kamalcph opened a new pull request, #20345: URL: https://github.com/apache/kafka/pull/20345
During broker restarts, the topic-based RemoteLogMetadataManager (RLMM) constructs the state by reading the internal __remote_log_metadata topic. When the partition is not ready to perform remote storage operations, then ReplicaNotAvailableException thrown back to the consumer. The clients retries the request immediately. This results in a lot of FETCH requests on the broker and utilizes the request handler threads. Using the CountdownLatch to reduce the frequency of ReplicaNotAvailableException thrown back to the clients. This will improve the request handler thread usage on the broker. Previously, when RLMM is not ready for a partition, then ~9K FetchConsumer requests / sec are received on the broker. With this patch, the number of FETCH requests come down to 600 / sec. -- 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
