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 2019/09/12 23:18:00 UTC

[jira] [Resolved] (KAFKA-8899) Optimize Partition.maybeIncrementLeaderHW

     [ https://issues.apache.org/jira/browse/KAFKA-8899?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Lucas Bradstreet resolved KAFKA-8899.
-------------------------------------
    Resolution: Duplicate

Duplicate of https://issues.apache.org/jira/browse/KAFKA-8841

> Optimize Partition.maybeIncrementLeaderHW
> -----------------------------------------
>
>                 Key: KAFKA-8899
>                 URL: https://issues.apache.org/jira/browse/KAFKA-8899
>             Project: Kafka
>          Issue Type: Task
>          Components: core
>    Affects Versions: 2.3.0, 2.2.1
>            Reporter: Lucas Bradstreet
>            Priority: Major
>
> Partition.maybeIncrementLeaderHW is in the hot path for ReplicaManager.updateFollowerFetchState. When replicating between brokers with high partition counts, maybeIncrementLeaderHW becomes expensive, with much of the time going to calling Partition.remoteReplicas which performs a toSet conversion. maybeIncrementLeaderHW should avoid generating any intermediate collections when calculating the new HWM.



--
This message was sent by Atlassian Jira
(v8.3.2#803003)