[
https://issues.apache.org/jira/browse/FLINK-35631?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17856110#comment-17856110
]
elon_X commented on FLINK-35631:
--------------------------------
[~robyoung] Ah, I have already fixed this issue. Once it is confirmed to be a
bug, I will submit my code:(
> KafkaSource parameter partition.discovery.interval.ms with a default value of
> 5 minutes does not take effect
> ------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-35631
> URL: https://issues.apache.org/jira/browse/FLINK-35631
> Project: Flink
> Issue Type: Improvement
> Components: Connectors / Kafka
> Affects Versions: kafka-3.1.0
> Reporter: elon_X
> Priority: Major
>
> When I start a stream program to consume Kafka
> (flink-connector-kafka-3.1-SNAPSHOT) the Flink task does not automatically
> detect new partitions after Kafka adds partitions.
>
> *Reason*
> In the {{{}KafkaSourceBuilder{}}}, this parameter is checked to see if it has
> been overridden. Since I did not set this parameter, even though it is
> {{{}CONTINUOUS_UNBOUNDED{}}}, it still sets
> {{{}partition.discovery.interval.ms = -1{}}}.
> In the {{{}KafkaSourceEnumerator{}}}, the value of
> {{partition.discovery.interval.ms}} is then -1, instead of the default value
> of 5 minutes, so automatic partition discovery does not work, and the default
> value of 5 minutes for {{partition.discovery.interval.ms}} is meaningless.
>
> A possible solution is to set {{partition.discovery.interval.ms = -1}} only
> if {{boundedness == Boundedness.BOUNDED}} is true.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
