[
https://issues.apache.org/jira/browse/KAFKA-8931?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
HeryLong updated KAFKA-8931:
----------------------------
Description:
Recently, we are testing Kafka Connect 2.3.0, checking the incremental
cooperative rebalancing mechanism.
We found that when new connector was putted via REST endpoint, Kafka Connect
2.3.0 will restart all cached tasks;
And the issuse is 'KAFKA-8449', the PR is
'[#6850|[https://github.com/apache/kafka/pull/6850]0]';
So, the result is, new connector putted will lead to all cached tasks
restarted, which conflict to the motivation of incremental cooperative
rebalancing.
was:
Recently, we are testing Kafka Connect 2.3.0, checking the incremental
cooperative rebalancing mechanism.
We found that when new connector was putted via REST endpoint, Kafka Connect
2.3.0 will restart all cached tasks;
And the issuse is 'KAFKA-8449', the PR is
'[#6850|[https://github.com/apache/kafka/pull/6850]|https://github.com/apache/kafka/pull/6850]';
So, the result is, new connector putted will lead to all cached tasks
restarted, which conflict to the motivation of incremental cooperative
rebalancing.
> PR '#6850' fix issue 'KAFKA-8449', but lead to restart all cached tasks,
> which conflict to the motivation of incremental cooperative rebalancing.
> --------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: KAFKA-8931
> URL: https://issues.apache.org/jira/browse/KAFKA-8931
> Project: Kafka
> Issue Type: Bug
> Components: KafkaConnect
> Affects Versions: 2.3.0
> Reporter: HeryLong
> Priority: Blocker
>
> Recently, we are testing Kafka Connect 2.3.0, checking the incremental
> cooperative rebalancing mechanism.
>
> We found that when new connector was putted via REST endpoint, Kafka Connect
> 2.3.0 will restart all cached tasks;
> And the issuse is 'KAFKA-8449', the PR is
> '[#6850|[https://github.com/apache/kafka/pull/6850]0]';
>
> So, the result is, new connector putted will lead to all cached tasks
> restarted, which conflict to the motivation of incremental cooperative
> rebalancing.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
