Manual install fail when using encrypted LVM

[Stéphane Graber]

Binary package hint: debian-installer

That happens with Intrepid alpha-3 candidates.

When doing a manual partitioning and using encrypted LVM with the following partitioning:
/boot
lukfs partition
 --> LVM
       ---> /
       ---> /home
       ---> swap

The installer complains that the swap space is unsafe and should be put inside the encrypted LVM which it actually is.
As it's a major error the installer stops here and I have no way of continuing the installation.

Attached is a complete syslog

[Han Chung (han+c)]

I can confirm this problem also happens for me.

If I let it automatically decide the size of the partitition and create encrypted lvm volume, it will install fine, but it can not mount the encrypted LVM after typing in the passphrase during the boot process right after install.

[Han Chung (han+c)]

Forgot to comment I tested this on Intrepid Ipex Alpha-3 AMD64/x86-64 on:
Qemu/KVM on a Intel Core 2 Duo T5600,
with: 512MB allocated for KVM,
5.4GB disk space
-smp 2

[Phil Norbeck (ptn107)]

Confirmed on intrepid alpha 4 (daily 8-16-08) x86_64 alternate cd.

[Colin Watson (cjwatson)]

I've tried to reproduce this with the 2008-09-10 daily build of the server CD, and had no luck; both automatic and manual encrypted LVM setups seem to work fine. It's possible that we fixed this in the process of fixing something else, without realising that it was connected to this bug. In particular I wonder if this was related:

partman-lvm (61ubuntu1) intrepid; urgency=low

  * Backport from trunk (Jérémy Bobbio):
    - Fix size substitution when displaying partman-lvm/lvcreate_error.
    - Don't fiddle with label on devices of type loop (e.g. RAID or crypto)
      in update.d/lvm_sync_flag. (Closes: #494910, LP: #267048)

 -- Colin Watson <email address hidden> Sun, 07 Sep 2008 20:27:57 +0100

Could somebody who was able to reproduce this beforehand please try it again with a current daily build (*not* Alpha 5, which predated the suspected fix above)? If you can reproduce it, please try to give me exact directions so that I can reproduce it inside a clean virtual machine. Thanks in advance.

[Steve Langasek (vorlon)]

I reproduced this bug, altogether incidentally, while doing an install with alpha-5; and in alpha-6, I confirm that it now appears to be fixed. Marking 'fix released', thanks!