[03:51] ubiquity: evand * r2954 ubiquity/debian/changelog: releasing version 1.11.0 [03:51] superm1: it should :) [03:51] There was one remaining issue with grub segfaulting, but I cannot reproduce it on the latest i386 CD. I'm downloading the amd64 version now just to be sure it's not a arch specific issue. [13:15] evand I have here the python branch for wubi: https://code.edge.launchpad.net/~ubuntu-installer/wubi/jaunty.python [13:15] you might want to go through it quickly and have a look in the readme so we can discuss at UDS [13:17] it does not yet complete the installation and there are a few TBD around the place, but it's getting there, I'll need a couple of days to finish it off and a few more to get rid of the TBD. [13:18] fantastic. ill have a look later today [13:23] feel free to ask if you have any question [13:23] ok, will do [13:24] fyi I will rename that branch as trunk once I have it working, as mentioned some time ago' I was also planning to experiment with grub2 [20:56] Ugh. Just drove myself crazy for an hour trying to figure out why manual partitioning is broken in ubiquity, only to find out that I forgot to upload partman-base :/ [21:01] partman-base: evand * r119 ubuntu/debian/changelog: releasing version 128ubuntu3 [22:16] hey [22:17] question, im building a custom install initrd, would anyone be able to help me with it? [22:19] all i need to do it edit /etc/event.d/tty1 in the installed system before it reboots [23:06] dmarkey: doesn't need a customised initrd, just write a preseed/late_command script that changes /target/etc/event.d/tty1 as you want [23:07] and preseed that in the usual way (see the installation guide, help.ubuntu.com) [23:07] cjwatson: is this from the install cd? [23:07] if you like, yes [23:08] not an option, im booting the install initrd only onto xen [23:08] so all i have to work with is the initrd [23:08] I didn't say it only worked from the install cd; you asked ... [23:09] in practice most people use preseeding from something like netboot installs since it's usually applied to mass deployments [23:09] can you pass arbitrary kernel arguments? [23:10] i want this to be a generic initrd, but with the xen modules within, i have it that far, now all i need is to change /etc/event.d/tty1 when the install is finished. [23:10] cjwatson: i suppose i could [23:10] if you're customising the initrd anyway, you can put a preseed file in /preseed.cfg in the initrd [23:11] so just do that [23:12] but that will give me a fully unattended install? [23:13] if you write the preseed file appropriately, sure ... [23:13] you need to read the installation guide :) [23:14] appendix b [23:14] bah, [23:14] preseed files in and of themselves do not make the whole install unattended [23:14] you can preseed just one thing and that does not affect the rest of the installation in any way [23:16] perfect [23:16] e.g. [23:16] d-i preseed/late_command string sed -i 's/foo/bar/' /target/etc/event.d/tty1 [23:16] or [23:17] d-i preseed/late_command string chroot /target perl -pi -e 's/foo/bar/' /etc/event.d/tty1 [23:17] that sort of thing [23:18] so i just bang that into /preseed.cfg in the initrd [23:20] lemme try that [23:22] right [23:25] one last thing, can i get the insaller to dump an answerfile, like RHEL anacond.cfg [23:26] i'll just RTFM that [23:31] sort of, but unfortunately not in a way that you can easily just feed back in [23:31] it'll have a lot of junk that shouldn't be preseeded [23:32] you can use 'debconf-set-selections --installer' but strictly as a starting point [23:32] err, get not set [23:33] ok, that worked thanks, i'll put it up on my blog [23:33] http://dmarkey.com/wordpress/ ? [23:33] you should probably consider doing this, if you have any say on the ubuntu installer [23:33] yes [23:34] doing which? [23:34] oh, the answerfile thing? [23:34] no no [23:35] allowing the installer to run under xen, which is just 2 easy steps [23:35] oh, if we ever get an Ubuntu xen-enabled kernel that works for more than one release in succession, I'll consider it [23:36] i dont understand, im talking about in a xen guest [23:36] oh [23:36] what am I missing then? [23:36] so manual partitioning still seems to be broken, despite my changes. I'm going to take a further look at the airport / on the plane as I have to pack. [23:37] all you need to do is include the xen modules in the initrd and change tty1 to hvc0 in /etc/event.d/tty1 [23:37] I thought we already arranged for an /etc/event.d/hvc0 if that was console= [23:38] oh really? [23:38] (it shouldn't be /etc/event.d/tty1, that should be, well, tty1) [23:38] oh, only as of jaunty [23:38] [ Ian Campbell ] [23:38] * Add a getty to hvc0 if Xen is detected. [23:38] finish-install 2.19 [23:39] cool, so just include the kernel modules and we'd be set [23:39] xen modules in initrd> isn't that from the xen-enabled kernel though? [23:39] i dont think theres a xen enabled initrd? [23:40] and its a unified kernel now [23:40] oh, it is? which modules then? [23:40] sorry, bear with me, sleeping baby in arms => typing hard [23:41] http://pastebin.com/m14b2760d [23:41] those do not appear to be in 2.6.27-9-generic [23:41] basically find /lib/modules/`uname -r` -name "*xen*" [23:42] well apart from netxen_nic [23:42] I can only easily include the ones built from our -generic kernel build [23:42] they are from the generic tree [23:42] $ find -name \*xen\* [23:42] ./2.6.27-9-generic/kernel/drivers/net/netxen [23:42] ./2.6.27-9-generic/kernel/drivers/net/netxen/netxen_nic.ko [23:42] ./2.6.27-7-generic/kernel/drivers/net/netxen [23:42] ./2.6.27-7-generic/kernel/drivers/net/netxen/netxen_nic.ko [23:43] oh, hmm [23:43] where di i get those then [23:43] 2.6.27-7-generic/kernel/drivers/block/xen-blkfront.k [23:43] ?? [23:43] I don't know where you got that - it isn't in the stock package [23:44] including netxen_nic is a matter of the kernel guys listing it in debian/d-i/modules/nic-modules (file a bug on linux for that one) [23:44] umm.. they wouldnt insert correctly if they wer for another kernel [23:45] oh, I'm on i386 here [23:45] ah ok [23:45] it appears that stuff is built on amd64 but for i386 it's only built for -server [23:45] oh, hmm [23:45] which seems a trifle odd [23:46] could you ask the kernel team about that? I'm not really keen on switching the installer initrd over to -server :-) [23:46] well, in fairness most xen setups now are amd64 [23:47] ah, well in that case ... [23:47] which of those modules are actually needed in d-i? [23:47] xen-blkfront I assume [23:47] is the i386 pae by default? [23:47] no [23:48] xen-fbfront sort of looks like it would need some special userspace support to load it? [23:49] umm.. dunno, its the xen frame buffer, if you enable vfb in the config file for that domain, you can vnc into it [23:49] worked seamlessly when i tried it [23:49] just wondering what loads the module [23:50] oh, just came across a bug [23:50] anyway, basically all of this sounds totally sensible to include by default, but I'll need to get the kernel guys to actually add the modules; perhaps you could send a mail to ubuntu-installer@lists.ubuntu.com about it and I'll deal with it when I get back from holiday [23:51] when i change tty1 then i dont get a login on the virtual frame buffer, makes sense [23:51] right, it's probably better to have both tty1 and hvc0 [23:51] so copy that file to hvc0 and then change it [23:52] can i have more than 1 preseed/late_command string [23:52] no, but it's passed to the shell for evaluation, so you can separate statements with semicolons [23:53] ah ok [23:55] http://pastebin.com/m2cc36efa [23:56] ok [23:56] maybe udev is doing it or something [23:58] d-i preseed/late_command string chroot /target cp /etc/event.d/tty1 /etc/event.d/hvc0 ; chroot /target sed -i 's/tty1/hvc0/' /etc/event.d/hvc0 [23:58] does that look alright?