[ 
https://issues.apache.org/jira/browse/KAFKA-10339?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17767334#comment-17767334
 ] 

Greg Harris commented on KAFKA-10339:
-------------------------------------

Please note that this KAFKA ticket only applies to data replication being 
exactly once. This means that if you replicate data between two topics, and 
consume from the target topic from the beginning with read_committed, you will 
read each record in the source topic exactly once.

It does _not_ apply to the MirrorCheckpointTask, offset translation, or 
consumer group auto sync. These processes are _not_ exactly once, they are 
at-least-once. This means that if you commit offsets on your upstream topic and 
then start consuming from the target topic with the translated offsets, you may 
read the same record more than once. You should read every record at least once.

> MirrorMaker2 Exactly-once Semantics
> -----------------------------------
>
>                 Key: KAFKA-10339
>                 URL: https://issues.apache.org/jira/browse/KAFKA-10339
>             Project: Kafka
>          Issue Type: New Feature
>          Components: KafkaConnect, mirrormaker
>            Reporter: Ning Zhang
>            Assignee: Ning Zhang
>            Priority: Major
>              Labels: needs-kip
>             Fix For: 3.5.0
>
>
> MirrorMaker2 is currently implemented on Kafka Connect Framework, more 
> specifically the Source Connector / Task, which do not provide exactly-once 
> semantics (EOS) out-of-the-box, as discussed in 
> https://github.com/confluentinc/kafka-connect-jdbc/issues/461,  
> https://github.com/apache/kafka/pull/5553, 
> https://issues.apache.org/jira/browse/KAFKA-6080  and 
> https://issues.apache.org/jira/browse/KAFKA-3821. Therefore MirrorMaker2 
> currently does not provide EOS.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to