You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "yikaifei (Jira)" <ji...@apache.org> on 2023/05/05 12:38:00 UTC
[jira] [Updated] (SPARK-43385) The Generator's statistics should be ratio times greater than the child nodes
[ https://issues.apache.org/jira/browse/SPARK-43385?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
yikaifei updated SPARK-43385:
-----------------------------
Description:
The Generator's statistics should be ratio times greater than the child nodes.
Generator is an expression that produces zero or more rows given a single input row.
If `UserDefinedGenerator` and `HiveUDTF` were used, the output could be N times that of the child node, resulting in a statistical error.
Because of incorrect statistics, Spark may select an incorrect execution plan. For example, if `BroadcastHashJoinExec` is selected, the Job fails to broadcast buildSide.
was:The Generator's statistics should be ratio times greater than the child nodes.
> The Generator's statistics should be ratio times greater than the child nodes
> -----------------------------------------------------------------------------
>
> Key: SPARK-43385
> URL: https://issues.apache.org/jira/browse/SPARK-43385
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 3.4.0
> Reporter: yikaifei
> Priority: Minor
> Fix For: 3.5.0
>
>
> The Generator's statistics should be ratio times greater than the child nodes.
> Generator is an expression that produces zero or more rows given a single input row.
> If `UserDefinedGenerator` and `HiveUDTF` were used, the output could be N times that of the child node, resulting in a statistical error.
> Because of incorrect statistics, Spark may select an incorrect execution plan. For example, if `BroadcastHashJoinExec` is selected, the Job fails to broadcast buildSide.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org