Your cluster has 3 dedicated master nodes. After a network partition, the master was on node-master-1 which got isolated. When the partition heals, you observe that both node-master-2 and node-master-3 formed their own cluster (they reached quorum of 2) and were writing data during the partition. Now node-master-1 rejoins but has diverged cluster state. What mechanism prevents this "split-brain" scenario in Elasticsearch 8.x?
