[01:33] <CIA-4> debian-installer: cjwatson * r1384 ubuntu/ (5 files in 2 dirs): Move to 2.6.37-5 kernels.
[01:52] <CIA-4> debian-installer: cjwatson * r1385 ubuntu/debian/changelog: releasing version 20101020ubuntu3
[09:47] <sumitkv2> i installed ubuntu 9.10 alongside windows.....but m stuck with sh:grub>...
[10:06] <ev> sumitkv2: why 9.10?  Are you not able to use Ubuntu 10.10?
[10:08] <sumitkv2> i didnt had time to install 10.10!!but noe m stuck with this thing!!
[10:09] <sumitkv2> ev, ?
[10:10] <ev> sumitkv2: was this a full install whereby you resized windows, or did you install using Wubi?
[10:11] <sumitkv2> ev, i used wubi!
[10:11] <ev> sumitkv2: you're not stuck with it, then.  If you go into Add/Remove programs in Windows, you can select Ubuntu and remove it.
[10:12] <ev> I would suggest that you give Ubuntu 10.10 a try, as it's entirely possible the bug you're encountering has since been fixed.
[10:12] <sumitkv2> ev,  yes!but i want an ubuntu alongside windows using wubi!is there any way of reparing this?
[10:13] <sumitkv2> ev,ok!!i think i ll do that!
[10:13] <sumitkv2> thnx :)
[10:13] <ev> sure thing
[10:13] <ev> if it's still not working with 10.10, please file a bug at http://launchpad.net/wubi/+filebug
[10:18] <soren> ev: I've never used btrfs. Does it actually let me add another block device to the same btrfs filesystem?
[10:19] <soren> "btrfs device add" sure suggests that that is the case.
[10:19] <ev> indeed, I believe it is, but I haven't used it enough myself to be sure
[10:22] <ev> https://btrfs.wiki.kernel.org/index.php/Multiple_Device_Support
[11:49] <soren> ev: I can't help but wonder why it's suddenly kosher to do all this stuff in the filesystem layer, but when there was talk of putting ZFS in Linux, there was a lot of resistance for exactly this reason.
[11:50] <soren> I tend to agree with the resistance, actually. It seems really weird to have the filesystem support this, and not put it in a lower layer.
[11:50] <ev> I thought ZFS was shunned because it was licensed under the CDDL
[11:50] <soren> Well, that was the reason the /implementation/ was rejected.
[11:50] <ev> sure
[11:50] <soren> IIRC, Linus didn't like the /design/.
[11:52] <ev> I don't know if I agree.  Sure, there are performance implications (assuming you're talking about doing it in hardware), but you can innovate in the space a hell of a lot faster at the filesystem level.
[11:53] <soren> I just think that thinks like mapping multiple block devices as one belongs in the device-mapper layer.
[11:54] <soren> s/thinks/things/, obviously.
[11:54] <ev> ahh, I gotcha
[11:54] <soren> I can't seem to find the reference where Linus says something along those lines. Maybe I dreamed it. :(
[11:54] <ev> haha, do you often dream about Linus? :-P
[11:56] <soren> Didn't think so, but maybe I do :)
[11:57] <soren> Just found this: https://btrfs.wiki.kernel.org/index.php/Multiple_Device_Support
[11:58] <soren> It provides their rationale for not using dm, in case you're interested.
[11:58] <soren> Oh, well.
[12:06] <ev> ah, that makes sense
[12:19] <soren> I can't help but wonder why a checksumming dm-linear couldn't be whipped up, but meh.
[15:30] <shtylman> cjwatson: is dmraid not installed in the livecd installer anymore?
[15:34] <cjwatson> was it ever?
[16:00] <shtylman> cjwatson: I thought so
[16:00] <shtylman> but I could be wrong
[16:05] <shtylman> I was under the impression that it was installed if dmraid was detected
[17:05] <davmor2> ev, cjwatson:  http://davmor2.co.uk/ev_logs/debug I'm just grabbing the other logs
[17:06] <ev> ah, parted_server went away before m-a appeared
[17:06] <ev> can you please file a bug and attach those, then mark it as high and assign it to me
[17:06] <davmor2> ev will do
[17:06] <ev> thanks!
[17:07] <superm1> ev, so interestingly enough i discovered recently that our firmware guys have started using hudson too
[17:07] <davmor2> ev also the install has just stopped
[17:07] <superm1> at least for uEFI based firmwares.
[17:09] <superm1> seems to be quite powerful for everything that they've managed to scale out of it
[17:21] <davmor2> ev: No items matched "Evan Dandrea". you broked LP
[17:26] <davmor2> ev: https://bugs.launchpad.net/ubuntu/+bug/677100 I can't set the status and it won't let me assign it to you fsckin lp
[17:26] <ubot2> Launchpad bug 677100 in migration-assistant (Ubuntu) "Migration-assistant flags a warning even though it shouldn't of run (affects: 1) (heat: 6)" [Undecided,New]
[17:27] <davmor2> ev: any other logs you need
[17:46] <ev> sorry about that, had to help Magda with Spotify in Unity.
[17:46] <ev> davmor2: that should be enough, thanks!
[17:46] <ev> superm1: it is made of awesome
[17:47] <ev> I'm a bit blocked at the moment on our IS team, otherwise I'd be showing you reports of installer tests
[18:38] <superm1> ev, can you potentially share how you've got things setup with it while waiting on being able to see the reports?  i am thinking it might be worthwhile to set something up that does something quite similar but triggers off new ISOs rather with my ubiquity plugins layered in place
[18:40] <superm1> hm actually looking over closer, that probably wouldn't work for what i'd like to do
[18:41] <ev> this does effectively generate new ISOs as part of the process (but stops short of mkisofs as it uses NFS), and I'm more than happy to share the configuration with you
[18:43] <ev> superm1: some of the configuration is in here: lp:~ev/+junk/hudson-infrastructure/
[18:44] <ev> I'll work on extracting the hudson jobs into a branch as well
[18:44] <superm1> so does it actually get tested on real hardware then that's PXE booted with an NFS root mount?
[18:44] <ev> correct
[18:45] <ev> the copy of Ubuntu they're running has a post-resume job that tells them to reboot
[18:45] <superm1> ah
[18:45] <ev> when it suspends, it loads as much of shutdown and its dynamically linked libraries into memory as it can
[18:45] <ev> oh and uses reboot -f
[18:46] <superm1> using the same idea /etc/init.d/casper stop does i would guess to load as much as it can
[18:46] <ev> so when a job is ready, it forces a reboot by doing a wake on lan, which means it doesn't depend on the connection to the NFS server persisting (as it will likely have timed out)
[18:46] <ev> exactly
[18:46] <ev> the code is copied straight from there
[18:47] <superm1> is the install preseeded, or is there something that actually fakes out clicks in the UI?
[18:48] <ev> the latter
[18:48] <ev> sikuli
[18:49] <ev> but that bit is interchangeable
[18:49] <ev> it could be preseeded
[18:49] <superm1> ah i think i need to read through a bit of your prepare-cd to see a little better
[18:49] <ev> it could do a series of post-install probes of /target
[18:49] <ev> there's also an i386 chroot involved here (which you'll see in the jobs that I'm about to paste), as the hudson master is amd64
[18:57] <ev> superm1: http://paste.ubuntu.com/533953/ - rough setup of the hudson jobs
[18:57] <ev> I was using a cloned workspace to work around the problem that the netbooks don't have access to the internet
[18:57] <ev> but this deadlocks in hudson
[18:58] <ev> and we'll need access to the archive anyway if we want to test langpack grabbing
[18:58] <ev> and this is where I'm blocked on IS
[18:59] <ev> getting the hudson master to hand the slaves via its dhcp server a route to archive.ubuntu.com and bazaar.launchpad.net
[18:59] <ev> well, that and using pbuilder for the i386 chroot rather than what we've got now, which requires manual intervention for setting up new ubuntu releases
[18:59] <superm1> why not just mirror the archive on the hudson master?  Or at least cache the packages there
[19:00] <ev> I think there's IS voodoo here that effectively does that
[19:00] <superm1> ah
[19:00] <ev> squid and the whatnot
[19:00] <ev> the real trick here was figuring out how to get the machines to boot on command without tearing them open
[19:00] <ev> it was first assumed that they could just be suspended, woken up, and rebooted
[19:01] <superm1> scott ran into the same problems with the stuff he did for http://people.canonical.com/~scott/daily-installer/ didn't he?
[19:01] <ev> but as mentioned, nfs and tcp don't have yearlong timeouts, so thus the load everything into memory trick
[19:01] <ev> he had the advantage of rebooting into a full system
[19:01] <superm1> oh right
[19:02] <ev> also, I thought we could use file referencing to write the squashfs over top of itself, as existing slaves would use the old copy until they closed it
[19:03] <ev> this doesn't work across nfs, as it turns out
[19:03] <ev> but doesn't matter
[19:03] <ev> as we assume the nfs link to be dead once we suspend
[19:03] <ev> so it does rsync over top of the existing nfs root
[19:04] <superm1> have you considered pulling some of this directly into natty's casper and some others with just a different kernel command line option that casper could trigger to do a lot of what you do?  at least try to get away from having to break apart the squashfs
[19:05] <superm1> oh but you need to push the new ubiquity in, duh.
[19:05] <ev> there may still be value in putting some of this in there (incidentally, is casper surviving?  I could've sworn I heard from Scott that we were ditching it), but I just wanted to get it working first
[19:06] <ev> I did manage to discover that cifs support in casper is horrendously broken at the moment
[19:06] <ev> first because of cifs-utils not being there (fixed in natty)
[19:06] <ev> and second due to a bug in mount.cifs that drops existing mount options on a remount
[19:07] <superm1> casper has potential to not be surviving?  there's a lot that lives there that needs to find a new home then.
[19:07] <ev> indeed, though I could be wrong
[19:08] <ev> one goal with all of this was to be able to just drop a machine in, connect it to the network, add it's system uuid to the hudson config, power it on and have it just work
[19:08] <ev> just work> join the next build
[19:08] <superm1> well you could actually get away without needing to respin the squashfs though if you just install the new ubiquity packages in the early command (or if you include a 99hudson in the initramfs that does that exact thing)
[19:09] <ev> true
[19:09] <ev> I've added a note to do that
[19:09] <superm1> that's actually what i do for my primary plugin to allow it to try to upgrade itself if it sees a newer version added to the media than what's in the squashfs
[19:09] <superm1> you just have to be careful on low memory machines that you don't fill up the tmpfs aufs sets up for you
[19:10] <ev> indeed
[19:13] <superm1> was there a particular reason you chose to launch out of xdg/autostart rather than spawning hudson, sikuli and what not on an upstart job that started while ubiquity started in maybe mode?
[19:14] <superm1> it's not clear to me if the UI has to be ready immediately for things to work when hudson is spawned
[19:15] <superm1> if so, then if you'd be open to moving more of it into natty proper, then in ubiquity-dm it could make sense to just key off the same kernel commandline option while setting up the session to spawn hudson at that time too
[19:15] <sumitkv2> i installed ubuntu 10.10 inside windows Xp....but when i boot into it....i end up with a prompt... GRUB> ......what is wrong....??please help!!
[19:21] <ev> superm1: yes
[19:21] <ev> I think it was because it needed network manager applet to be running
[19:21] <ev> but no, that cannot be it
[19:22] <ev> as we're guaranteed a network connection by the very nature of the setup
[19:22] <ev> hm
[19:22] <superm1> some race condition or sorts though
[19:22] <ev> I'll have to look at my paper notes
[19:26] <superm1> well i've got a lot to play with here on this, thanks! if i can get this replicated with our special ISOs i'll see if I can help come up with some ideas how to improve it or where it would make most sense to pull pieces into natty
[19:28] <ev> that would be amazing
[19:28] <ev> thanks!
[19:28] <ev> heading out, have a good afternoon and evening
[19:29] <superm1> you too
[20:45] <apparle> guys I'm having problem installing kubuntu via wubi....
[20:46] <apparle> after setup, I am dropped at intrafms
[23:47] <MHJessen> I'm currently using eeebuntu 3 (Jaunty - Kernel 2.6.29-1-netbook) on an Asus eeepc 1000 and want to know if I can upgrade to Ubuntu Netbook 10.10 or do I need to do a complete reinstall?
[23:51] <cjwatson> this is speaking as a user rather than a developer, but for my wife's eeepc I found it easier to reinstall
[23:52] <cjwatson> two reasons: (1) the upgrade path from 9.04 to 10.10 goes via 9.10 and 10.04 and there was no eeebuntu for those releases anyway, not to mention that it would have taken ages (2) my wife's eeepc was seriously running out of disk space and didn't have enough temporary space to do the upgrade
[23:52] <cjwatson> it may be possible I suppose but I doubt it would be fun
[23:55] <MHJessen> Ah, OK. I thought that might be the case. The problem for me was that I have a limited situation as far as being able to tinker. I wondered if it was possible to do it without fouling up the 32 GB SDHC partition I'm using for /home. The root is on the 8GB SSD and I still have 47% free.
[23:56] <cjwatson> you can do an install preserving /home
[23:56] <cjwatson> use manual partitioning, mount the partition in question on /home, and make sure the format checkbox is cleare
[23:56] <cjwatson> *cleared
[23:57] <MHJessen> That's true but what happens to the packages that are in Jaunty? Do they get upgraded at the same time?
[23:58] <MHJessen> I guess I'm asking a lot because I'm looking for as painless a way to get to 10.10 as possible.