You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Manikumar (Jira)" <ji...@apache.org> on 2020/06/05 20:34:00 UTC
[jira] [Updated] (KAFKA-10112) Consider making the number of
threads configurable for offset/group metadata cache loading
[ https://issues.apache.org/jira/browse/KAFKA-10112?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Manikumar updated KAFKA-10112:
------------------------------
Description: Currently we use [single-thread scheduler|https://github.com/apache/kafka/blob/trunk/core/src/main/scala/kafka/coordinator/group/GroupMetadataManager.scala#L84] to handle offset/group metadata cache loading and unloading. If there are leadership changes for multiple offset topic partitions, overall loading time will be high. So if you have to load 10 partitions, the 10th one will have to wait for the previous ones. We can consider making the number of threads configurable. (was: Currently we use [single-thread scheduler|https://github.com/apache/kafka/blob/trunk/core/src/main/scala/kafka/coordinator/group/GroupMetadataManager.scala#L84] to handle offset/group metadata cache loading and unloading. If there are leadership changes for multiple offset topic partitions, overall loading time will be high. So if you have to load 10 partitions, the 10th one will have to wait
for the previous ones. We can consider making the number of threads configurable.)
> Consider making the number of threads configurable for offset/group metadata cache loading
> ------------------------------------------------------------------------------------------
>
> Key: KAFKA-10112
> URL: https://issues.apache.org/jira/browse/KAFKA-10112
> Project: Kafka
> Issue Type: Task
> Components: core
> Reporter: Manikumar
> Priority: Minor
>
> Currently we use [single-thread scheduler|https://github.com/apache/kafka/blob/trunk/core/src/main/scala/kafka/coordinator/group/GroupMetadataManager.scala#L84] to handle offset/group metadata cache loading and unloading. If there are leadership changes for multiple offset topic partitions, overall loading time will be high. So if you have to load 10 partitions, the 10th one will have to wait for the previous ones. We can consider making the number of threads configurable.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)