Moving a XEN-Instance from etch to lenny
With the upgrade from Debian etch to lenny i’ve set up an new machine to serv my XEN-Instances. First install the Dom0 on the new hardware and create a logical volume (here my-lvm-vg on/dev/sdb) for further expansion:
pvcreate /dev/sdb
vgcreate my-lvm-vg /dev/sdb
Then bring up your network by editing the /etc/xen/xend-config.sxp similar to the old system (e. g. network-bridge). If you have broken network interfaces in Dom0, follow this link.
Now create new logical volumes with the same size as the old ones:
lvcreate -n host1.fqdn-disk -L 10G my-lvm-vg
lvcreate -n host1.fqdn-swap -L 256M my-lvm-vg
Copy the DomU’s using scp to the new lenny machine:
On the old etch-machine
xm shutdown <domain-id>
dd if=/dev/my-lvm-vg/host1.fqdn-disk bs=1k | ssh -C root@<new-machine> dd of=/dev/my-lvm-vg/host1.fqdn-disk bs=1k
scp /etc/xen/host1.fqdn.cfg root@<new-machine>:/etc/xen/
After that check the filesystem of the new DomU’s with _e2fsck -f /dev/my-lvm-vg/host1.fqdn-disk_and edit your host1.fqdn.cfg to use the new kernel version. At the moment e. g. :
kernel = '/boot/vmlinuz-2.6.26-2-xen-686'
ramdisk = '/boot/initrd.img-2.6.26-2-xen-686'
All your DomU’s needs the suitable kernel module. Simultaneous edit the _/etc/init.d/hwclock.sh_inside the DomU to prevent hanging on boot. Insert a exit 0 in the second line after #!/bin/sh.
mount -o loop /dev/my-lvm-vg/host1.fqdn-disk /mnt/
cp -vdpR /lib/modules/2.6.26-2-xen-686 /mnt/lib/modules/
vi /mnt/etc/init.d/hwclock.sh
#!/bin/sh
exit 0
umount /mnt/
Another pitfall is the crond-daemon start. It hangs also at boot. The solution is to add a line extra = ‘console=hvc0 xencons=tty’ in your host1.fqdn.cfg (see xen-support).
The last two errors are the swap- and ssh-login-error. So boot the DomU an do this (my swap is here /dev/sda2):
mkswap /dev/sda2
swapon -a
If you try to login in your DomU using SSH, you get an ssh error: PTY allocation request failed …. Install udev on the DomU will fix it.
Done!
Tom