From 60977889eaecdf28adc6164310eaa5afed488fa1 Mon Sep 17 00:00:00 2001
From: Reynold Xin <rxin@databricks.com>
Date: Wed, 29 Mar 2017 19:06:51 -0700
Subject: [PATCH] [SPARK-20136][SQL] Add num files and metadata operation
timing to scan operator metrics
## What changes were proposed in this pull request?
This patch adds explicit metadata operation timing and number of files in data source metrics. Those would be useful to include for performance profiling.
Screenshot of a UI with this change (num files and metadata time are new metrics):
<img width="321" alt="screen shot 2017-03-29 at 12 29 28 am" src="https://cloud.githubusercontent.com/assets/323388/24443272/d4ea58c0-1416-11e7-8940-ecb69375554a.png">
## How was this patch tested?
N/A
Author: Reynold Xin <rxin@databricks.com>
Closes #17465 from rxin/SPARK-20136.
---
.../sql/execution/DataSourceScanExec.scala | 18 ++++++++++++++++--
1 file changed, 16 insertions(+), 2 deletions(-)
diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
index 28156b277f..239151495f 100644
--- a/sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
+++ b/sql/core/src/main/scala/org/apache/spark/sql/execution/DataSourceScanExec.scala
@@ -171,8 +171,20 @@ case class FileSourceScanExec(
false
}
- @transient private lazy val selectedPartitions =
- relation.location.listFiles(partitionFilters, dataFilters)
+ @transient private lazy val selectedPartitions: Seq[PartitionDirectory] = {
+ val startTime = System.nanoTime()
+ val ret = relation.location.listFiles(partitionFilters, dataFilters)
+ val timeTaken = (System.nanoTime() - startTime) / 1000 / 1000
+
+ metrics("numFiles").add(ret.map(_.files.size.toLong).sum)
+ metrics("metadataTime").add(timeTaken)
+
+ val executionId = sparkContext.getLocalProperty(SQLExecution.EXECUTION_ID_KEY)
+ SQLMetrics.postDriverMetricUpdates(sparkContext, executionId,
+ metrics("numFiles") :: metrics("metadataTime") :: Nil)
+
+ ret
+ }
override val (outputPartitioning, outputOrdering): (Partitioning, Seq[SortOrder]) = {
val bucketSpec = if (relation.sparkSession.sessionState.conf.bucketingEnabled) {
@@ -293,6 +305,8 @@ case class FileSourceScanExec(
override lazy val metrics =
Map("numOutputRows" -> SQLMetrics.createMetric(sparkContext, "number of output rows"),
+ "numFiles" -> SQLMetrics.createMetric(sparkContext, "number of files"),
+ "metadataTime" -> SQLMetrics.createMetric(sparkContext, "metadata time (ms)"),
"scanTime" -> SQLMetrics.createTimingMetric(sparkContext, "scan time"))
protected override def doExecute(): RDD[InternalRow] = {
--
GitLab