Skip to content

[SPARK-24166][SQL] InMemoryTableScanExec should not access SQLConf at executor side #21223

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
cloud-fan wants to merge 1 commit into from
Closed

[SPARK-24166][SQL] InMemoryTableScanExec should not access SQLConf at executor side #21223

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
11 changes: 8 additions & 3 deletions sql/core/src/main/scala/org/apache/spark/sql/execution/columnar/InMemoryTableScanExec.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,10 +78,12 @@ case class InMemoryTableScanExec(

private lazy val columnarBatchSchema = new StructType(columnIndices.map(i => relationSchema(i)))

private def createAndDecompressColumn(cachedColumnarBatch: CachedBatch): ColumnarBatch = {
private def createAndDecompressColumn(
cachedColumnarBatch: CachedBatch,
offHeapColumnVectorEnabled: Boolean): ColumnarBatch = {
val rowCount = cachedColumnarBatch.numRows
val taskContext = Option(TaskContext.get())
val columnVectors = if (!conf.offHeapColumnVectorEnabled || taskContext.isEmpty) {
val columnVectors = if (!offHeapColumnVectorEnabled || taskContext.isEmpty) {
OnHeapColumnVector.allocateColumns(rowCount, columnarBatchSchema)
} else {
OffHeapColumnVector.allocateColumns(rowCount, columnarBatchSchema)
Expand All @@ -101,10 +103,13 @@ case class InMemoryTableScanExec(

private lazy val inputRDD: RDD[InternalRow] = {
val buffers = filteredCachedBatches()
val offHeapColumnVectorEnabled = conf.offHeapColumnVectorEnabled
if (supportsBatch) {
// HACK ALERT: This is actually an RDD[ColumnarBatch].
// We're taking advantage of Scala's type erasure here to pass these batches along.
buffers.map(createAndDecompressColumn).asInstanceOf[RDD[InternalRow]]
buffers
.map(createAndDecompressColumn(_, offHeapColumnVectorEnabled))
.asInstanceOf[RDD[InternalRow]]
} else {
val numOutputRows = longMetric("numOutputRows")

Expand Down