[01:06] <twb> cjwatson: ping?
[01:06] <twb> So I am netbooting the 8.04 and 8.04.1 desktop i386 images, basically unmodified.  And while the former works, the latter completely explodes and dies -- gdm doesn't start, and the tty1-6 upstart events are disabled because they restart too frequently.
[01:06] <twb> I get "user not known to underlying authentication system" or similar -- presumably in reference to the "ubuntu" user that casper tries to create on boot.
[01:07] <cjwatson> twb: it's 1am and I'm about to go to bed - I don't think I should attempt to dive into this now
[01:07] <twb> OK, no worries.
[01:07] <cjwatson> twb: try to extract /var/log/casper.log if you can
[01:07] <twb> cjwatson: when do you next expect to be pester-able?
[01:08] <cjwatson> twb: may have to play about a bit to boot into something where you can get hold of that - maybe try booting with 'single' on the command line?
[01:08] <cjwatson> that should get you into an environment where you can check if the user actually exists, too
[01:08] <twb> Yeah, I'm about to seriously attack this.
[01:08] <cjwatson> twb: for various personal reasons I can't give you a reliable time at the moment
[01:08] <twb> No problem.
[01:09] <cjwatson> asynchronous methods are best - feel free to send me / ubuntu-installer@lists mail
[01:09] <twb> OK..
[01:33] <twb> OK, I just realized there are TWO casper.logs -- after unionfsization, there's a second one in /var/log
[01:33] <twb> It says "sudo: no passwd entry for ubuntu!"
[01:34] <twb> groupdel: group ubuntu does not exist
[01:34] <twb> adduser: `groupdel ubuntu' returned error code 6.  Exiting.
[02:34] <CIA-2> debian-installer: cjwatson * r1001 ubuntu/ (build/Makefile debian/changelog): Apply Ubuntu branding to version.info.
[02:55] <twb> Problem solved.  Some jackass had set umask 077 in /etc/profile on the server, so the netboot dir was being created readable and executable by root, but not by anyone else.
[06:12] <CarlFK> twb: does 8.10 netboot for you?
[06:20] <twb> I've no idea.
[06:20] <twb> Since these systems are deployed to secure facilities, we generally stick to LTS.
[08:50] <evand> davmor2: re USB> Yes, that would be a big help.
[08:50] <evand> I just need to confirm that Wubi from a usb-creator generated USB disk works
[08:51] <evand> need someone* (I don't have a Windows VM)
[08:51] <davmor2> No Probs I'll hit it in a bit then need to take the mother-in-law to the docs first :)
[08:52] <evand> sure thing.  Thanks!
[08:52] <twb> davmor2: to the docks?
[08:52] <twb> davmor2: are you gonna make her sleep with the fishies?
[09:02] <cjwatson> davmor2: pretty sure I fixed your grub installation problem of the other day, by the way, once I figured out how to trigger it myself
[09:02] <soren> win 4
[09:02] <soren> gah..
[10:35] <evand> ah, I take back what I said yesterday.  It was a broken version of partman-auto sitting in my local archive causing trouble.
[11:03] <CIA-2> ubiquity: evand * r2952 ubiquity/debian/ (changelog ubiquity.install-any): Install check-missing-firmware as part of hw-detect.
[12:12] <evand> hrm, jaunty grub is segfaulting
[12:21] <davmor2> cjwatson: evand: Right back :)
[12:28] <davmor2> evand: Are you wanting the usb testing on hardy.2 or intrepid?
[12:29] <evand> intrepid please
[12:35] <evand> hrm, it happens with intrepid's grub in a jaunty live CD as well
[12:36] <davmor2> cjwatson: Was the issue that grub was writing to boot but not to mbr?
[12:43] <davmor2> evand right firing up windows fingers crosseed
[12:43]  * evand crosses his fingers
[12:44] <davmor2> installing
[12:46] <davmor2> evand: bugger it's downloading the files
[12:46] <davmor2> but wubi and umenu work okay
[12:47] <evand> downloading the files?  It doesn't detect the USB disk as a source then?
[12:47] <davmor2> That's what I'm assuming yes
[12:48] <evand> if you start wubi and click on the desktop environment drop down list, is there more than just Ubuntu in there?
[12:48] <davmor2> I'll let it complete anyway
[12:48] <davmor2> hang on
[12:49] <davmor2> Ubuntu/Kubuntu/Xubuntu/Mythbuntu are listed
[12:50] <evand> yup, doesn't see it as a source
[12:50] <evand> err
[12:50] <evand> right, doesn't see the USB disk as a source
[12:51] <evand> I'll file a bug
[12:51] <evand> thanks for your help
[12:52] <davmor2> evand: Np's
[16:02] <cjwatson> davmor2: yeah, it was a side-effect of it misdetecting the system as using SATA RAID
[16:02] <cjwatson> davmor2: which was due to a change in dmraid's output format
[16:36] <CIA-2> console-setup: cjwatson * r84 ubuntu/debian/ (changelog console-setup.templates):
[16:36] <CIA-2> console-setup: Default console-setup/ask_detect to false, since it's hard to back up
[16:36] <CIA-2> console-setup: out of the detection widget once you've entered it.
[16:38] <cjwatson> Haegin: you're running into bug 18225. The simple workaround is to set mirror/country to "manual" rather than to "IE"
[16:40] <Haegin> cjwatson: ah, thanks, if i set it to manual will i have to select the country manually? i am aiming for a completely automated install
[16:40] <cjwatson> no, when preseeding "manual" effectively just means that you can set it using mirror/http/hostname and mirror/http/directory as you're trying to do
[16:40] <cjwatson> if you select "manual" in an interactive installation, that causes it to ask the hostname and directory questions - but preseeding them skips that
[16:45] <Haegin> awesome, that seems to have worked
[16:46] <Haegin> now i am getting the keyboard choice dialog coming up again but i blame that on me changing the netboot install thing i am using to the correct set of files for intrepid
[16:50] <CIA-2> console-setup: cjwatson * r85 ubuntu/debian/changelog: releasing version 1.28ubuntu2
[16:50] <cjwatson> Haegin: you're using Debian-style keyboard preseeding rather than Ubuntu-style (I guess you're reading the wrong documentation) and furthermore you need to set keyboard preseeding on the kernel command line rather than in a preseed file (unless you're using initrd preseeding)
[16:51] <CIA-2> usb-creator: evand * r66 usb-creator/usbcreator/gtk_frontend.py: Fix a grammatical typo in the GTK frontend.
[16:54] <Haegin> cjwatson: ah yes, got it now thanks. do you know how i can get rid of the encrypted home dir dialog and the auto update dialog?
[17:03] <cjwatson> https://help.ubuntu.com/8.10/installation-guide/i386/preseed-contents.html already documents the second (search for "Policy for applying updates"
[17:03] <cjwatson> )
[17:04] <cjwatson> I forgot to document the first in time for 8.10 (it's in the jaunty guide): 'd-i user-setup/encrypted-private boolean false'
[17:12] <Haegin> thanks again
[17:25] <CIA-2> usb-creator: evand * r67 usb-creator/debian/changelog: Correct grammatical typo in the GTK frontend (LP: #297569).
[17:29] <Haegin> just out of interest is it possible to automatically run an oem install from netboot?
[17:30] <cjwatson> well, the first stage yes, obviously not counting the end-user configuration step
[17:31] <cjwatson> oem-config/enable=true on the kernel command line, plus 'd-i oem-config-udeb/instructions seen true'
[17:33] <Haegin> thanks cjwatson
[17:37] <Haegin> cjwatson: sorry to keep asking questions but is it possible for the installer to turn the pc off after the end of the install rather than reboot?
[17:45] <Haegin> is the debian-installer/exit/poweroff boolean true option still valid in ubuntu?
[17:48] <cjwatson> Haegin: "still"? it's a very recent option and only exists in jaunty
[17:49] <cjwatson> added by me in response to another user's question in this channel
[17:49] <cjwatson> so not in <=8.10, I'm afraid
[17:50] <Haegin> cjwatson: ah ok, i figured it was a debian option. thanks
[17:51]  * Haegin ponders how stable jaunty is then decides that selling jaunty machines isn't a good idea
[18:04] <cjwatson> I wouldn't recommend it :)
[18:10]  * evand tries to get the upper hand on Haegin by selling Jaunty+1 machines.
[18:35] <evand> cjwatson: Isn't bug 92014 really a bug in libata?  As far as I can tell, there's really no way to discern a SCSI block device from a SATA one.
[18:36] <evand> Even HAL sees the storage.bus (somewhat rightly) as scsi.  I suspect this is something that need to be fixed further up the chain.
[18:37] <evand> Though I also suspect this has been discussed before, and am really just looking for a better understanding of the problem :)
[18:43] <cjwatson> evand: the bug is in the description used by partman-base; /dev/sd* realistically should no longer be described as exclusively SCSI
[18:45] <cjwatson> I don't think we can expect libata to change
[18:47] <evand> Any idea if the kernel could expose an extra bit in /sys/block to allow userspace to discern between SCSI and SATA?
[18:47] <cjwatson> sounds feasible but I don't know ...
[18:49] <evand> ok, thanks for explanation.  I'll ask in -kernel whether it's feasible and if someone with familiarity in that subsystem has the time and willingness to try to get it upstream.
[19:04] <cjwatson> I think it'd be useful to say, but if we can't, then it would be better to be less specific rather than being more specific but possibly wrong :-)
[19:04] <cjwatson> "SCSI/SATA" wouldn't be that bad I guess
[19:05] <cjwatson> if you do change it, send it to debian-boot@ so we can get translations updated in sync if possible
[19:05] <evand> noted, thanks