Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.Original topic: 想问问configchange的时候，这种情况该怎么处理
|
username: wzharies
When performing a config change and adding a new node 5, node 5 is successfully created but due to a network partition, it remains unaware of the existence of other nodes and thus becomes the leader and writes a no-op. In another partition, nodes 1, 2, 3, and 4 have node 1, 2, and 3 already removed, leaving only node 4. When the partitions merge, node 4 becomes aware of node 5’s existence, but node 4 cannot successfully send a vote request to node 5 because node 4’s LogTerm is smaller.
As a result, the entire cluster experiences a cycle where node 4’s requestVote is rejected, it increases its Term, and sends it to node 5. Node 5 steps down, then re-elects itself as the leader. Node 4’s requestVote is rejected again, and this cycle continues.
If a new peer is added and the leader steps down or no leader is elected, the new node is likely to time out and initiate an election, leading to its own election as the leader.
|
username: 
|
username: