A repair server provides a potential repair mechanism after a system experiences a problem that extends beyond its redundancy tolerance (e.g. simultaneous, multiple drive failures spanning multiple nodes). Typically, a repair server is a sync destination that contains a fairly recent replication of the given system. A repair server will automatically attempt to pull back any needed blocks from the remote system, potentially avoiding the need to roll back using a snapshot. It is generally recommended that you have a repair server in place whenever possible.