[jira] [Comment Edited] (FLINK-8236) Allow to set the parallelism of table queries

["Xingcan Cui (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-8236?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16289369#comment-16289369 ] 

Xingcan Cui edited comment on FLINK-8236 at 12/13/17 3:15 PM:
--------------------------------------------------------------

Hi [~twalthr], thanks for raising this. I got a question about the {{setParallelism()}} method. In {{ExecutionConfig.java}}, I saw this default value
{code:java}
/**
* The flag value indicating use of the default parallelism. This value can
* be used to reset the parallelism back to the default state.
*/
public static final int PARALLELISM_DEFAULT = -1;
{code}
However, almost all the {{setParallelism()}} methods (e.g., SingleOutputStreamOperator.setParallelism(), StreamTransformation.setParallelism()) require the input to be at least 1. Is that a contradiction？

Thanks, Xingcan


was (Author: xccui):
Hi [~twalthr], thanks for raising this. I got a question about the {{setParallelism()}} method. In {{ExecutionConfig.java}}, I saw this default value
{code:java}
/**
* The flag value indicating use of the default parallelism. This value can
* be used to reset the parallelism back to the default state.
*/
public static final int PARALLELISM_DEFAULT = -1;
{code}
However, all the {{setParallelism()}} methods require the input to be at least 1. Is that a contradiction？

Thanks, Xingcan

> Allow to set the parallelism of table queries
> ---------------------------------------------
>
>                 Key: FLINK-8236
>                 URL: https://issues.apache.org/jira/browse/FLINK-8236
>             Project: Flink
>          Issue Type: Improvement
>          Components: Table API & SQL
>    Affects Versions: 1.4.0
>            Reporter: Timo Walther
>            Assignee: Xingcan Cui
>
> Right now the parallelism of a table program is determined by the parallelism of the stream/batch environment. E.g., by default, tumbling window operators use the default parallelism of the environment. Simple project and select operations have the same parallelism as the inputs they are applied on.
> While we cannot change forwarding operations because this would change the results when using retractions, it should be possible to change the parallelism for operators after shuffling operations.
> It should be possible to specify the default parallelism of a table program in the {{TableConfig}} and/or {{QueryConfig}}. The configuration per query has higher precedence that the configuration per table environment.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)