reach 33%, then take a very long time to do the huge one
[21:44] <zertyui> hello there
[21:45] <zertyui> iptable  is it easy ?
[21:45] <TheEvilPhoenix> zertyui:  what do you mean by "easy"
[21:47] <zertyui> this is what i mean  : http://www.thefreedictionary.com/easy TheEvilPhoenix
[21:47] <TheEvilPhoenix> zertyui:  short answer, no.  long answer, maybe.
[21:48] <RoyK> zim: just use ufw
[21:48] <TheEvilPhoenix> ^ that
[21:48] <RoyK> zertyui: that one was for you
[21:48] <TheEvilPhoenix> zertyui:  i'd recommend you use ufw rather than just iptables
[21:48] <TheEvilPhoenix> zertyui:  iptables is complicated in comparison to ufw
[21:48] <RoyK> ufw can do most iptables stuff
[21:48] <TheEvilPhoenix> indeed
[21:48] <zertyui> what is the difference between ufw and iptable  ?
[21:48] <TheEvilPhoenix> zertyui:  ufw isnt complicated :P
[21:48] <RoyK> ufw is an iptables wrapper
[21:49] <TheEvilPhoenix> zertyui:  ufw lets you define rules simply, but it will automatically fill in iptables' rules based on the uncomplicated rules you specify to ufw
[21:50] <zertyui> define rules simply, like what ?
[21:51] <RoyK> zertyui: man ufw
[21:53] <zertyui>        This program is for managing a Linux firewall and aims to provide an easy to use interface for the user.
[21:53] <zertyui> how easy it is ?
[21:54] <zertyui> is there any web based application for that ?
[21:55] <RoyK> zertyui: did you read the manpage or just the header?
[21:56] <zertyui> just the header
[21:56] <RoyK> bingo
[21:56] <RoyK> read on
[21:56] <zertyui> i don't have any idea about what can firewall can do
[21:56] <zertyui> give me some example kind of task
[21:57] <RoyK> see EXAMPLES in the manpage
[21:57] <zertyui> can we do with ufw
[21:57] <RoyK> does 'RTFM' sound familiar?
[21:59] <zertyui> no
[22:00] <zertyui> i know this one  : WTF
[22:04] <RoyK> zertyui: just read the manual and stop bothering us
[22:04] <zertyui> ok fine
[22:16] <jMCg> mmmmmmm....
[22:17] <jMCg> Wonder if I should RTFM too, or ask stupid questions first on ufw..
[22:29] <hallyn> zul: all right, please try with my newest lxcguest (just pushed to bzr) as well.  That gives you a console when you do 'virsh -c lxc:/// console n1', so you can actually snoop around
[22:29] <hallyn> after i did 'rm -rf /etc/init/cloud*' it came up fine for me with eth0 having netaddr :)
[22:29] <hallyn> i'm going to go look just a bit more at vga stuff, bbl.
[22:44] <mrevd> every couple of days apache OOMs on me. always at the same time. how can i determine what specific apache process is causing this?
[22:53] <zul> hallyn: cool will do
[22:54] <jMCg> mrevd: how many do you have running? Doesn't syslog say which process it killed?
[22:57] <mrevd> jmcg: i see 6 apache process running
[22:58] <jMCg> mrevd: but those aren't separate deamons, right?
[22:58] <mrevd> syslog states 'Kill process 2932 (apache2) score 180 or sacrifice'
[22:59] <mrevd> jmcg: how can i tell if they're separate deamons?
[22:59] <jMCg> mrevd: judging from the problem statement and the way you're asking, I'd say they're not.
[23:00] <mrevd> jmcg: sorry, this is new to me
[23:01] <jMCg> mrevd: what I find weird is the score. It seems low to me. How much memory do you have. What are you apache2 processes doing and how? Are they alone on that machine or do you also have other daemons (a DBMS, an MTA, etc..)
[23:03] <mrevd> 5 or 6 other apache processes were killed at the same time
[23:03] <mrevd> http://p.linode.com/5149
[23:04] <jMCg> mrevd: so you have what, ~500M memory?
[23:05] <mrevd> 512
[23:06] <jMCg> mrevd: it seems a bit too low for me, either that or you're doing something severely wrong. Might want to answer the rest of my questions too :)
[23:08] <mrevd> there are other daemons on the machine, apache is server a couple of small traffic sites
[23:09] <jMCg> What daemons? Should they be running on that machine, given that that one's going down regularly?
[23:09] <jMCg> Are the sites dynamic? How is httpd serving them?
[23:10] <zul> hallyn: the libvirt patch worked as well?
[23:11] <mrevd> the sites are dynamic. i'm not sure what you mean, how httpd serves them
[23:11] <jMCg> mrevd: mod_php, mod_wsgi, mod_perl, mod_cgi, etc..
[23:12] <mrevd> mod_php
[23:12] <mrevd> mostly, some cgi
[23:12] <markatto> snmpd got installed with the wrong permissions due to my umask. is this normal?
[23:12] <markatto> well, the config file for it
[23:13] <markatto> at least I assume it's due to my umask, as the permissions on it matched my umask
[23:20] <jMCg> mrevd: you could throttle MaxRequestsPerChild a bit and see if that gives you any head space.
[23:22] <mrevd> this just started happening, i was hoping to find the process that's responsible and clean/remove it
[23:25] <jMCg> mrevd: the process responsible is, most probably your apache2 process, because it gets bloated by some ugly wordpress site with mod_php
[23:25] <ZaclnxNewb> Hey, what's the best secure and easy email server for ubuntu headless 10.10?
[23:26] <jMCg> mrevd: here's a good summary of how it works: http://prefetch.net/blog/index.php/2009/09/30/how-the-linux-oom-killer-works/
[23:27] <ZaclnxNewb> I'm currently installing postfix
[23:29] <mrevd> jmcg: it's definitely apache2, but how can i narrow that down. i'm running 6 sites
[23:30] <jMCg> mrevd: I do that by running them with different daemons.
[23:31] <mrevd> jmcg: multiple apache daemons?
[23:31] <jMCg> mrevd: yes.
[23:31] <mrevd> jmcg: could you share how? i wouldn't know where to begin
[23:32] <jMCg> 00:39:20 <fajita> http://wiki.apache.org/httpd/ExtendingPrivilegeSeparation
[23:49] <jMCg> This here: http://thoughts.j-davis.com/2009/11/29/linux-oom-killer/ suggests, among other things, to: turn off overcommit, run FreeBSD or Solaris (and Oracle instead of PostgreSQL), do not turn off overcommit and have more memory.