Skip to content

[SPARK-29189][SQL] Add an option to ignore block locations when listing file #25869

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
wangshisan wants to merge 9 commits into from
Closed

[SPARK-29189][SQL] Add an option to ignore block locations when listing file #25869

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
e07c230
[SPARK-29189] Add an option to ignore block locations when listing file
wangshisan Sep 20, 2019
b7f9f03
add ut
wangshisan Sep 22, 2019
0da2903
fix ut
wangshisan Sep 22, 2019
a659e67
tiny fix
wangshisan Sep 23, 2019
22dacc6
Trivial fix
wangshisan Sep 24, 2019
eb1a802
fix code style
wangshisan Sep 25, 2019
475abba
Trivial fix
wangshisan Sep 26, 2019
e500bcd
remove unused import
wangshisan Sep 29, 2019
00ad219
Enhance comment
wangshisan Oct 6, 2019
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
10 changes: 10 additions & 0 deletions sql/catalyst/src/main/scala/org/apache/spark/sql/internal/SQLConf.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1979,6 +1979,14 @@ object SQLConf {
.doc("When true, the ArrayExists will follow the three-valued boolean logic.")
.booleanConf
.createWithDefault(true)

val IGNORE_DATA_LOCALITY =
Comment thread
wangshisan marked this conversation as resolved.
Outdated
buildConf("spark.sql.ignore.datalocality")
.doc("If it is set to true, Spark won't fetch the block locations for each file on " +
Comment thread
wangshisan marked this conversation as resolved.
Outdated
"listing files, which will greatly accelerate the list file operation, while loss " +
"data locality.")
.booleanConf
.createWithDefault(false)
}

/**
Expand Down Expand Up @@ -2475,6 +2483,8 @@ class SQLConf extends Serializable with Logging {

def defaultV2Catalog: Option[String] = getConf(DEFAULT_V2_CATALOG)

def ignoreDataLocality: Boolean = getConf(SQLConf.IGNORE_DATA_LOCALITY)

/** ********************** SQLConf functionality methods ************ */

/** Set Spark SQL configuration properties. */
Expand Down
7 changes: 5 additions & 2 deletions sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/InMemoryFileIndex.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -290,6 +290,7 @@ object InMemoryFileIndex extends Logging {
isRootPath: Boolean): Seq[FileStatus] = {
logTrace(s"Listing $path")
val fs = path.getFileSystem(hadoopConf)
val ignoreLocality = sessionOpt.map(_.sqlContext.conf.ignoreDataLocality).getOrElse(false)

// Note that statuses only include FileStatus for the files and dirs directly under path,
// and does not include anything else recursively.
Expand All @@ -299,7 +300,7 @@ object InMemoryFileIndex extends Logging {
// to retrieve the file status with the file block location. The reason to still fallback
// to listStatus is because the default implementation would potentially throw a
// FileNotFoundException which is better handled by doing the lookups manually below.
case _: DistributedFileSystem =>
case _: DistributedFileSystem if !ignoreLocality =>
val remoteIter = fs.listLocatedStatus(path)
new Iterator[LocatedFileStatus]() {
def next(): LocatedFileStatus = remoteIter.next
Expand Down Expand Up @@ -376,7 +377,7 @@ object InMemoryFileIndex extends Logging {
// - Here we are calling `getFileBlockLocations` in a sequential manner, but it should not
// be a big deal since we always use to `bulkListLeafFiles` when the number of
// paths exceeds threshold.
case f =>
case f if !ignoreLocality =>
// The other constructor of LocatedFileStatus will call FileStatus.getPermission(),
// which is very slow on some file system (RawLocalFileSystem, which is launch a
// subprocess and parse the stdout).
Expand All @@ -400,6 +401,8 @@ object InMemoryFileIndex extends Logging {
missingFiles += f.getPath.toString
None
}

case f => Some(f)
}

if (missingFiles.nonEmpty) {
Expand Down