popsunlake commented on PR #16712: URL: https://github.com/apache/kafka/pull/16712#issuecomment-2256080608
Hi, @chia7712 I hope this finds you well. I've made some modifications to the zk-based assignment logic in Kafka, but I've left the KRaft-based assignment unchanged due to its distinct mechanism. In KRaft, once all unfenced brokers are assigned to preferred replicas, the broker list undergoes a shuffle. This design effectively distributes the creation and expansion of topics when viewed separately. However, when considered together, the issues described in KAFKA-17198 still emerge, as verified in my environment. Considering that the ReplicaPlacer interface allows for customized replica assignment policies, I am evaluating whether to implement a separate assignment strategy for KRaft. I would appreciate your insights on this matter, especially regarding the necessity and potential impact of a distinct approach for KRaft. Looking forward to your valuable suggestions. -- 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: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
