Since many people interested in GlusterFS also seem interested in
fault-tolerance (and probably have clients that can't run GlusterFS), I
though I'd point out an interesting project. Take a look at:
http://ctdb.samba.org
It is designed to use a cluster filesystem (such as GlusterFS) shared
amongst a cluster that can reexport the filesystem from multiple members
of the cluster as if they were one (LVS-style). Although obviously
intended for Samba, it should work for NFS, too (they have instructions
for it).
So, you could have a cluster exporting GlusterFS, parallel CIFS, and
parallel NFS! A fault-tolerant, high-performance fileserving
jack-of-all-trades; very cool (could make an incredible NAS product; add
more bricks, serve them up in parallel, handle redundancy and failover)...
Of course, NFS reexport has to work first (SAMBA already seems to work
fine on GlusterFS).
Thanks,
Brent Nelson
Director of Computing
Dept. of Physics
University of Florida
