[
https://issues.apache.org/jira/browse/FLINK-23309?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dian Fu closed FLINK-23309.
---------------------------
Resolution: Fixed
Merged to master via eab562bd3c58089e8bc945a83826def3438dc25f
> Optimize the finish bundle logic in PyFlink
> -------------------------------------------
>
> Key: FLINK-23309
> URL: https://issues.apache.org/jira/browse/FLINK-23309
> Project: Flink
> Issue Type: Improvement
> Components: API / Python
> Reporter: Huang Xingbo
> Assignee: Huang Xingbo
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.14.0
>
>
> Whenever the `bundle size` or `bundle time` is reached, the data in the input
> buffer needs to be sent from the jvm to the pvm, and then waits for the pym
> to be processed and sent back to the jvm to send all the results to the
> downstream operator, which leads to a large delay in current implementation
> of main thread triggering `finish bundle`, especially when it is a small size
> event as small messages are hard to processed in pipeline.
> We need to move the logic of `finish bundle` from the main thread to the
> asynchronous sending thread, so as to avoid the problem of the main thread
> being blocked.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
