Replacing dead nodes in a cluster

To replace a dead node, we should first remove that node using the nodetool removenode command as described earlier, and then we should add the new node as discussed in the previous section.

Note

If the new node's IP address is different to the previous dead node's IP address, then start Cassandra on the new node with the startup parameter replace_address=<IP-address-of-dead-node>.

To replace a node that is alive and being replaced due to hardware upgrade or another such reason, we should first add the new node and then decommission the old node using the nodetool decommission command, as discussed previously.