[
https://issues.apache.org/jira/browse/SPARK-38406?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
XiDuo You updated SPARK-38406:
------------------------------
Description:
If shuffle is skewed with tens of thousands of map partitions and reduce
partitions in AQE, the method `ShufflePartitionsUtil#createSkewPartitionSpecs`
will be very slow. More unfortunately, it is running at driver side.
was:
If shuffle is skewed with tens of thousands of map partitions and reduce
partitions in AQE, the method `ShufflePartitionsUtil#createSkewPartitionSpecs`
will be very slow. More unfortunately, it is running at driver side.
See the Flame Graph of my local test env:
!image-2022-03-03-20-24-48-244.png!
We can see, a lot of cpu is running for build seq.
> Improve perfermance of ShufflePartitionsUtil createSkewPartitionSpecs
> ---------------------------------------------------------------------
>
> Key: SPARK-38406
> URL: https://issues.apache.org/jira/browse/SPARK-38406
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 3.3.0
> Reporter: XiDuo You
> Priority: Major
>
> If shuffle is skewed with tens of thousands of map partitions and reduce
> partitions in AQE, the method
> `ShufflePartitionsUtil#createSkewPartitionSpecs` will be very slow. More
> unfortunately, it is running at driver side.
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
