[SPARK-14363] Executor OOM due to a memory leak in Sorter

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.6.1
• Fix Version/s: 1.6.2, 2.0.0
• Component/s: Shuffle, Spark Core
• Assignee: Sital Kedia
• Reporter: Sital Kedia
• Created: 04/Apr/16 05:54
• Updated: 17/May/20 18:30
• Resolved: 12/Apr/16 23:13

Description

While running a Spark job, we see that the job fails because of executor OOM with following stack trace -

java.lang.OutOfMemoryError: Unable to acquire 76 bytes of memory, got 0
	at org.apache.spark.memory.MemoryConsumer.allocatePage(MemoryConsumer.java:120)
	at org.apache.spark.util.collection.unsafe.sort.UnsafeExternalSorter.acquireNewPageIfNecessary(UnsafeExternalSorter.java:326)
	at org.apache.spark.util.collection.unsafe.sort.UnsafeExternalSorter.insertRecord(UnsafeExternalSorter.java:341)
	at org.apache.spark.sql.execution.UnsafeExternalRowSorter.insertRow(UnsafeExternalRowSorter.java:91)
	at org.apache.spark.sql.execution.UnsafeExternalRowSorter.sort(UnsafeExternalRowSorter.java:168)
	at org.apache.spark.sql.execution.Sort$$anonfun$1.apply(Sort.scala:90)
	at org.apache.spark.sql.execution.Sort$$anonfun$1.apply(Sort.scala:64)
	at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$21.apply(RDD.scala:728)
	at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$21.apply(RDD.scala:728)
	at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
	at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
	at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
	at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
	at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
	at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
	at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
	at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
	at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
	at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
	at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
	at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
	at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
	at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:306)
	at org.apache.spark.rdd.RDD.iterator(RDD.scala:270)
	at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:66)
	at org.apache.spark.scheduler.Task.run(Task.scala:89)
	at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:214)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at java.lang.Thread.run(Thread.java:745)

The issue is that there is a memory leak in the Sorter. When the UnsafeExternalSorter spills the data to disk, it does not free up the underlying pointer array. As a result, we see a lot of executor OOM and also memory under utilization.

This is a regression partially introduced in PR https://github.com/apache/spark/pull/9241

Issue Links

• relates to: SPARK-18181 Huge managed memory leak (2.7G) when running reduceByKey
• links to: [Github] Pull Request #12285 (sitalkedia)