Does anyone else wish they had doubled the number of drives and used RAID?
10 nodes @ 1 disk per node = 1 drive loss kills the node
It seems you can only buy 'extra' disks for 4 of the 10 nodes, which makes disk redundancy across all nodes impossible.