In the current distributed cluster, when each segment writes topic data, is it necessary to set a replication factor for the distributed cluster to support a scenario where there are a total of 2N+1 nodes, where N nodes fail, but the topic data can still be recovered? This is consistent with the reliability of Raft metadata.
In the current distributed cluster, when each segment writes topic data, is it necessary to set a replication factor for the distributed cluster to support a scenario where there are a total of 2N+1 nodes, where N nodes fail, but the topic data can still be recovered? This is consistent with the reliability of Raft metadata.