[00:00] cjwatson: I think best at this point is I setup an env., to build the installer myself so I can investigate [00:01] cjwatson: oh and the live option doesn't boot all the way to X, but I'll look at that later [00:01] cjwatson: could be X busted [00:02] sounds like a kernelspace bug to me, TBH [00:02] could well be [00:02] either that or something wrong with setfont [00:02] I can't see how the installer would even have any control over that [00:02] setfont isn't called on VT switch [00:02] so still a kernel bug :) [00:02] yeah [00:02] possibly [00:02] mjg59 adjusted font restoration on VT switch handling in the Ubuntu kernel recently [00:02] could be specific to fbcon [00:02] that could be busted [00:02] ok [00:03] I can try using a different kernel see if it makes a difference [00:03] I've noticed something similar in kvm, FWIW [00:03] ok [00:03] but I had put it down to kvm damage of some kind [00:03] I haven't reproduced with real VGA text mode [00:03] but I haven't tried hard on the x86 [00:03] in the kvm case, I don't see brokenness on vt1 [00:03] only on the non-newt vts [00:03] kvm uses fbcon too ? [00:03] yeah, that's what I see on the G5 too [00:04] but I wonder if the ps3 is just showing a variant of that problem that happens to affect the newt vt [00:04] d-i uses vga16fb+fbcon [00:05] I'd been meaning to try out http://paste.ubuntu.com/6540/ at some point, having noticed that mjg59's patch leaks memory [00:06] could conceivably have something to do with it [00:35] I'll have a look [00:35] the console code is ugly as hell but heh [00:39] thanks. nothing new there :) [00:46] cjwatson: ok [00:46] so [00:46] http://www.codon.org.uk/~mjg59/tmp/fix_fonts.diff [00:46] it's a bug [00:46] the ifdef VGA CON makes the font save/restore code be always called when vgacon doesn't exist [00:47] instead of the other way around [00:47] which breaks fbcon [00:47] matthew mailed the kernel team, it should be fixed soon hopefully [00:52] ah! thanks [00:52] I'll chase that and make sure it gets in [00:59] ok [08:33] hi all [08:33] iam try to add more than one Repos into the preseed file, can i do this or is better to create a post install script for this? [11:07] if you plan to fix some of the remaining i18n bugs (https://lists.ubuntu.com/archives/ubuntu-installer/2008-April/000180.html), remember that Thursday is theoretically NonLanguagePackTranslationDeadline. of course, in the installer's case it just depends on a) when you do the last translation template upload to Rosetta and b) when you do the last translations download from Rosetta, so it's adjustable if you want I guess [11:11] Silicium: hi [11:11] Silicium: you can [11:12] d-i apt-setup/local0/repository string ftp://your.local.repo [11:12] and [11:12] err, no [11:12] d-i apt-setup/local0/comment string This is my repo [11:13] no ? [11:13] the value of apt-setup/local0/repository needs to be a valid sources.list line, minus the leading "deb " [11:13] oh yes [11:13] so you probably need ftp://your.local.repo/ubuntu hardy main or similar [11:13] sorry I didn't paste correctly [11:13] (but otherwise yes, though I didn't see the question ...) [11:13] my local repo only ends with "/" [11:14] grrrreg: tjamls [11:15] thanks [11:15] damn qwerty [11:41] Mirv: thanks. I'm all too aware of the deadline and had been largely avoiding those as a result, but I might hoover up some small things. [11:41] Mirv: I've had a fair few complaints about the automatic language prompt in the boot menu, I think mostly because it looks like an accident. I think maybe that adding "Select your language" or something at the top might help [11:42] I would definitely have to notify translators and documenters about that [11:43] cjwatson: yes, I think so Select your language could help those people who are puzzled about it (though it might be mainly people who have used earlier versions of Ubuntu installers, not new people) [11:43] but usually a selection has some kind of title so it makes sense [11:43] it's a little difficult to squeeze everything in, and of course I don't actually have a widget set at that point, it all has to be drawn by hand [11:44] I could just put the menu title at the top to avoid requiring translation changes [11:44] so the Language and Keymap menus would change [11:45] maybe only add the menu title if the menu is bigger than one column [11:45] since otherwise it's rendered fairly clearly as a pop-up from the menu bar [11:50] the menu title at the top would work too, quite nicely even. and also "Select your language" in English wouldn't be too bad since most people use the language selection only once and it's English anyway then before another language is selected... but both solutions are probably fine for people [11:51] I'll try figuring that out today, but I need some more REALLY STRONG COFFEE before attacking that bit of gfxboot-theme-ubuntu [11:53] how i can create a script running after installing ubuntu (on the fresh system) [11:53] ? [11:53] cjwatson: good luck :) [11:56] Silicium: you can download your script with wget from another server, during preseed/late_command [12:01] grrrreg: yea, i try [12:01] but does not work [12:02] ah [12:02] i used base-config/late-command [12:02] don't forget to chmod 755 your script before running it ;) [12:03] i already done on the server :) [12:03] and if you want to run your script from inside the target, I have done this way: [12:04] of course i want run inside [12:04] i only done a apt-get install [12:04] d-i preseed/late_command string wget blah; . /lib/chroot-setup.sh; chmod 755 /target/tmp/postinstall.sh; chroot_setup; chroot /target /tmp/postinstall.sh; chroot_cleanup [12:04] thanks [12:04] /lib/chroot-setup.sh is very useful [12:05] and is actually used by apt-setup [12:05] ok :) [12:06] depending on the complexity of the script, you may not need chroot_setup/chroot_cleanup [12:06] but it's useful if you might care about e.g. /proc being mounted [12:08] blah; . /lib/chr [12:08] should be lah; ./lib/chr ? [12:09] no. [12:09] '.' is a command [12:09] mhm ok :D [12:10] (it means "read and execute commands from this file, without starting a subprocess" [12:10] ) [12:13] ok thanks [12:22] cjwatson: should the d-i late_command show anythin on the installer? [12:22] so i cant see any reaction [12:36] Silicium: not unless you explicitly add user interaction (which is a bit non-trivial) [12:36] Mirv: how does http://people.ubuntu.com/~cjwatson/tmp/gfxboot-language.png look to you? [12:37] I might also arrange for the F4 text not to be displayed until after you exit the language menu [12:37] since it looks a bit funny there [12:50] cjwatson: no problem [12:50] cjwatson: but the script does not start [12:50] check syslog for errors [12:52] cjwatson: looks very good. yes the F4 text looks a bit funny though not that big of a problem. [12:54] cjwatson: no entrie in the logfile [12:54] is this before or after install grub? [12:57] after [13:00] and after user-setup? [13:01] yes [13:01] and before what? :) [13:02] it would be easier if you just posted the syslog rather than going back and forth [13:02] :D [13:04] http://marco.2am.ch/syslog [13:08] Silicium: search for /usr/lib/finish-install.d/07preseed [13:08] Silicium: you need to change your wget to output to /target/tmp/postinstall.sh rather than /tmp/postinstall.sh [13:08] Apr 7 11:53:34 log-output: chmod: /target/tmp/postinstall.sh: No such file or directory [13:08] Apr 7 11:53:34 log-output: chroot: cannot execute /tmp/postinstall.sh: No such file or directory [13:18] uuh [13:18] sorry [13:18] that i could find self :/ [13:54] cjwatson: the package postfix want a user interaction to select wun type "internet site" [13:54] so can i automate this? [13:54] is not in the preseed i think [13:54] its runs from postinstall [13:58] Silicium: there's two preseed lines to add to pre-configure postfix's installation [13:58] postfix postfix/main_mailer_type select No configuration [13:58] ("No Configuration" is an example) [13:58] and: [13:58] postfix postfix/mailname string yourhost [13:59] hmm can i use this if i do apt-get install in a script, outside the installer? [13:59] mhm or i install postfix with the installer [13:59] maybe [14:00] I have a script that executes debconf-set-selection before the aptitude install [14:00] oh [14:00] can i see? [14:01] well, it's a cfengine module, but it's very simple [14:01] yea great [14:01] put your debconf conf in a file [14:01] can i see? :) [14:01] then download it [14:01] then run a shellcommand action like this: [14:02] /bin/cat /var/cfengine/inputs/debco [14:02] nf/* | /usr/bin/debconf-set-selections [14:02] oops [14:02] /bin/cat /var/cfengine/inputs/debconf/* | /usr/bin/debconf-set-selections [14:02] see what I mean ? [14:02] yep [14:02] can i see you debconf file? [14:02] or is this a d-i like file? [14:02] it is [14:03] ok great [14:03] in will try :9 [14:03] it's exactly the 2 lines I put before, beginning by "postfix" [14:09] echo "d-i postfix postfix/main_mailer_type select internet site" >> /tmp/postfixconf [14:09] echo "d-i postfix postfix/mailname string localhost.localdomain.int" >> /tmp/postfixconf [14:09] like this? [14:13] yup [14:13] hm [14:13] no [14:13] without "d-i" [14:13] ok [14:14] why not just put the configuration in your preseed file? [14:14] there's no need to mess about with calling debconf-set-selections externally [14:14] does this works? [14:14] yes [14:14] ok great [14:14] I wouldn't suggest it if it didn't [14:14] ok great [14:15] doesn't matter how you install postfix - any entry in the preseed file whose owner is not "d-i" automatically gets preseeded in the installed system's debconf database [14:15] ah [14:16] fist is the owner? [14:16] (the owner being the first field, "postfix" above) [14:16] d-i is debian installer [14:16] yes [14:16] cool :D [14:16] the format is "owner question_name type value" [14:16] grat [14:16] other question [14:16] how i can get the "question_name"? [14:16] if i dont have [14:18] most of the questions you care about will be documented in the installation guide [14:18] for other packages, if it isn't documented, try /var/lib/dpkg/info/.templates [14:18] ok thx [14:18] as a general rule, only preseed the ones you need, don't go preseeding everything just because it's there :) [14:18] yep [14:19] i only need a base system mostly unattanded (only ip adress and diskpart) and automatically install a self programmed application [14:32] ben suffisait de l'acheter ya jours [14:32] arg [14:32] oops. [14:35] Mirv: are you sure about bug 203909? debian/po/fi.po in current ubiquity has those strings marked as translatable - they just have empty msgstrs. [14:35] Launchpad bug 203909 in ubiquity "Network Proxy settings in Advanced dialog lack translations/i18n" [Medium,Confirmed] https://launchpad.net/bugs/203909 [14:36] Mirv: as a general rule, please don't file bugs when ubiquity simply isn't up to date with Launchpad translations; it's very likely that we'll miss those bugs when we do update ubiquity, and then the bugs end up lying around as cruft [14:37] hmm [14:38] all config files got the name *.dpkg-new after automated install [14:38] strange [14:41] Silicium: sounds like some packages didn't get configured [14:41] check syslog (again) [14:42] yes is it [14:42] strange [14:43] I don't see anything problematic in that regard in your previous syslog [14:43] yep [14:44] i will check that all [14:47] cjwatson: isn't it installer team's job to upload a new template to rosetta, still? the current template in rosetta does not have the proxy settings translatable, which is why I filed the bug. [14:48] Mirv: sure it is, but a bug report doesn't help [14:48] we do it semi-automatically every so often [14:49] actually, technically it's something Carlos has been doing [14:50] and if there's a problem with that process, there isn't really a good place for the bug report - better to talk to us directly [14:52] ok, well, it's hard to know for most people if something's a i18n bug in the ubiquity code or if the translation template in Rosetta is not up-to-date (since it's being done manually, unlike with other packages where each upload updates the templates too) [14:53] I understand for most people, but you know more :-) [14:53] I've asked on #launchpad for an urgent update [14:53] since it does seem to be way out of date :-( [14:56] ok, I will keep in mind to check the current bzr templates etc. before filing the bugs. also, I do keep my bugs systematically up-to-date enough so they don't get left as cruft [14:56] fair enough [14:56] carlos is going to update the template in LP [14:56] it's actually surprisingly easy to forget that most ubuntu-specific packages have the bzr repository available for checking... [14:57] I think, actually, that you can't really be expected to do that; I had incorrectly assumed that the LP template would be up to date and had been surprised that you hadn't checked that === ebel_ is now known as ebel [15:57] aah [15:57] :9 [15:58] cjwatson: all works now :) [15:58] good stuff [15:58] yea [15:58] what was the problem? [15:59] i dont know, i added the debconf settings into the preseed file and then works [15:59] and removed the echo >> [16:02] the software selection is named "tasksel" right? [16:07] yes, though why do you need to know? [16:07] so i need to skip that [16:10] surely you need to know the individual questions to preseed, rather than what the software is called [16:10] mhm? [16:10] i cant only set to skip? [16:12] could you elaborate? [16:13] https://help.ubuntu.com/community/Tasksel?action=AttachFile&do=get&target=tasksel.jpg [16:13] i want to skip this dialog [16:13] :D [16:13] withot any selection [16:13] without [16:13] you can't just "skip" installer components in general, but you can preseed any questions they ask [16:13] and how i do that? [16:13] ah [16:13] search for tasksel/first in the preseeding appendix to the installation guide [16:13] i cant [16:14] ok [16:15] https://help.ubuntu.com/7.04/installation-guide/i386/appendix-preseed.html (7.10 guide isn't on the web yet, but is in the installation-guide-i386 package in 7.10 itself) [16:16] any i need for 8.04 [16:16] but its the same i think [16:16] it's the same [16:16] you can look at the installation-guide-i386 package in hardy too [16:17] where is located the guide aftern install? [16:18] it's not installed by default, but you can install the installation-guide-i386 package [16:18] i have it [16:18] then it is in /usr/share/doc/installation-guide-i386/ [16:18] thx [16:28] ok fixed [16:28] and the "download language support" only can be skipped with boot options, right? [16:28] pkgsel/language-pack-patterns= pkgsel/install-language-support=false [16:34] you can use the equivalent preseed file entries too ('d-i pkgsel/language-pack-patterns string', 'd-i pkgsel/install-language-support boolean false') [16:34] ok [16:35] thanks for your help mate [16:35] so, i know i can use other ressources but this take a lot of time, and its not normal to get help like this :) [17:01] ubiquity: cjwatson * r2603 ubiquity/ (6 files in 5 dirs): [17:01] ubiquity: * Add a "Format the partition?" checkbox to the partition edit dialog [17:01] ubiquity: (LP: #184838). [17:12] cjwatson, i didn't see it merged or any comments from you, so i just wanted to remind you I updated that branch for partman-auto's EDD improvement at http://bazaar.launchpad.net/~superm1/partman-auto/edd-improvements . I tested it on a two drive system this time around and made sure that it was still picking the right drive. [17:12] oh, right, thanks for the reminder [17:18] once that is merged, would you or evan be able to do a ubiquity release and DVD generation sometime before RC to make sure there is still time before the big rush at RC time to catch last pending issues? [17:19] we definitely have to do that anyway [17:19] okay great [17:20] ok, that partman-auto branch looks fine [17:21] merging [17:22] Is FinalFreeze tomorrow or Thursday? I was under the impression it was Tuesday, but the MOTU announcement of it on ubuntu-devel@l.u.c has me confused. [17:22] partman-auto: cjwatson * r263 ubuntu/ (2 files in 2 dirs): merge Mario's edd-improvements branch [17:23] evand: Thu, afaik [17:23] cjwatson: thanks for clairfying [17:23] still, no time to lose [17:23] indeed [17:23] whenever these freezes happen there is always the confusion of which time zone they are referring to as well [17:23] and what time on that "day" [17:24] (Just waiting on the new m-a to hit the archive before I start a new ubiquity release) [17:25] indeed [17:25] mario_limonciell: it's always been kind of deliberate, to try to discourage people from pushing deadlines too tightly [17:27] partman-auto: cjwatson * r264 ubuntu/debian/changelog: releasing version 73ubuntu7 [17:33] evand: I want to get my fix for the usplash fiasco in before the next release if possible; in progress on that [17:33] (you have to copy .Xauthority into /target somewhere, set XAUTHORITY to point to it, and bind-mount /tmp/.X11-unix. Ugh.) [17:34] cjwatson: ack'ed [17:34] heh [17:35] cjwatson, you can't just xhost + local:"" for the time being? [17:35] and bind mount .X11-unix [17:36] I had to do a similarish hack on when we launch setup inside the chroot for mythbuntu: http://bazaar.launchpad.net/~ubuntu-installer/ubiquity/trunk/annotate/cjwatson%40canonical.com-20080407160143-xh1hutjtdhainnod?file_id=mythbuntusetup-20070702142439-ttoc7269gdy3xks7-5 [17:36] I suppose I could but I have religious objections to that [17:36] and seeing as it isn't that hard to do it properly :) [17:37] right [17:37] it was more difficult to do it properly on the mythbuntu stuff because the new hostname had to take effect immediately [17:38] otherwise the launched app would fail to store settings properly [20:39] bdmurry, when you have a few mins, can we resume on bug 204133? [20:39] Launchpad bug 204133 in wubi "wubi install unusable - Buffer I/O error on device loop0" [High,Confirmed] https://launchpad.net/bugs/204133 [20:40] cjwatson, I suspect I will need some help on that... [20:41] xivulon: already took it to the kernel team, see my PM [20:41] oh, I'm not idenitified [20:41] odd [20:46] ah great [20:47] bdmurray, as mentioned last time it would be useful if you could boot with init=/bin/sh and run a few tests [20:47] evand re segfaults [20:47] xivulon: Is serial access necessary for that? [20:47] I will try to make a build with (nsis) debugging symbols [20:48] bdmurray no [20:48] when you boot, press esc and use "e" to edit the kernel line in menu.lst [20:48] append init=/bin/sh [20:49] that should bring you to a console after run-init (~chroot) [20:49] assuming you can get that far... [20:50] This is still a 2.6.24.12 kernel [20:51] not sure if it is worth reintstalling with a recent build, just in case the issue was addressed [20:51] What did the other commenters say regarding media? [20:53] Other than the ones that commented on the bug a few others commented on the forum, but do not know much other than the last 1 or 2 error lines: either buffer I/O or "respawning to fast" (which I assume is related) [20:53] okay, I'm at the init prompt fwiw [20:54] can you check the r/w status of /host and / [20:56] /host is 777 and . is 755 - is that what you are looking for? [20:57] if you can try to write a few files [20:57] touch /host.trash [20:57] touch /host/trash [20:59] I used 'dd if=dev/urandom of=/host/trash bs=512 count=20' w/o any problem [21:00] and on / as well? [21:01] next is to run ntfs-3g.probe on the /host device [21:04] and if that works I'd try to run manually the scripts in rcS.d [21:04] hmm, I'm not having any luck writing to / [21:08] hmm what do you have in /proc/mounts [21:09] /dev/loop0 is mounted at / and is whoing ro [21:10] showing that is [21:10] can you pls reboot with break=init? [21:13] done [21:19] bdmurray, lost connection, not sure if you posted any msg after my last reply [21:19] I've rebooted with break=init [21:19] can you do the same exercise with /root ? [21:20] should also be the mountpoint of /dev/loop0 [21:22] there is no /dev or /proc mounted now but I was able to touch and edit a file on / [21:22] the relevant mountpoint in this case is /root [21:22] which will then be chrooted to / [21:23] ah, with /root I see now [21:24] The filesystem is read-only [21:24] umount /root [21:24] losetup -d /dev/loop0 [21:25] I can't unmount it because the device is busy [21:25] ah sorry [21:25] mount -o move /root/host /host [21:26] then umount and remove loop device [21:26] brb [21:31] xivulon: there is more than just host mounted under /root do they are need to be moved? [21:36] bdmurray let me see what other breakpoint we can use, one sec [21:39] break=bottom should do [21:44] rebooting again ;) [21:46] thx for that [21:47] its weird going through 3 boot loaders. ;) [21:49] xivulon: you are looking for the results of losetup correct? [21:50] losetup -d is to undo the loop device, [21:51] I'd like to redo the operations one by one to see if there is any relevant message (syslog) [21:51] okay, I've done mount -o move and umount /root so far [21:51] losetup -d /dev/loop0 [21:51] losetup -d /dev/loop0 reports "ioctl: LOOP_CLR_FD: No such device or address" [21:53] losetup /dev/loop0 /host/ubuntu/disks/root.disk [21:54] mount -tv ext3 -o rw /dev/loop0 /root [21:57] it has mounted fine [21:57] and I can write to it now [22:00] hmm [22:01] * xivulon wonders if there is any reason why roflag might have been set -r [22:02] shouldn't make any difference but we might try to do in one go [22:02] umount /root [22:02] losetup -d /dev/loop0 [22:03] mount -w -o loop -t ext3 /host/ubuntu/disks/root.disk /root [22:06] What are you looking now? [22:06] if /root is rw [22:06] the above line is the same as used in initramfs [22:07] yes, it is rw [22:09] did you test for rw after break=bottom [22:10] hmm, I'll test again just to be sure [22:10] ubiquity: evand * r2604 ubiquity/ (debian/changelog partman/check.d/03partition_too_small): [22:10] ubiquity: * Don't miscalculate with nested partitions in 03partition_too_small (LP: [22:10] ubiquity: #198039). [22:12] after rebooting with break=bottom /root is ro [22:12] hmm the only thing I can think of is that roflag is set to -r but I do not see how [22:13] can you boot with: DEBUG=1 break=bottom [22:14] ehm no [22:14] debug break=bottom [22:15] and look for roflag in /tmp/initramfs.debug (when /host/ubuntu/disks/root.disk gets mounted) [22:15] looking at dmesg it says recovery required ... recovery complete [22:16] hmm can it be that if we have a sleep=nsecs before mounting the loop device it will work? [22:16] because some fs operation does not have time to complete? [22:17] you may try to boot with break=mount [22:17] wait a few secs and type exit [22:17] sounds good [22:20] no change [22:21] Then I'd boot with: debug break=bottom [22:21] and look into syslog and dmesg [22:21] and /tmp/initramfs.debug [22:24]