I doubt that “Remove and Rebalance” is exactly the same as “Graceful failover and Rebalance”. Could any one know the different between this 2 things?
No, they are different. Graceful failover drops client connections ungracefully, but handles transfer of the replication with checkpoints gracefully between the nodes. Doing a remove node, rebalance ensures there is time for applications to receive topology updates, services to switch over, etc.
truly thank you for the answer.
So from what I understand from answer is that with Remove node option, Couchbase will rebalance the data first and then send the updates to client. So if I want to do Graceful failover followed by Rebalance, Remove node is better choice for me right?
Generally yes. Note that I wouldn’t call that “failover”. Technically speaking, it’s more of a “switchover” in general distributed system terms. There is no fail in this case. Remove node, rebalance is more graceful.