[
https://issues.apache.org/jira/browse/FLINK-20777?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17255869#comment-17255869
]
Qingsheng Ren commented on FLINK-20777:
---------------------------------------
>From my perspective, I prefer enabling discovering partitions by default. This
>is helpful when upstream Kafka topic scales out for some reason, or a new
>topic matching the topic pattern is created, so that the Flink job doesn't
>have to restart to discover the new topic/partition. I think this discovering
>action won't introduce a strong overhead to performance since it will be
>running in a thread on SplitEnumerator, and won't interweave with data
>consuming process.
> Default value of property "partition.discovery.interval.ms" is not as
> documented in new Kafka Source
> ----------------------------------------------------------------------------------------------------
>
> Key: FLINK-20777
> URL: https://issues.apache.org/jira/browse/FLINK-20777
> Project: Flink
> Issue Type: Bug
> Components: Connectors / Kafka
> Reporter: Qingsheng Ren
> Priority: Major
> Fix For: 1.12.1
>
>
> The default value of property "partition.discovery.interval.ms" is documented
> as 30 seconds in {{KafkaSourceOptions}}, but it will be set as -1 in
> {{KafkaSourceBuilder}} if user doesn't pass in this property explicitly.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
