- Create a single node cluster
- Scale up the cluster to 3 nodes
- Kill the new node as soon as it gets added into the cluster
Here, Operator will wait for auto-failover to failover the killed node. But auto-failover will not be triggered to the node which is not yet rebalanced into the couchbase-cluster.
So before rebalance is successful for the new nodes, operator should remove the killed nodes in the cluster.