[12:15] <xnox> cjwatson: Tajik translation was initially started in ubuntu, but then re-directed to debian. It's now at 100% in debian. What needs to happen to enable Tajik translation by default in d-i & ubiquity on ubuntu side? (apart from well merging/syncing all the udeb packages)
[12:15] <cjwatson> Just that.
[12:16] <xnox> awesome =)
[12:16] <cjwatson> Well, there might be a few tweaks in things like gfxboot-theme-ubuntu, but it's basically all about merging everything up
[15:38] <jbicha> xnox: please disable the U1 plugin for Ubuntu GNOME also
[17:30] <stgraber> cjwatson: around?
[17:30] <cjwatson> Just going for dinner now
[17:30] <cjwatson> Back in a bit
[17:31] <stgraber> ok, I'll dump my brain in here for now and then go with a workaround :)
[17:31] <stgraber> I'm reinstalling a 12.04 box that's booting through UEFI. I started d-i by having the existing grub2-efi boot into the kernel+initrd of the netboot image
[17:31] <stgraber> dmesg confirms the firmware is EFI, the partitioning step found the EFI boot partition and marked it as such
[17:32] <stgraber> but at the end of the install, d-i tried to install the BIOS variant of grub and failing quite badly (not too surprisingly)
[17:32] <stgraber> checking /target, none of the usual efi packages have been installed
[17:33] <stgraber> so it looks like something misdetected that machine at some point during the install
[17:35] <stgraber> I'll now try to install from a USB stick, see if that works any better... That box is my home router so I can't really spend much time debugging d-i on it
[17:36] <stgraber> syslog: http://paste.ubuntu.com/6055948/
[18:54] <stgraber> cjwatson: solved it, apparently the 3.2 kernel uses pstore for kernel panics and managed to use all of the available nvram space
[18:54] <stgraber> cjwatson: as a result no more settings could be changed and the kernel got quite confused about it
[18:55] <stgraber> cjwatson: I now managed to manually remove all the dump-* variables from nvram and I'm suspecting d-i will be much happier
[19:01] <antarus> stgraber: mmm so you are working on EFI based installs then?
[19:01] <antarus> (netboots?)
[19:01] <stgraber> antarus: not working, just using. Most of my systems are UEFI.
[19:02] <stgraber> *not working on
[19:03] <antarus> we have repeatedly kicked the can on that problem, managed to avoid it so far
[19:03] <antarus> ;p
[19:03] <antarus> also why am I on IRC on a holiday
[19:18] <stgraber> cjwatson: hmm, spoke too soon, d-i still installed grub-pc instead of grub-efi...
[19:21] <stgraber> cjwatson: I worked around it by telling d-i not to install grub2, then manually installed grub-efi
[19:38] <cjwatson> The EFI test in d-i is simply whether /sys/firmware/efi exists
[19:39] <cjwatson> So presumably the kernel has got confused by something and hasn't created that ...
[19:45] <stgraber> nope, that was there, otherwise I wouldn't have been able to wipe the dump-* variables
[19:45] <stgraber> also:
[19:45] <stgraber> stgraber@castiana:~$ grep grub.*archite Desktop/syslog
[19:45] <stgraber> Sep  2 17:26:02 grub-installer: info: architecture: amd64/efi
[19:46] <cjwatson> Blink
[19:50] <cjwatson> That's incredibly freaky, I can't see why it's doing that
[19:50] <cjwatson> Oh.  LVM
[19:50] <cjwatson> Bug in precise's grub-installer
[19:51] <cjwatson>     lvm:*:*:*)
[19:51] <cjwatson>         # Booting from LVM requires grub2
[19:51] <cjwatson>         grub_package="grub-pc"
[19:51] <cjwatson>         ;;
[19:51] <cjwatson> for minimal-change-fix that probably ought to be lvm:*:*:grub)
[19:51] <stgraber> oh, that'd explain it!
[19:51] <cjwatson> saucy doesn't have this bug
[19:51] <cjwatson> and I was looking there first so I didn't notice