You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "Stephan Ewen (Jira)" <ji...@apache.org> on 2020/10/01 12:00:15 UTC

[jira] [Commented] (FLINK-18044) Add the subtask index information to the SourceReaderContext.

    [ https://issues.apache.org/jira/browse/FLINK-18044?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17205439#comment-17205439 ] 

Stephan Ewen commented on FLINK-18044:
--------------------------------------

I am not strictly opposed to exposing the Subtask Id.

It is more that this encourages users to write code that is not oblivious to parallelism, as code ideally should be.
Code that makes explicit assumptions about the current parallelism and the current subtask index would break when we try to introduce mode dynamic models (where some subtasks can split/scale out without affecting other tasks).

> Add the subtask index information to the SourceReaderContext.
> -------------------------------------------------------------
>
>                 Key: FLINK-18044
>                 URL: https://issues.apache.org/jira/browse/FLINK-18044
>             Project: Flink
>          Issue Type: Sub-task
>          Components: Connectors / Common
>            Reporter: Jiangjie Qin
>            Priority: Major
>              Labels: pull-request-available
>
> It is useful for the `SourceReader` to retrieve its subtask id. For example, Kafka readers can create a consumer with proper client id.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)