In distributed mode, consider the reliability of cluster logs? #21

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.