If the “master” node that ran the couchbase-cli cluster-init and couchbase-cli bucket-create is killed sometime later, the cluster seems to carry on with the other nodes that have joined. Is there anything special about the original master node or any reason why the cluster couldn’t continue as normal? I assume not, but just want to double check.
No, the nodes are all identical, it’s just that one will be designated the orchestrator at any one point in time. If you remove the node which is currently the orchestrator a new one will be elected.