[ 
https://issues.apache.org/jira/browse/SPARK-52486?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

zhixingheyi-tian updated SPARK-52486:
-------------------------------------
    Description: 
During the join operations of a dozen partitioned fact tables in Spark-SQL, 8w+ 
subqueries were generated, ultimately causing a Driver OOM (Out-Of-Memory) 
error.

 

for example
{code:java}
Subquery:88573 Hosting operator id = 94 Hosting Expression = ds#22 IN 
dynamicpruning#76
AdaptiveSparkPlan (76)
+- Project (75)
   +- Filter (74)
      +- Scan parquet spark_catalog.dlctest.table_1 (73) {code}
 

 

The attachment "reproducing_sqlcase" can trigger substantial memory consumption:

 

!image-2025-06-18-15-23-39-175.png|width=611,height=109!

  was:
During the join operations of a dozen partitioned fact tables in Spark-SQL, 8w+ 
subqueries were generated, ultimately causing a Driver OOM (Out-Of-Memory) 
error.

 

for example

、、、

Subquery:88573 Hosting operator id = 94 Hosting Expression = ds#22 IN 
dynamicpruning#76
AdaptiveSparkPlan (76)
+- Project (75)
   +- Filter (74)
      +- Scan parquet spark_catalog.dlctest.table_1 (73)

、、、


The attachment "reproducing_sqlcase" can trigger substantial memory consumption:


!image-2025-06-18-15-23-39-175.png!


> AQE Planning caused Spark Driver OOM before Execution
> -----------------------------------------------------
>
>                 Key: SPARK-52486
>                 URL: https://issues.apache.org/jira/browse/SPARK-52486
>             Project: Spark
>          Issue Type: Bug
>          Components: SQL
>    Affects Versions: 3.5.3
>            Reporter: zhixingheyi-tian
>            Priority: Major
>         Attachments: image-2025-06-18-15-23-39-175.png, reproducing_sqlcase
>
>
> During the join operations of a dozen partitioned fact tables in Spark-SQL, 
> 8w+ subqueries were generated, ultimately causing a Driver OOM 
> (Out-Of-Memory) error.
>  
> for example
> {code:java}
> Subquery:88573 Hosting operator id = 94 Hosting Expression = ds#22 IN 
> dynamicpruning#76
> AdaptiveSparkPlan (76)
> +- Project (75)
>    +- Filter (74)
>       +- Scan parquet spark_catalog.dlctest.table_1 (73) {code}
>  
>  
> The attachment "reproducing_sqlcase" can trigger substantial memory 
> consumption:
>  
> !image-2025-06-18-15-23-39-175.png|width=611,height=109!



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

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to