I second other people's recommendations of rsync. rsync is great.

There's another possibility, though, that has some of the same advantages. With rsync, you can do a preliminary copy that does everything while the system is live, and then schedule a hopefully short downtime to update just those things that have changed. Dump/restore should be able to do the same thing. While your filer is up and serving, dump/restore everything at level 0. (You can do this by rsh'ing the dump to the old filer and the restore to the new server, of course.)

Then, when you have downtime scheduled, dump/restore at level 1 (making sure there were no intervening level 0 dumps). That way, you're only copying the changes, and the dump/restore will be much quicker.

(Obviously, because you're going over the net, this is going to be slower than something like "vol copy". However, you can do the level 0 dump without interrupting service, so probably you only care about the time taken by the level 1 dump.)

Basically, that amounts to doing by hand what rsync does automatically. Probably a bit more efficient than rsync (because less general).

-j.

Jay Sekora jay@ccs.neu.edu Unix Systems Administrator Northeastern University College of Computer Science