I'm trying to think of a solution for this as well. I have several droplets and I would like to share *some* disk space between them, in a redundant way (I don't want one droplet going down to render some files inaccessible).
I'm used to do this in high-availability clusters with OCFS2 and a space on a SAN, with which (OCFS2) I've had great success (not so much with NFS v2). It is my understanding DRBD and GlusterFS are a little tricky to configure (they require exactly the same amount of blocks on each disk) and a little inefficient for large system (I read from GlusterFS that for anything serious you should have 8GB per machine).
Ideally, I would use one droplet as an OCFS2 "server", but I don't think I can do that (you need another type of connectivity to mount the disks on the hosts).
I'll try reading more about NFS3 for now, see if it brings anything better that would fix the caching issues I had with its previous version.