Expanding virtual disks presented to a MCS cluster on W2K from a SAN works well - simply increase the virtual disk on the SAN then use diskpart (not version 1.0!) and SQL continues without a problem. Decreasing the size of a volume is not so easy. Has anyone else done this without backing up the data, reinstalling the instance etc? I've been using the new disk recovery tool [in our lab] that comes with W2K3 (also works with W2K MCS) to create a new virtual disk, offline SQL Instance, swap disk with old one using recovery tool, copy data from old disk to new, delete old disk and bring SQL on-line. This works okay but it requires a random number of reboots of all nodes and seems to make things quite unstable until everything settles down. Not exactly desirable for in a production environment. Has anyone perfected the process for W2K MCS?

Why would you want to decrease the size of a volume to be
lower? The reality is whether expanding or contracting
disk space, it's a low level operation that's a
fundamental change to the OS. You won't get around
reboots.

on the SAN then use diskpart (not version 1.0!) and SQL
continues without a problem. Decreasing the size of a
volume is not so easy. Has anyone else done this without
backing up the data, reinstalling the instance etc? I've
been using the new disk recovery tool [in our lab] that
comes with W2K3 (also works with W2K MCS) to create a new
virtual disk, offline SQL Instance, swap disk with old one
using recovery tool, copy data from old disk to new,
delete old disk and bring SQL on-line. This works okay
but it requires a random number of reboots of all nodes
and seems to make things quite unstable until everything
settles down. Not exactly desirable for in a production
environment. Has anyone perfected the process for W2K MCS?