Thanks ernie.

Just one question - I assume that UUIDs are specific to the actual hardware, rather than just the ISO image of the partition within that hardware?

If this is the case, then I wouldn't be able to use UUIDs in my case, because in my application I have to clone the ISO image across many different drives.

O.K., then perhaps I do not quite understand, but in a situation like that, I think I would try to create a script to do the work for me. The script might run the cloning process on a drive, then launch a chrooted child process to get the drives UUID and configure that system's /etc/lilo.conf file. Since I do not know the specific limitations under which you are working, I will not try to be more specific, but I hope this helps you work out a solution.

Thanks. That might work, though with speed limitations...so there's no way I can use label instead of UUID? Seems like that would be easier, though I can't seem to get it working.

Well, you must have the partition labeled or accessing the partition by its label name will not work. Labeling the partition requires addressing the drive by its device node which may be (if I recall correctly) either sdb2 or hdb2. The partition already has a UUID associated with it, so getting that may be easier. If you prefer setting the partition label, you may be able to get it to work if you know which device node is assigned to each computer with which you will work, but then if you know that, you do not need to set the label in the first place, do you?

If you choose to access the partition using the UUID, you can use this scenario: try to get the partition's UUID number from the sdb2 device node. Grep the output of dumpe2fs for the UUID. Put the UUID number into a variable that has been initialized with a null value, test the variable for a null value. if it is null, you did not get a UUID so try to get the UUID number using the hdb2 device node. This way you can cover both possibilities and make sure you have a value other than null.

HTH,