Address code review comments, change event logging to stage end.

Author: edwinalu

core/src/main/scala/org/apache/spark/executor/Executor.scala

@@ -19,12 +19,13 @@ package org.apache.spark.executor
 
 import java.io.{File, NotSerializableException}
 import java.lang.Thread.UncaughtExceptionHandler
-import java.lang.management.ManagementFactory
+import java.lang.management.{BufferPoolMXBean, ManagementFactory}
 import java.net.{URI, URL}
 import java.nio.ByteBuffer
 import java.util.Properties
 import java.util.concurrent._
 import javax.annotation.concurrent.GuardedBy
+import javax.management.ObjectName
 
 import scala.collection.JavaConverters._
 import scala.collection.mutable.{ArrayBuffer, HashMap, Map}
@@ -36,7 +37,7 @@ import org.apache.spark._
 import org.apache.spark.deploy.SparkHadoopUtil
 import org.apache.spark.internal.Logging
 import org.apache.spark.internal.config._
-import org.apache.spark.memory.{SparkOutOfMemoryError, TaskMemoryManager}
+import org.apache.spark.memory.{MemoryManager, SparkOutOfMemoryError, TaskMemoryManager}
 import org.apache.spark.rpc.RpcTimeout
 import org.apache.spark.scheduler.{DirectTaskResult, IndirectTaskResult, Task, TaskDescription}
 import org.apache.spark.shuffle.FetchFailedException
@@ -71,6 +72,12 @@ private[spark] class Executor(
 
   private val conf = env.conf
 
+  // BufferPoolMXBean for direct memory
+  private val directBufferPool = Executor.getBufferPool(Executor.DIRECT_BUFFER_POOL_NAME)
+
+  // BufferPoolMXBean for mapped memory
+  private val mappedBufferPool = Executor.getBufferPool(Executor.MAPPED_BUFFER_POOL_NAME)
+
   // No ip or host:port - just hostname
   Utils.checkHost(executorHostname)
   // must not have port specified.
@@ -773,10 +780,8 @@ private[spark] class Executor(
     val curGCTime = computeTotalGcTime()
 
     // get executor level memory metrics
-    val executorUpdates = new ExecutorMetrics(System.currentTimeMillis(),
-      ManagementFactory.getMemoryMXBean.getHeapMemoryUsage().getUsed(),
-      env.memoryManager.onHeapExecutionMemoryUsed, env.memoryManager.offHeapExecutionMemoryUsed,
-      env.memoryManager.onHeapStorageMemoryUsed, env.memoryManager.offHeapStorageMemoryUsed)
+    val executorUpdates = Executor.getCurrentExecutorMetrics(env.memoryManager,
+      directBufferPool, mappedBufferPool)
 
     for (taskRunner <- runningTasks.values().asScala) {
       if (taskRunner.task != null) {
@@ -814,4 +819,43 @@ private[spark] object Executor {
   // task is fully deserialized. When possible, the TaskContext.getLocalProperty call should be
   // used instead.
   val taskDeserializationProps: ThreadLocal[Properties] = new ThreadLocal[Properties]
+
+  val DIRECT_BUFFER_POOL_NAME = "direct"
+  val MAPPED_BUFFER_POOL_NAME = "mapped"
+
+  /** Get the BufferPoolMXBean for the specified buffer pool. */
+  def getBufferPool(pool: String): BufferPoolMXBean = {
+    val name = new ObjectName("java.nio:type=BufferPool,name=" + pool)
+    ManagementFactory.newPlatformMXBeanProxy(ManagementFactory.getPlatformMBeanServer,
+      name.toString, classOf[BufferPoolMXBean])
+  }
+
+  /**
+   * Get the current executor level memory metrics.
+   *
+   * @param memoryManager the memory manager
+   * @param direct the direct memory buffer pool
+   * @param mapped the mapped memory buffer pool
+   * @return the executor memory metrics
+   */
+  def getCurrentExecutorMetrics(
+      memoryManager: MemoryManager,
+      direct: BufferPoolMXBean,
+      mapped: BufferPoolMXBean) : ExecutorMetrics = {
+    val onHeapExecutionMemoryUsed = memoryManager.onHeapExecutionMemoryUsed
+    val offHeapExecutionMemoryUsed = memoryManager.offHeapExecutionMemoryUsed
+    val onHeapStorageMemoryUsed = memoryManager.onHeapStorageMemoryUsed
+    val offHeapStorageMemoryUsed = memoryManager.offHeapStorageMemoryUsed
+    new ExecutorMetrics(System.currentTimeMillis(),
+      ManagementFactory.getMemoryMXBean.getHeapMemoryUsage().getUsed(),
+      ManagementFactory.getMemoryMXBean.getNonHeapMemoryUsage().getUsed(),
+      onHeapExecutionMemoryUsed,
+      offHeapExecutionMemoryUsed,
+      onHeapStorageMemoryUsed,
+      offHeapStorageMemoryUsed,
+      onHeapExecutionMemoryUsed + onHeapStorageMemoryUsed, // on heap unified memory
+      offHeapExecutionMemoryUsed + offHeapStorageMemoryUsed, // off heap unified memory
+      direct.getMemoryUsed,
+      mapped.getMemoryUsed)
+  }
 }

core/src/main/scala/org/apache/spark/executor/ExecutorMetrics.scala

@@ -26,18 +26,29 @@ import org.apache.spark.annotation.DeveloperApi
  * This is sent to the driver periodically (on executor heartbeat), to provide
  * information about each executor's metrics.
  *
- * @param timestamp the time the metrics were collected
- * @param jvmUsedMemory the amount of JVM used memory for the executor
+ * @param timestamp the time the metrics were collected, or -1 for Spark history
+ *                  log events which are logged when a stage has completed
+ * @param jvmUsedHeapMemory the amount of JVM used heap memory for the executor
+ * @param jvmUsedNonHeapMemory the amount of JVM used non-heap memory for the executor
  * @param onHeapExecutionMemory the amount of on heap execution memory used
  * @param offHeapExecutionMemory the amount of off heap execution memory used
  * @param onHeapStorageMemory the amount of on heap storage memory used
  * @param offHeapStorageMemory the amount of off heap storage memory used
+ * @param onHeapUnifiedMemory the amount of on heap unified region memory used
+ * @param offHeapUnifiedMemory the amount of off heap unified region memory used
+ * @param directMemory the amount of direct memory used
+ * @param mappedMemory the amount of mapped memory used
  */
 @DeveloperApi
 class ExecutorMetrics private[spark] (
     val timestamp: Long,
-    val jvmUsedMemory: Long,
+    val jvmUsedHeapMemory: Long,
+    val jvmUsedNonHeapMemory: Long,
     val onHeapExecutionMemory: Long,
     val offHeapExecutionMemory: Long,
     val onHeapStorageMemory: Long,
-    val offHeapStorageMemory: Long) extends Serializable
+    val offHeapStorageMemory: Long,
+    val onHeapUnifiedMemory: Long,
+    val offHeapUnifiedMemory: Long,
+    val directMemory: Long,
+    val mappedMemory: Long) extends Serializable

core/src/main/scala/org/apache/spark/scheduler/DAGScheduler.scala

@@ -35,7 +35,7 @@ import org.apache.commons.lang3.SerializationUtils
 
 import org.apache.spark._
 import org.apache.spark.broadcast.Broadcast
-import org.apache.spark.executor.{ExecutorMetrics, TaskMetrics}
+import org.apache.spark.executor.{Executor, ExecutorMetrics, TaskMetrics}
 import org.apache.spark.internal.Logging
 import org.apache.spark.internal.config
 import org.apache.spark.network.util.JavaUtils
@@ -214,6 +214,12 @@ class DAGScheduler(
   private val heartbeater: Heartbeater = new Heartbeater(reportHeartBeat,
     sc.conf.getTimeAsMs("spark.executor.heartbeatInterval", "10s"))
 
+  /** BufferPoolMXBean for direct memory */
+  private val directBufferPool = Executor.getBufferPool(Executor.DIRECT_BUFFER_POOL_NAME)
+
+  /** BufferPoolMXBean for mapped memory */
+  private val mappedBufferPool = Executor.getBufferPool(Executor.MAPPED_BUFFER_POOL_NAME)
+
   /**
    * Called by the TaskSetManager to report task's starting.
    */
@@ -1766,12 +1772,8 @@ class DAGScheduler(
   /** Reports heartbeat metrics for the driver. */
   private def reportHeartBeat(): Unit = {
     // get driver memory metrics
-    val driverUpdates = new ExecutorMetrics(System.currentTimeMillis(),
-      ManagementFactory.getMemoryMXBean.getHeapMemoryUsage().getUsed(),
-      sc.env.memoryManager.onHeapExecutionMemoryUsed,
-      sc.env.memoryManager.offHeapExecutionMemoryUsed,
-      sc.env.memoryManager.onHeapStorageMemoryUsed,
-      sc.env.memoryManager.offHeapStorageMemoryUsed)
+    val driverUpdates = Executor.getCurrentExecutorMetrics(
+      sc.env.memoryManager, directBufferPool, mappedBufferPool)
     val accumUpdates = new Array[(Long, Int, Int, Seq[AccumulableInfo])](0)
     listenerBus.post(SparkListenerExecutorMetricsUpdate("driver", accumUpdates,
       Some(driverUpdates)))

core/src/main/scala/org/apache/spark/scheduler/EventLoggingListener.scala

@@ -94,8 +94,9 @@ private[spark] class EventLoggingListener(
   // Visible for tests only.
   private[scheduler] val logPath = getLogPath(logBaseDir, appId, appAttemptId, compressionCodecName)
 
-  // Peak metric values for each executor
-  private var peakExecutorMetrics = new mutable.HashMap[String, PeakExecutorMetrics]()
+  // map of live stages, to peak executor metrics for the stage
+  private val liveStageExecutorMetrics = mutable.HashMap[(Int, Int),
+    mutable.HashMap[String, PeakExecutorMetrics]]()
 
   /**
    * Creates the log file in the configured log directory.
@@ -162,7 +163,8 @@ private[spark] class EventLoggingListener(
   override def onStageSubmitted(event: SparkListenerStageSubmitted): Unit = {
     logEvent(event)
     // clear the peak metrics when a new stage starts
-    peakExecutorMetrics.values.foreach(_.reset())
+    liveStageExecutorMetrics.put((event.stageInfo.stageId, event.stageInfo.attemptNumber()),
+      new mutable.HashMap[String, PeakExecutorMetrics]())
   }
 
   override def onTaskStart(event: SparkListenerTaskStart): Unit = logEvent(event)
@@ -177,6 +179,27 @@ private[spark] class EventLoggingListener(
 
   // Events that trigger a flush
   override def onStageCompleted(event: SparkListenerStageCompleted): Unit = {
+    // log the peak executor metrics for the stage, for each executor
+    val accumUpdates = new ArrayBuffer[(Long, Int, Int, Seq[AccumulableInfo])]()
+    val executorMap = liveStageExecutorMetrics.remove(
+      (event.stageInfo.stageId, event.stageInfo.attemptNumber()))
+    executorMap.foreach {
+      executorEntry => {
+        for ((executorId, peakExecutorMetrics) <- executorEntry) {
+          val executorMetrics = new ExecutorMetrics(-1, peakExecutorMetrics.jvmUsedHeapMemory,
+            peakExecutorMetrics.jvmUsedNonHeapMemory, peakExecutorMetrics.onHeapExecutionMemory,
+            peakExecutorMetrics.offHeapExecutionMemory, peakExecutorMetrics.onHeapStorageMemory,
+            peakExecutorMetrics.offHeapStorageMemory, peakExecutorMetrics.onHeapUnifiedMemory,
+            peakExecutorMetrics.offHeapUnifiedMemory, peakExecutorMetrics.directMemory,
+            peakExecutorMetrics.mappedMemory)
+          val executorUpdate = new SparkListenerExecutorMetricsUpdate(
+            executorId, accumUpdates, Some(executorMetrics))
+          logEvent(executorUpdate)
+        }
+      }
+    }
+
+    // log stage completed event
     logEvent(event, flushLogger = true)
   }
 
@@ -205,12 +228,10 @@ private[spark] class EventLoggingListener(
   }
   override def onExecutorAdded(event: SparkListenerExecutorAdded): Unit = {
     logEvent(event, flushLogger = true)
-    peakExecutorMetrics.put(event.executorId, new PeakExecutorMetrics())
   }
 
   override def onExecutorRemoved(event: SparkListenerExecutorRemoved): Unit = {
     logEvent(event, flushLogger = true)
-    peakExecutorMetrics.remove(event.executorId)
   }
 
   override def onExecutorBlacklisted(event: SparkListenerExecutorBlacklisted): Unit = {
@@ -244,19 +265,13 @@ private[spark] class EventLoggingListener(
     }
   }
 
-  /**
-   * Log if there is a new peak value for one of the memory metrics for the given executor.
-   * Metrics are cleared out when a new stage is started in onStageSubmitted, so this will
-   * log new peak memory metric values per executor per stage.
-   */
   override def onExecutorMetricsUpdate(event: SparkListenerExecutorMetricsUpdate): Unit = {
-    var log: Boolean = false
+    // For the active stages, record any new peak values for the memory metrics for the executor
     event.executorUpdates.foreach { executorUpdates =>
-      val peakMetrics = peakExecutorMetrics.getOrElseUpdate(event.execId, new PeakExecutorMetrics())
-      if (peakMetrics.compareAndUpdate(executorUpdates)) {
-        val accumUpdates = new ArrayBuffer[(Long, Int, Int, Seq[AccumulableInfo])]()
-        logEvent(new SparkListenerExecutorMetricsUpdate(event.execId, accumUpdates,
-          event.executorUpdates), flushLogger = true)
+      liveStageExecutorMetrics.values.foreach { peakExecutorMetrics =>
+        val peakMetrics = peakExecutorMetrics.getOrElseUpdate(
+          event.execId, new PeakExecutorMetrics())
+        peakMetrics.compareAndUpdate(executorUpdates)
       }
     }
   }

core/src/main/scala/org/apache/spark/scheduler/PeakExecutorMetrics.scala

@@ -21,17 +21,40 @@ import org.apache.spark.executor.ExecutorMetrics
 import org.apache.spark.status.api.v1.PeakMemoryMetrics
 
 /**
- * Records the peak values for executor level metrics. If jvmUsedMemory is -1, then no values have
- * been recorded yet.
+ * Records the peak values for executor level metrics. If jvmUsedHeapMemory is -1, then no
+ * values have been recorded yet.
  */
 private[spark] class PeakExecutorMetrics {
-  private var jvmUsedMemory = -1L;
-  private var onHeapExecutionMemory = 0L
-  private var offHeapExecutionMemory = 0L
-  private var onHeapStorageMemory = 0L
-  private var offHeapStorageMemory = 0L
-  private var onHeapUnifiedMemory = 0L
-  private var offHeapUnifiedMemory = 0L
+  private var _jvmUsedHeapMemory = -1L;
+  private var _jvmUsedNonHeapMemory = 0L;
+  private var _onHeapExecutionMemory = 0L
+  private var _offHeapExecutionMemory = 0L
+  private var _onHeapStorageMemory = 0L
+  private var _offHeapStorageMemory = 0L
+  private var _onHeapUnifiedMemory = 0L
+  private var _offHeapUnifiedMemory = 0L
+  private var _directMemory = 0L
+  private var _mappedMemory = 0L
+
+  def jvmUsedHeapMemory: Long = _jvmUsedHeapMemory
+
+  def jvmUsedNonHeapMemory: Long = _jvmUsedNonHeapMemory
+
+  def onHeapExecutionMemory: Long = _onHeapExecutionMemory
+
+  def offHeapExecutionMemory: Long = _offHeapExecutionMemory
+
+  def onHeapStorageMemory: Long = _onHeapStorageMemory
+
+  def offHeapStorageMemory: Long = _offHeapStorageMemory
+
+  def onHeapUnifiedMemory: Long = _onHeapUnifiedMemory
+
+  def offHeapUnifiedMemory: Long = _offHeapUnifiedMemory
+
+  def directMemory: Long = _directMemory
+
+  def mappedMemory: Long = _mappedMemory
 
   /**
    * Compare the specified memory values with the saved peak executor memory
@@ -43,36 +66,44 @@ private[spark] class PeakExecutorMetrics {
   def compareAndUpdate(executorMetrics: ExecutorMetrics): Boolean = {
     var updated: Boolean = false
 
-    if (executorMetrics.jvmUsedMemory > jvmUsedMemory) {
-      jvmUsedMemory = executorMetrics.jvmUsedMemory
+    if (executorMetrics.jvmUsedHeapMemory > _jvmUsedHeapMemory) {
+      _jvmUsedHeapMemory = executorMetrics.jvmUsedHeapMemory
+      updated = true
+    }
+    if (executorMetrics.jvmUsedNonHeapMemory > _jvmUsedNonHeapMemory) {
+      _jvmUsedNonHeapMemory = executorMetrics.jvmUsedNonHeapMemory
+      updated = true
+    }
+    if (executorMetrics.onHeapExecutionMemory > _onHeapExecutionMemory) {
+      _onHeapExecutionMemory = executorMetrics.onHeapExecutionMemory
       updated = true
     }
-    if (executorMetrics.onHeapExecutionMemory > onHeapExecutionMemory) {
-      onHeapExecutionMemory = executorMetrics.onHeapExecutionMemory
+    if (executorMetrics.offHeapExecutionMemory > _offHeapExecutionMemory) {
+      _offHeapExecutionMemory = executorMetrics.offHeapExecutionMemory
       updated = true
     }
-    if (executorMetrics.offHeapExecutionMemory > offHeapExecutionMemory) {
-      offHeapExecutionMemory = executorMetrics.offHeapExecutionMemory
+    if (executorMetrics.onHeapStorageMemory > _onHeapStorageMemory) {
+      _onHeapStorageMemory = executorMetrics.onHeapStorageMemory
       updated = true
     }
-    if (executorMetrics.onHeapStorageMemory > onHeapStorageMemory) {
-      onHeapStorageMemory = executorMetrics.onHeapStorageMemory
+    if (executorMetrics.offHeapStorageMemory > _offHeapStorageMemory) {
+      _offHeapStorageMemory = executorMetrics.offHeapStorageMemory
       updated = true
     }
-    if (executorMetrics.offHeapStorageMemory > offHeapStorageMemory) {
-      offHeapStorageMemory = executorMetrics.offHeapStorageMemory
+    if (executorMetrics.onHeapUnifiedMemory > _onHeapUnifiedMemory) {
+      _onHeapUnifiedMemory = executorMetrics.onHeapUnifiedMemory
       updated = true
     }
-    val newOnHeapUnifiedMemory = (executorMetrics.onHeapExecutionMemory +
-      executorMetrics.onHeapStorageMemory)
-    if (newOnHeapUnifiedMemory > onHeapUnifiedMemory) {
-      onHeapUnifiedMemory = newOnHeapUnifiedMemory
+    if (executorMetrics.offHeapUnifiedMemory > _offHeapUnifiedMemory) {
+      _offHeapUnifiedMemory = executorMetrics.offHeapUnifiedMemory
       updated = true
     }
-    val newOffHeapUnifiedMemory = (executorMetrics.offHeapExecutionMemory +
-      executorMetrics.offHeapStorageMemory)
-    if ( newOffHeapUnifiedMemory > offHeapUnifiedMemory) {
-      offHeapUnifiedMemory = newOffHeapUnifiedMemory
+    if (executorMetrics.directMemory > _directMemory) {
+      _directMemory = executorMetrics.directMemory
+      updated = true
+    }
+    if (executorMetrics.mappedMemory > _mappedMemory) {
+      _mappedMemory = executorMetrics.mappedMemory
       updated = true
     }
 
@@ -84,23 +115,13 @@ private[spark] class PeakExecutorMetrics {
    *         values set.
    */
   def getPeakMemoryMetrics: Option[PeakMemoryMetrics] = {
-    if (jvmUsedMemory < 0) {
+    if (_jvmUsedHeapMemory < 0) {
       None
     } else {
-      Some(new PeakMemoryMetrics(jvmUsedMemory, onHeapExecutionMemory,
-        offHeapExecutionMemory, onHeapStorageMemory, offHeapStorageMemory,
-        onHeapUnifiedMemory, offHeapUnifiedMemory))
+      Some(new PeakMemoryMetrics(_jvmUsedHeapMemory, _jvmUsedNonHeapMemory,
+        _onHeapExecutionMemory, _offHeapExecutionMemory, _onHeapStorageMemory,
+        _offHeapStorageMemory, _onHeapUnifiedMemory, _offHeapUnifiedMemory,
+        _directMemory, _mappedMemory))
     }
   }
-
-  /** Clears/resets the saved peak values. */
-  def reset(): Unit = {
-    jvmUsedMemory = -1L;
-    onHeapExecutionMemory = 0L
-    offHeapExecutionMemory = 0L
-    onHeapStorageMemory = 0L
-    offHeapStorageMemory = 0L
-    onHeapUnifiedMemory = 0L
-    offHeapUnifiedMemory = 0L
-  }
 }