You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Lucas Bradstreet (Jira)" <ji...@apache.org> on 2020/02/26 23:37:00 UTC
[jira] [Commented] (KAFKA-8963) Benchmark and optimize incremental
fetch session handler
[ https://issues.apache.org/jira/browse/KAFKA-8963?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17045989#comment-17045989 ]
Lucas Bradstreet commented on KAFKA-8963:
-----------------------------------------
Hi [~agam], I saw that you were just assigned this issue. I added [https://github.com/apache/kafka/blob/8966d066bd2f80c6d8f270423e7e9982097f97b9/jmh-benchmarks/src/main/java/org/apache/kafka/jmh/fetchsession/FetchSessionBenchmark.java#L18] in an earlier PR, so maybe we can close it. I think to optimize the replica fetcher further we need to be smarter about how we maintain the incremental fetch sessions, and make it part of the replica fetcher itself. [~guozhang] was doing some work in this area in [https://github.com/apache/kafka/pull/7581/files].
> Benchmark and optimize incremental fetch session handler
> --------------------------------------------------------
>
> Key: KAFKA-8963
> URL: https://issues.apache.org/jira/browse/KAFKA-8963
> Project: Kafka
> Issue Type: Task
> Reporter: Lucas Bradstreet
> Assignee: Agam Brahma
> Priority: Major
>
> The FetchSessionHandler is a cause of high CPU usage in the replica fetcher for brokers with high partition counts. A jmh benchmark should be added and the incremental fetch session handling should be measured and optimized.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)