It is frustrating. The OES2 SP1 docs talk about install as a XEN VM. They also talk about installing with EVMS managed partitions so that one may use un-partitioned disk space to create NSS volumes. This requires a few additional configuration steps to make sure that your /boot partition is managed by the evms.boot.

However, when one goes to install an SLES/OES server with the instructions for EVMS partitioning, all of the sudden your device(s), assigned as /dev/sdb for example, appear in the VM as /dev/xvda.

Proceeding on the hope that this will not screw anything up, the VM will no longer boot after you have executed the instructions in

Novell Documentation

...A3.3 After the linux install.

That set of instructions also says "use evms_activate to start evms if the system does not boot". Try doing that from the sh prompt, with no file systems mounted, when the boot loader cannot find root. (Pointers welcome.)

Is there something special I need to do on either the dom0 or domU filesystem to make this work? I lost a pre-production system after a kernel update, and do not want to repeat the experience.