[
https://issues.apache.org/jira/browse/FLINK-19099?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17188410#comment-17188410
]
Aljoscha Krettek commented on FLINK-19099:
------------------------------------------
Did you completely cancel your job and start it fresh? Without starting it from
a savepoint or retained checkpoint?
If you want the Flink job to keep the read offset you need to take a savepoint
before stopping the job or stop it with a savepoint and then restart the job
later from that savepoint.
> consume kafka message repeat
> ----------------------------
>
> Key: FLINK-19099
> URL: https://issues.apache.org/jira/browse/FLINK-19099
> Project: Flink
> Issue Type: Bug
> Components: API / DataStream, Connectors / Kafka
> Affects Versions: 1.11.0
> Reporter: zouwenlong
> Priority: Major
> Attachments: image-2020-08-31-19-23-59-038.png,
> image-2020-08-31-19-25-09-548.png
>
>
> when taksmanager be killed ,my job consume some message , but the offset in
> not commit,
> then restart it ,my job consume kafka message repeat, I used checkpoint and
> set 5 seconds ,
> I think this is a very common problem，how to solve this problem？
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
