Github user yhuai commented on a diff in the pull request:
https://github.com/apache/spark/pull/13444#discussion_r66683717
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/fileSourceInterfaces.scala
---
@@ -419,56 +419,79 @@ private[sql] object HadoopFsRelation extends Logging {
def listLeafFilesInParallel(
paths: Seq[Path],
hadoopConf: Configuration,
- sparkContext: SparkContext): mutable.LinkedHashSet[FileStatus] = {
- logInfo(s"Listing leaf files and directories in parallel under:
${paths.mkString(", ")}")
-
+ sparkSession: SparkSession): mutable.LinkedHashSet[FileStatus] = {
+ val sparkContext = sparkSession.sparkContext
+ val sqlConf = sparkSession.sessionState.conf
val serializableConfiguration = new
SerializableConfiguration(hadoopConf)
val serializedPaths = paths.map(_.toString)
- val fakeStatuses =
sparkContext.parallelize(serializedPaths).mapPartitions { paths =>
- // Dummy jobconf to get to the pathFilter defined in configuration
- // It's very expensive to create a
JobConf(ClassUtil.findContainingJar() is slow)
- val jobConf = new JobConf(serializableConfiguration.value,
this.getClass)
- val pathFilter = FileInputFormat.getInputPathFilter(jobConf)
- paths.map(new Path(_)).flatMap { path =>
- val fs = path.getFileSystem(serializableConfiguration.value)
- Try(listLeafFiles(fs, fs.getFileStatus(path),
pathFilter)).getOrElse(Array.empty)
- }
- }.map { status =>
- val blockLocations = status match {
- case f: LocatedFileStatus =>
- f.getBlockLocations.map { loc =>
- FakeBlockLocation(
- loc.getNames,
- loc.getHosts,
- loc.getOffset,
- loc.getLength)
- }
-
- case _ =>
- Array.empty[FakeBlockLocation]
+ // Set the number of parallelism to prevent following file listing
from generating many tasks
+ // in case of large #defaultParallelism.
+ val numParallelism = paths.size /
Math.max(sqlConf.parallelPartitionDiscoveryThreshold, 1) + 1
--- End diff --
For now, how about we just add `val numParallelism = Math.min(paths.size,
10000)` (we set the upper bound to 10000 tasks).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
