This howto explains how I migrated/converted a Debian Lenny vserver container to a KVM (libvirt) virtual machine. The instructions can only be followed as-is if you used the newvserver script (from the vserver-debiantools package) to create your vserver container.

First, create and start a new VM with enough diskspace and a suitable liveCD. I chose grml, which I can highly recommend for this type of task. It contains a SSH daemon as well as LVM and SWRAID support.

kvmbox$ virt-install --connect=qemu:///system -n $vmname -r 256 \
--os-variant=debianlenny -c grml_2010.04.iso --livecd --disk \
path=... -w bridge=br_eth0 -k $kblayout --vnc

Find and connect to the VNC display.

kvmbox$ virsh --connect=qemu:///system vncdisplay $vmname
:3
yourbox$ ssh -f -L 5903:localhost:5903 kvmbox sleep 60m
yourbox$ vncviewer localhost:3

In the liveCD on the VM, set a root password and start the SSH daemon.

grml$ sudo passwd root
grml$ sudo /etc/init.d/ssh start

SSH into the liveCD on the VM, it's more handy to work that way. Partition, create and mount filesystems. /proc and /dev are bind mounted in order for grub-install to work. I am opting to use a swapfile, thus there's no mkswap command.

yourbox$ ssh root@$vmip
grml# cfdisk /dev/vda
grml# mkfs.ext3 /dev/vda1
grml# mount /dev/vda1 /mnt
grml# mount --bind /dev /mnt/dev
grml# mount --bind /proc /mnt/proc

Now it's time to stop and disable the vserver container, and to copy the filesystem over to the VM.

vserverbox$ sudo vserver $vmname stop
vserverbox$ sudo rm /etc/vservers/$vmname/apps/init/mark
vserverbox$ sudo tar --numeric-owner -c /etc/vservers/$vmname/vdir/ | \
ssh root@$vmip "tar -x --strip=4 -C /mnt"

In the VM, chroot into the newly copied filesystem and remove the vserver customizations. GRUB and kernel are also needed.

grml# chroot /mnt /bin/bash

Re-enable hardware-related initscripts:
chroot# pkgs=$(for f in /etc/rc3.d/K90*; do \
 dpkg -S $(readlink -f $f); done | cut -d: -f1 | sort | uniq)
chroot# for f in /etc/rc3.d/K90*; do update-rc.d -f $(basename $(readlink -f $f)) remove; done
chroot# for pkg in $pkgs; do dpkg-reconfigure $pkg; done

The above commands depend on the fact that the newvserver script disables
such initscripts by deleting all links and setting them to be stopped in
runlevel 3 at priority 90.

Add root filesystem and swap to /etc/fstab:
/dev/sda1       /       ext3    errors=remount-ro       0       1
/swapfile   none   swap   sw   0   0

Configure network interfaces in /etc/network/interfaces:
# The loopback network interface
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet static
address 192.168.0.xyz
netmask 255.255.255.0
gateway 192.168.0.1

Re-add ttys to /etc/inittab:
1:2345:respawn:/sbin/getty 38400 tty1
2:23:respawn:/sbin/getty 38400 tty2
3:23:respawn:/sbin/getty 38400 tty3
4:23:respawn:/sbin/getty 38400 tty4
5:23:respawn:/sbin/getty 38400 tty5
6:23:respawn:/sbin/getty 38400 tty6

Install grub and kernel and acpid:
chroot# mkdir /boot/grub
chroot# aptitude install grub linux-image-2.6-686 acpid

Abort initrd kernel image installation? No.

Install grub to the VM's MBR:
chroot# grub-mkdevicemap
chroot# grub-install /dev/vda
chroot# update-grub

Exit the chroot, and shutdown the VM.
chroot# exit
grml# umount /mnt/proc
grml# umount /mnt/dev
grml# umount /mnt
grml# halt

To make it easier to restart the process, the virt-install command I used earlier specified that the boot device should always be CD. To boot from HD, follow these steps.

kvmbox$ virsh --connect=qemu:///system edit $vmname

Change the boot dev to hd.

You can now start the VM.