peterxcli commented on PR #18175: URL: https://github.com/apache/kafka/pull/18175#issuecomment-2567096696
I think my original thought was wrong `unclean.leader.election.enable` has two scopes: topic-level and cluster-wide. If the cluster is configured with `unclean.leader.election.enable=true` at startup, the interval setting (`unclean.leader.election.interval.ms`) becomes irrelevant. This is because whenever the leader of a topic partition’s ISR is fenced, an unclean election will occur immediately. IMO, `unclean.leader.election.interval.ms` only become necessary when the cluster disable unclean election and enable it dynamically. > `unclean.leader.election.interval.ms` is merely a periodic task that scans for topic partitions without a leader and, based on the value of `unclean.leader.election.enable`(topic's || cluster's), determines whether to perform unclean leader elections for those partitions. --- Regarding enabling unclean elections at the cluster or topic level: Currently, in `RebootstrapTest`, only the relevant topics have unclean election enabled. I believe it is unnecessary to enable cluster-wide unclean election because those occurrence of unclean election for some topics are expected. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
