[
https://issues.apache.org/jira/browse/SPARK-58138?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Nikolina Vraneš updated SPARK-58138:
------------------------------------
Description:
BinByExec does not override outputPartitioning, so it falls back to
UnknownPartitioning. A downstream operator clustered on a pass-through column
that the child already partitions on then gets an unnecessary shuffle.
This makes BinByExec preserve a child partitioning whose keys are pass-through
(or range) columns, while dropping any partitioning on a scaled DISTRIBUTE
column (its output attribute has a fresh ExprId).
Sub-task of SPARK-57133.
> Preserve child output partitioning through BIN BY
> -------------------------------------------------
>
> Key: SPARK-58138
> URL: https://issues.apache.org/jira/browse/SPARK-58138
> Project: Spark
> Issue Type: Sub-task
> Components: SQL
> Affects Versions: 5.0.0
> Reporter: Nikolina Vraneš
> Priority: Major
> Labels: pull-request-available
>
> BinByExec does not override outputPartitioning, so it falls back to
> UnknownPartitioning. A downstream operator clustered on a pass-through column
> that the child already partitions on then gets an unnecessary shuffle.
> This makes BinByExec preserve a child partitioning whose keys are
> pass-through (or range) columns, while dropping any partitioning on a scaled
> DISTRIBUTE column (its output attribute has a fresh ExprId).
> Sub-task of SPARK-57133.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
