[
https://issues.apache.org/jira/browse/FLINK-28934?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yufan Sheng updated FLINK-28934:
--------------------------------
Description:
!image-2022-08-11-23-27-04-268.png|width=500px!
The image here shows if we start a Flink application with four parallelism and
four splits. All the splits would be sent to the first added reader. This is
because we don't assign splits by pre-divide splits according to the size of
parallelism. The readers are added to the enumerator one by one in the first
bootstrap.
was:
!image-2022-08-11-23-27-04-268.png|width=300px!
The image here shows if we start a Flink application with four parallelism and
four splits. All the splits would be sent to the first added reader. This is
because we don't assign splits by pre-divide splits according to the size of
parallelism. The readers are added to the enumerator one by one in the first
bootstrap.
> Pulsar Source put all the splits to only one parallelism when using Exclusive
> subscription
> ------------------------------------------------------------------------------------------
>
> Key: FLINK-28934
> URL: https://issues.apache.org/jira/browse/FLINK-28934
> Project: Flink
> Issue Type: Bug
> Reporter: Yufan Sheng
> Priority: Major
> Attachments: image-2022-08-11-23-27-04-268.png
>
>
> !image-2022-08-11-23-27-04-268.png|width=500px!
> The image here shows if we start a Flink application with four parallelism
> and four splits. All the splits would be sent to the first added reader. This
> is because we don't assign splits by pre-divide splits according to the size
> of parallelism. The readers are added to the enumerator one by one in the
> first bootstrap.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
