smjn opened a new pull request, #23964: URL: https://github.com/apache/kafka/pull/23964
Copying records into the DLQ was dominated by work on the persister thread, which also completes write responses for every share partition: - isDlqEnabledOnTopic and dlqTopicMaxMessageBytes built a LogConfig, which parses every topic config, on each DLQ request. Read the one key directly. (Affects both copy on and off case) - The source record read of ran synchronously on the persister thread. Run it (and later rounds) on a single-threaded fetch executor owned and shut down by BrokerServer. (Affects copy on case) - ShareGroupDLQRecordFetcher.collectRecords visited every batch of a 1 MiB log read, loading each file-backed batch from disk, although a request wants a few records. We can simply reject batches where the batch baseOffset > dlq endoffset (V2 only, Affects copy on case). Co-Authored-By: Claude Sonnet 5.5 <[email protected]> -- 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]
