Description
Remote cluster reference contains a host name, which points to a specific node on target. In order to keep remote cluster reference useful when this particular node becomes inaccessible, we keep a list of alternative nodes on target within the remote cluster reference. Periodically, we check if the host name is still accessible, if not, we try alternative nodes and update host name with useable alternative node.
This is currently done in replication code when we check if target bucket is still valid. This is not adequate or efficient. If should be done outside of replication.
1. It needs to be done even if no replication is running.
2. It needs to be done only once (per time period) for each remote cluster reference, instead of once per replication.