Gluster users - Jun 2012 - suggestions to improve our existing setup

Hi,

We brought our two node gluster replication down as we were blaming
poor gluster for what turns out to be filesystem issues, can't blame
gluster for replicating disappearing folders.

While the system is down there's a chance to rethink things. Would
anyone suggest a better approach regarding OS and filesystem than our
original file server setup?

:: Xen Cloud Platform
:: Two CentOS 5.7 Virtual Machines with latest gluster (one on iscsi
san and other using host local drive for storage) replicating 1TB of
files on same lan
:: 1TB accessed by separate VM serving 10000s of files under 1Mb using
samba and glusterfs mount

Thanks,

Chris