I have 2 node Cassandra cluster. I wish to replicate the complete data into a 3rd node so that the 3rd node will have complete data (for some special needs). All the 3 nodes will have Cassandra version 2.1.8.
With replication factor 3, I plan to follow as per doc in Adding nodes to an existing cluster
Question:
Is it correct approach ? If so, when can I confidently say data replication is complete (assuming I restrict inflow of data)
Can I replicate only a subset of tables from a cassandra keyspace ?
Yes, that is the correct approach. Although, I don't see a step to adjust the replication factor on your keyspace for your datacenter. Make sure you do that just before adding your new node. Also, as your RF = number of nodes, you won't need to run nodetool cleanup on the existing nodes.
No, you cannot. You can replicate only a subset of keyspaces differently, but not actual tables themselves. If you have tables that need to be replicated differently, you should put them in their own keyspaces.