@Gaurav JainThe URL provided when adding the Remote Process Group (RPG) to your canvas must be successful only when initially added. Once a successful connection is established the target instance will return a list of currently connected cluster nodes. The source instance with the RPG will record those hosts in peer files. From that point forward the RPG constantly updates the list of available nodes and will not only load-balance to those nodes but will also use anyone of them to get an updated status. Lets assume your source instance of NiFi has trouble getting a status update from any of the nodes, it will still attempt to load-balance with failover delivery of data to the last known set of nodes until communication is successful in getting an updated list.
In addition, NiFi will also allow you to specify multiple URLs in the RPG when you create it. Simply provide a comma separated list of URLS for the nodes in the same target cluster. This does not change how the RPG works. It will still constantly retrieve a new listing of available nodes. This allows the target cluster to scale up or down without affecting your Site-To-Site (S2S) functionality.
Thanks,
Matt
