[
https://issues.apache.org/jira/browse/FLINK-5018?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16110090#comment-16110090
]
Eron Wright commented on FLINK-5018:
-------------------------------------
An idle timeout for case (2) has some interesting semantic implications. When
an operator goes idle it releases its hold on the event-time clock, thus
allowing time to advance. Thus the processing-time clock (which governs the
idleness timeout) will have an increased interplay with the event-time clock,
and the timeout will act as a limit on out-of-orderness.
> Make source idle timeout user configurable
> ------------------------------------------
>
> Key: FLINK-5018
> URL: https://issues.apache.org/jira/browse/FLINK-5018
> Project: Flink
> Issue Type: Sub-task
> Components: DataStream API
> Reporter: Tzu-Li (Gordon) Tai
>
> There are 2 cases where sources are considered idle and should emit an idle
> {{StreamStatus}} downstream, taking Kafka consumer as example:
> - The source instance was not assigned any partitions
> - The source instance was assigned partitions, but they currently don't have
> any data.
> For the second case, we can only consider it idle after a timeout threshold.
> It would be good to make this timeout user configurable besides a default
> value.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
