Hi!
Some weeks ago the "Running Xen"-book was recommended to me on this list. A nice book, but it gave me crazy idea: One of the recommendations (which sounds very reasonable to me) was to have instead of a single file that serves as a disk-image to have to files: one of them (which is mounted to hda1) serves as a partition that holds the data, the other one (which mounts to hda2) is the swap partition. The text in the books (admittedly it's a bit vague there) lead me to the conclusion that to the Xen-machine they will look like two partitions of a single drive hda. The advantage should be that it is much easier to extend the data partition (Take the machine offline. Extend the image with dd. Extend the filesystem on the image)
I created the two files as described in book. "Formatted" them with mkswap and mkfs.ext3 (as described in the book) and added them to the configuration file for the host. Now when I start the installation of the host machine right in the beginning I get a message "/dev/hda1 has a loop partition layout. To use this disk for the installation of CentOS it must be initialized". When I don't allow formatting I get caught in a loop ("Is a loop partition. Initialize?"), when I allow formatting I get a partition on each of the devices (hda11 and hda21) and I would be allowed to go on with the installation, but IMHO this would defeat the purpose of the exercise
So my question: - is there some error in thinking on my side? - is this a situation that the installer can't deal with? (in other words: would it work if I copied a complete installation into the root partition-image and then booted that?) - or is there a reason that using partition images is not very popular (Googling around did not reveal anything useful)
Bernhard