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

Posted to issues@flink.apache.org by "Rogan Morrow (Jira)" <ji...@apache.org> on 2023/09/14 21:58:00 UTC

[jira] [Updated] (FLINK-33091) Limit on outgoing connections to 64 seems unnecessary

     [ https://issues.apache.org/jira/browse/FLINK-33091?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Rogan Morrow updated FLINK-33091:
---------------------------------
    Affects Version/s: 1.17.1

> Limit on outgoing connections to 64 seems unnecessary
> -----------------------------------------------------
>
>                 Key: FLINK-33091
>                 URL: https://issues.apache.org/jira/browse/FLINK-33091
>             Project: Flink
>          Issue Type: Improvement
>    Affects Versions: 1.17.1
>            Reporter: Rogan Morrow
>            Priority: Major
>
> We have a job that results in a node having more than 64 outputs. However submitting the job fails with error "Cannot currently handle nodes with more than 64 outputs.". The error originates from this line:
> [https://github.com/apache/flink/blob/e7eeea033a68e1ff6bf82132b5a59eb0a5a2d0ed/flink-optimizer/src/main/java/org/apache/flink/optimizer/dag/OptimizerNode.java#L355]
> There is no explanation in the code for why this arbitrary limit is set, so I'm wondering what the purpose of it is.
> After forking Flink and removing the above line so that there is no limit, the job works. So it seems that the limit is unnecessary and can be removed.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)