[07:23] <CIA-2> ubiquity: superm1 * r2941 ubiquity/debian/ (changelog rules): install mythbuntu's .desktop file as the appropriate name to make sure it gets copied into /home/ubuntu by casper on squashfs files created by livecd-rootfs
[11:04] <TheMuso> cjwatson: Just wondering what procedure you use to merge d-i bits with debian, consdiering debian uses svn and we use bzr. I tried creating a bzr import branch from debian svn for one of my merges, but bzr claims there is no shared root. Am I msising something?
[11:05] <cjwatson> TheMuso: I have Launchpad do imports for me
[11:05] <cjwatson> look on code.launchpad.net/whatever for the vcs-imports branch
[11:05] <TheMuso> cjwatson: ah ok.
[11:05] <TheMuso> That should be enough then, thanks.
[11:06] <cjwatson> it's usually just lp:whatever, in fact
[11:43] <CIA-2> oem-config: cjwatson * r559 oem-config/ (5 files in 2 dirs):
[11:43] <CIA-2> oem-config: Create a BaseFrontend class and move various things to it; it now counts
[11:43] <CIA-2> oem-config: as a relatively clear specification of what a frontend has to implement.
[11:54] <sakle> cjwatson: Has the format of the kickstart files changed significantly in intrepid? Ours is broken in multiple locations.
[11:55] <cjwatson> sakle: no, it hasn't
[11:55] <cjwatson> sakle: I'm happy to investigate a specific example of breakage
[11:55] <cjwatson> sakle: it will help if you can also produce /var/log/syslog from the installer
[11:56] <sakle> http://pastebin.com/d7f17db2c
[11:57] <sakle> for exampel
[11:57] <cjwatson> and the syslog?
[11:59] <sakle> http://pastebin.com/ma1fffae
[11:59] <sakle> It's not very informative
[12:00] <cjwatson> no, the whole syslog
[12:00] <cjwatson> not just a snippet
[12:00] <sakle> Sure, one second
[12:01] <cjwatson> what is --size=0 without --grow supposed to mean? that indicates a zero-sized partition not permitted to grow
[12:01] <sakle> Well, we need file based swaps
[12:02] <sakle> not partitions
[12:02] <cjwatson> not supported by d-i
[12:02] <sakle> right, so I create a appendix swap, and remove it with puppet
[12:02] <cjwatson> if you don't want a swap partition then don't list it in the kickstart file, and set up a swap file yourself later
[12:03] <sakle> Yes, but then you have to tell d-i you really don't want a swap partition
[12:03] <cjwatson> yes, you do.
[12:03] <cjwatson> that's not hard though, you can preseed that
[12:03] <cjwatson> anyway, I still need the full syslog from you, this is just speculation
[12:04] <sakle> coming, 10 more seconds
[12:04] <sakle> needed a pastebin with upload support
[12:07] <sakle> more than 150k
[12:08] <sakle> need another one
[12:08] <cjwatson> paste.ubuntu.com?
[12:09] <sakle> http://www8.picsearch.com/syslog
[12:10] <cjwatson> ok, that's a successful install apparently - is the failure that partitions come out wrong?
[12:11] <cjwatson> sakle: I note that that got HTTP 503 when trying to retrieve things from string.lan.picsearch.com
[12:11] <sakle> the failure is that the root user is not made, and the partitions are wrong
[12:12] <cjwatson> I can't help with the root user unless I see your full kickstart file; feel free to replace passwords in it
[12:12] <sakle> In fact, the partitions are not made at all, It needs manual intervention at that stage
[12:12] <sakle> 1 sec
[12:13] <cjwatson> hm, there does seem to be some kind of systemic parsing problem here, that's odd
[12:13] <cjwatson> the exact same file worked in hardy?
[12:13]  * soren is somewhat surprised by the all caps output from apt
[12:13] <soren> I thought we fixed that.. :/
[12:14] <cjwatson> because http://patches.ubuntu.com/by-release/atomic/ubuntu/k/kickseed/kickseed_0.49.patch was the total kickseed patch from hardy to intrepid
[12:14] <sakle> yes, the file worked in hardy
[12:14] <sakle> and gutsy
[12:14] <cjwatson> soren: to my knowledge we never figured out what was calling the relevant ioctl
[12:14] <soren> cjwatson: kees and I found it last summer.
[12:14] <cjwatson> where?
[12:15] <sakle> http://www8.picsearch.com/ksnp.cfg
[12:15] <soren> cjwatson: Good question. We fixed it, and I went on my merry way :)
[12:15] <soren> cjwatson: I don't remember, to be honest. I'm quite sure it was apt.
[12:15]  * soren goes to look
[12:16] <soren>   * apt-pkg/deb/dpkgpm.cc:
[12:16] <soren>     - merged patch from Kees Cook to fix anoying upper-case display
[12:16] <soren>       on amd64 in sbuild
[12:16] <soren> ...and the patch is still there :(
[12:16]  * soren goes to find a bigger hammer
[12:17] <StevenK> Screaming terminal on amd64 came back?
[12:17] <soren> StevenK: http://www8.picsearch.com/syslog
[12:17] <soren> StevenK: In some cases, apparantly.
[12:19] <cjwatson> sakle: looking
[12:19] <sakle> Thanks
[12:19] <soren> Hm... Very interesting. It switched to all-caps in the middle of the output.
[12:19] <soren> Ah, no. My bad.
[12:21] <cjwatson> hmm, odd, I can't reproduce the parsing problems with the test harness
[12:22] <sakle> It complains to the console about getopt problems-- The console is not saved, is it?
[12:22] <sakle> (when parsing)
[12:22] <cjwatson> whoa
[12:23] <cjwatson> yes, just noticed that
[12:23] <cjwatson> that's clearly the problem
[12:23] <cjwatson> debian/config/config.udeb:# CONFIG_GETOPT_LONG is not set
[12:23] <cjwatson> well, shit
[12:24] <cjwatson> I'm sorry we didn't catch this
[12:25] <sakle> I can fix this by modifying that flag in the initrd, then
[12:26] <cjwatson> only by rebuilding busybox and the installer
[12:26] <cjwatson> you can't modify that trivially
[12:26] <cjwatson> this is fixed in Debian, apparently
[12:26] <cjwatson> sakle: could you please file a bug on https://bugs.launchpad.net/ubuntu/+source/busybox/+filebug about this?
[12:27] <cjwatson> sakle: are you using a netboot installation? if so then I can build an update for you
[12:27] <sakle> Yes and no. Netboot, except for the kernel and the initrd, which are on a usb stick, to prevent pxe accidents
[12:28] <cjwatson> right, I mean the netboot image rather than how you boot it
[12:28] <sakle> Is this irc convo linkable from somewhere, as evidence that this is a real problem for the bug report?
[12:29] <cjwatson> don't worry about that, just describe the problem and I'll expand on it
[12:29] <cjwatson> (it is, irclogs.ubuntu.com)
[12:29] <sakle> Yes, please, that would be great (If you would build a update)
[12:30] <sakle> I've registered on the bug thing-- I'm waiting for the email.
[12:32] <cjwatson> update is building
[12:32] <sakle> By the way, cjwatson, can kickstart support disk selection now
[12:32] <cjwatson> though not officially as yet, just on my laptop
[12:32] <sakle> Right now, as you can see in the ksnp i remove sda and rmmod usbstorage
[12:33] <sakle> to ensure that it doesn't install to the usb stick itself
[12:33] <cjwatson> you've always been able to select a disk, but I think what you're asking is selecting it in a reliable way that does not rely on enumerated device names
[12:35] <sakle> I don't really care about which disk it installs to, just as long as it is not /dev/sda, which is the installation usb stick. As I understand it, you can use the uuid, but you won't know it when installing on a random machine
[12:35] <cjwatson> our kickstart implementation does not support that natively, but if you can come up with a shell fragment that figures out which device to use then it is possible to have that run to decide which device to use
[12:35] <cjwatson> is the USB stick reliably /dev/sda on your system?
[12:36] <cjwatson> because if that's reliable, why not just say clearpart --drives /dev/sdb?
[12:36] <cjwatson> disks do not really have UUIDs; I think you picked that up from some different bit of documentation
[12:36] <cjwatson> *filesystems* have UUIDs
[12:37] <sakle> The  shell fragment you mentioned, would that run in the %pre?
[12:38] <sakle> I suppose I can take anything that doesn't have the same disk label of the usb stick?
[12:39] <sakle> Not always /dev/sda no. But we have a large number of machines where the usb stick is reliably /dev/sda
[12:39] <cjwatson> technically %pre is before disks have been detected; in intrepid, you can put it in 'preseed partman/early_command string ...' and just make sure that the last thing it does is 'db_set partman-auto/disk "$disk"'
[12:39] <cjwatson> (replace "$disk" with whatever)
[12:40] <sakle> Ok, thanks.
[12:45] <sakle> https://bugs.launchpad.net/ubuntu/+source/busybox/+bug/293586
[12:46] <sakle> That's this bug
[12:46] <cjwatson> thanks
[13:09] <cjwatson> sakle: try http://people.ubuntu.com/~cjwatson/tmp/intrepid-busybox-fix/netboot/
[13:22] <sakle> Thank you
[14:05] <tjaalton> I'm having a hard time netbooting hardy, it refuses to install linux-generic since -image-generic/l-r-m-generic are "not going to be installed"
[14:08] <tjaalton> oh wait, maybe the netboot image is too old
[14:11] <tjaalton> cjwatson: is it just me or is the latest d-i stuck in hardy-proposed when it should be in -updates?
[14:56] <cjwatson> tjaalton: haven't put a whole lot of effort into hardy-proposed recently, but will do soon since 8.04.2 is coming up
[14:57] <tjaalton> cjwatson: the inability to install is what bothers me now, but I remember that something similar happened in June(ish).. it had something to do with the kernel versions in updates/security..
[15:03] <tjaalton> oh well, need to install something now.. intrepid then
[15:11] <kirkland> cjwatson: speaking of 8.04.2 ...  i backported the boot-degraded-raid bits, and uploaded to my ppa for testing
[15:11] <kirkland> cjwatson: i opened a new bug to track this, bug #290885
[15:12] <kirkland> cjwatson: i have ported the minimal patches back to hardy's mdadm, initramfs-tools, and grub
[15:12] <kirkland> cjwatson: i still need to do grub-installer
[15:23] <kirkland> cjwatson: how do you want to review those three packages for simultaneous upload to hardy-proposed?
[16:48] <CIA-2> ubiquity: evand * r2942 ubiquity/ (debian/changelog ubiquity/misc.py ubiquity/segmented_bar.py): Provide operating system titles in segmented_bar.
[20:10] <kirkland> TheMuso: evand: hey guys, I might need some help from you guys to review https://bugs.edge.launchpad.net/ubuntu/+source/mdadm/+bug/290885, if cjwatson is out for a while (or soon to be so)
[23:56] <CIA-2> hw-detect: TheMuso * r90 ubuntu/ (75 files in 3 dirs): merge from Debian 1.70
[23:57] <CIA-2> hw-detect: TheMuso * r91 ubuntu/debian/changelog: releasing version 1.70ubuntu1