Nice -- to see this practice described as 'crazy'.



Have done it myself several times, and it's quite a safe method: if you make any mistakes you don't spoil the original disk, so there is a fallback until the end. (And nothing has changed on the original disk until you come to the lilo/grub changes.)



You could even postpone the wipe/attachment of the original disk a few days to ensure that things are running safely.



The best reason for this method is that one can do this from places like a comfortable home/office chair/bed, which eliminates a lot of mistakes. :)

Well, I am an old-fashioned sysadmin in many things... including switching a production server's storage strategy ;-)

I've done this numerous times some 2 years ago with redhat 7.3. Actually this technique is described in RAID HOWTO, and this isn't so risky, you just have to be VERY careful at performing procedure's steps.