Currently Backup/Restore requests are initiated and tracked on a single Infinispan node. If a status request for a backup/restore is issued to a node other than the initiator, then a 404 response will be returned.

We should modify BackupManagerImpl so that backup/restore meta information is distributed via a cache. Status requests can then be handled by all nodes in the cluster. Requests for state, e.g. downloading a backup, received by non-initiator node can then be forwarded to the initiator node who is able to access the required state.