Skip to content
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

Jump to bottom

[SPARK-35215][SQL] Update custom metric per certain rows and at the end of the task #32330

Closed
wants to merge 4 commits into from
Closed

[SPARK-35215][SQL] Update custom metric per certain rows and at the end of the task #32330

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
1a98660
Update metric per certain rows and at the end of the task.
viirya Apr 25, 2021
31262d7
Merge remote-tracking branch 'upstream/master' into metric-update
viirya May 4, 2021
4f0be6c
For review comment.
viirya May 4, 2021
5f6cf5a
For review comment.
viirya May 6, 2021
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
19 changes: 12 additions & 7 deletions sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/v2/DataSourceRDD.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@ import org.apache.spark.rdd.RDD
import org.apache.spark.sql.catalyst.InternalRow
import org.apache.spark.sql.connector.read.{InputPartition, PartitionReader, PartitionReaderFactory}
import org.apache.spark.sql.errors.QueryExecutionErrors
import org.apache.spark.sql.execution.metric.SQLMetric
import org.apache.spark.sql.execution.metric.{CustomMetrics, SQLMetric}
import org.apache.spark.sql.vectorized.ColumnarBatch

class DataSourceRDDPartition(val index: Int, val inputPartition: InputPartition)
Expand Down Expand Up @@ -66,7 +66,12 @@ class DataSourceRDD(
new PartitionIterator[InternalRow](rowReader, customMetrics))
(iter, rowReader)
}
context.addTaskCompletionListener[Unit](_ => reader.close())
context.addTaskCompletionListener[Unit] { _ =>
// In case of early stopping before consuming the entire iterator,
// we need to do one more metric update at the end of the task.
CustomMetrics.updateMetrics(reader.currentMetricsValues, customMetrics)
reader.close()
}
// TODO: SPARK-25083 remove the type erasure hack in data source scan
new InterruptibleIterator(context, iter.asInstanceOf[Iterator[InternalRow]])
}
Expand All @@ -81,6 +86,8 @@ private class PartitionIterator[T](
customMetrics: Map[String, SQLMetric]) extends Iterator[T] {
private[this] var valuePrepared = false

private var numRow = 0L

override def hasNext: Boolean = {
if (!valuePrepared) {
valuePrepared = reader.next()
Expand All @@ -92,12 +99,10 @@ private class PartitionIterator[T](
if (!hasNext) {
throw QueryExecutionErrors.endOfStreamError()
}
reader.currentMetricsValues.foreach { metric =>
assert(customMetrics.contains(metric.name()),
s"Custom metrics ${customMetrics.keys.mkString(", ")} do not contain the metric " +
s"${metric.name()}")
customMetrics(metric.name()).set(metric.value())
if (numRow % CustomMetrics.numRowsPerUpdate == 0) {
CustomMetrics.updateMetrics(reader.currentMetricsValues, customMetrics)
}
numRow += 1
valuePrepared = false
reader.get()
}
Expand Down
15 changes: 14 additions & 1 deletion sql/core/src/main/scala/org/apache/spark/sql/execution/metric/CustomMetrics.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,11 +17,13 @@

package org.apache.spark.sql.execution.metric

import org.apache.spark.sql.connector.metric.CustomMetric
import org.apache.spark.sql.connector.metric.{CustomMetric, CustomTaskMetric}

object CustomMetrics {
private[spark] val V2_CUSTOM = "v2Custom"

private[spark] val numRowsPerUpdate = 100
Copy link
Contributor

Choose a reason for hiding this comment

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

nit: NUM_ROWS_PER_UPDATE since it's a constant?


/**
* Given a class name, builds and returns a metric type for a V2 custom metric class
* `CustomMetric`.
Expand All @@ -41,4 +43,15 @@ object CustomMetrics {
None
}
}

/**
* Updates given custom metrics.
*/
def updateMetrics(
currentMetricsValues: Seq[CustomTaskMetric],
customMetrics: Map[String, SQLMetric]): Unit = {
currentMetricsValues.foreach { metric =>
customMetrics(metric.name()).set(metric.value())
}
}
}
9 changes: 6 additions & 3 deletions ...n/scala/org/apache/spark/sql/execution/streaming/continuous/ContinuousDataSourceRDD.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ import org.apache.spark.rdd.RDD
import org.apache.spark.sql.catalyst.InternalRow
import org.apache.spark.sql.connector.read.InputPartition
import org.apache.spark.sql.connector.read.streaming.ContinuousPartitionReaderFactory
import org.apache.spark.sql.execution.metric.SQLMetric
import org.apache.spark.sql.execution.metric.{CustomMetrics, SQLMetric}
import org.apache.spark.sql.types.StructType
import org.apache.spark.util.NextIterator

Expand Down Expand Up @@ -92,10 +92,13 @@ class ContinuousDataSourceRDD(

val partitionReader = readerForPartition.getPartitionReader()
new NextIterator[InternalRow] {
private var numRow = 0L

override def getNext(): InternalRow = {
partitionReader.currentMetricsValues.foreach { metric =>
customMetrics(metric.name()).set(metric.value())
if (numRow % CustomMetrics.numRowsPerUpdate == 0) {
CustomMetrics.updateMetrics(partitionReader.currentMetricsValues, customMetrics)
}
numRow += 1
readerForPartition.next() match {
case null =>
finished = true
Expand Down