System fails to reboot after eucalyptus preseeded instlation

I've attached a preseed file that fails to reboot correctly.

Re: "Installing without the eucalyptus-udeb leads to a working base system (non-UEC)"
Could you attach the preseed used in that case, and the resulting installation syslog ?

It's strange that eucalyptus-udeb would create a divergence here (it doesn't affect the part of the install that takes care of the bootloader), especially since a clc+walrus install doesn't download any preseed that could affect the debconf settings...

Also I just tried my own CLC/Walrus preseeded install on a cloud-in-a-bag laptop, and it reboots alright.
See attached preseed.

However there is definitely some tricks involved around disk order detection in netboot (partman shows the only disk as /dev/sdb -- SCSI2). See in my preseed:

d-i partman-auto/disk string /dev/sdb
d-i grub-installer/bootdev string /dev/sda

I don't see how that would be related to eucalyptus-udeb, though...

The netboot ordering is a red herring, but getting the install syslog for the non-uec case would still be interesting.

Marc Tardif suggested to add a "; true" at the end of the late_command.

Both the preseed and the log of a failing installation are attached to the bug report.

@Mathias: see my question at comment 3, the attached preseed and log are about the failing UEC install, not the corresponding (working) non-UEC install on the same hardware. I'm trying to understand what makes this bug UEC-specific.

Downgrading prio since that could not be reproduced outside the testrig

[Expired for eucalyptus (Ubuntu) because there has been no activity for 60 days.]