Skip to content

[SPARK-37530][Core] Spark reads many paths very slow though newAPIHadoopFile #34792

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
yaooqinn wants to merge 1 commit into from
Closed

[SPARK-37530][Core] Spark reads many paths very slow though newAPIHadoopFile #34792

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions core/src/main/scala/org/apache/spark/rdd/NewHadoopRDD.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -123,6 +123,10 @@ class NewHadoopRDD[K, V](

override def getPartitions: Array[Partition] = {
val inputFormat = inputFormatClass.getConstructor().newInstance()
// setMinPartitions below will call FileInputFormat.listStatus(), which can be quite slow when
// traversing a large number of directories and files. Parallelize it.
_conf.setIfUnset(FileInputFormat.LIST_STATUS_NUM_THREADS,
Runtime.getRuntime.availableProcessors().toString)

@yaooqinn yaooqinn Dec 3, 2021

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe we shall revamp the value for k8s?

@cloud-fan cloud-fan Dec 3, 2021

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you refer to the existing code if we have done this before?

@yaooqinn yaooqinn Dec 3, 2021

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes,

spark/core/src/main/scala/org/apache/spark/rdd/WholeTextFileRDD.scala

Lines 45 to 46 in 2d085c1

conf.setIfUnset(FileInputFormat.LIST_STATUS_NUM_THREADS,
Runtime.getRuntime.availableProcessors().toString)

spark/core/src/main/scala/org/apache/spark/rdd/BinaryFileRDD.scala

Lines 42 to 43 in 2d085c1

conf.setIfUnset(FileInputFormat.LIST_STATUS_NUM_THREADS,
Runtime.getRuntime.availableProcessors().toString)

@LuciferYang LuciferYang Dec 3, 2021

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will this availableProcessors get the actual number of available CPU cores in container instances, VM instances and physical machine instances?

@LuciferYang LuciferYang Dec 3, 2021

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not sure if using Runtime.getRuntime.availableProcessors() is really reasonable

@yaooqinn yaooqinn Dec 3, 2021

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes, as I mentioned #34792 (comment), maybe some magic number 8/16 works better for all scenarios if we choose to hardcode it

inputFormat match {
case configurable: Configurable =>
configurable.setConf(_conf)
Expand Down