If a DataNode fails, is it possible to just move the partition to another datanode? Does that work?
I've got a very small cluster and was considering using a few USB drives on extra DataNodes to increase my replication factor. Each USB drive would have its own dedicated partition to hadoop. If for whatever reason one of my backup machines running DataNodes failed, I'd like to just move the drive to another machine, configure it as a DataNode and start it back up. If I move the drive from one machine to another, do I have to move it to a drive that is not already running a DataNode?
