dongjoon-hyun commented on a change in pull request #24527: [SPARK-27635][SQL]
Prevent from splitting too many partitions smaller than row group size in
Parquet file format
URL: https://github.com/apache/spark/pull/24527#discussion_r281027718
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
##########
@@ -420,8 +420,12 @@ case class FileSourceScanExec(
selectedPartitions: Seq[PartitionDirectory],
fsRelation: HadoopFsRelation): RDD[InternalRow] = {
val openCostInBytes =
fsRelation.sparkSession.sessionState.conf.filesOpenCostInBytes
- val maxSplitBytes =
- FilePartition.maxSplitBytes(fsRelation.sparkSession, selectedPartitions)
+ val maxSplitBytes = relation.fileFormat match {
+ case _ : ParquetSource =>
+ fsRelation.sparkSession.sessionState.conf.filesMaxPartitionBytes //
parquet.block.size
+ case _ =>
+ FilePartition.maxSplitBytes(fsRelation.sparkSession,
selectedPartitions)
+ }
Review comment:
Hi, @LantaoJin . It would be very helpful if you provide a test case for
your following claim.
> Splitting RDD to too many small pieces doesn't make sense. Jobs will
launch too many partitions and never complete.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
