Merge branch 'master' of github.com:apache/spark into concurrent-sql-executions

Conflicts:
	core/src/test/scala/org/apache/spark/ThreadingSuite.scala

Author: Andrew Or

core/src/main/scala/org/apache/spark/annotation/AlphaComponent.java renamed to core/src/main/java/org/apache/spark/annotation/AlphaComponent.java

@@ -368,7 +368,7 @@ private[ui] class StagePage(parent: StagesTab) extends WebUIPage("stage") {
           val peakExecutionMemory = validTasks.map { case TaskUIData(info, _, _) =>
             info.accumulables
               .find { acc => acc.name == InternalAccumulator.PEAK_EXECUTION_MEMORY }
-              .map { acc => acc.value.toLong }
+              .map { acc => acc.update.getOrElse("0").toLong }
               .getOrElse(0L)
               .toDouble
           }

core/src/main/scala/org/apache/spark/annotation/DeveloperApi.java renamed to core/src/main/java/org/apache/spark/annotation/DeveloperApi.java

@@ -147,12 +147,12 @@ class ThreadingSuite extends SparkFunSuite with LocalSparkContext with Logging {
       }.start()
     }
     sem.acquire(2)
+    throwable.foreach { t => throw improveStackTrace(t) }
     if (ThreadingSuiteState.failed.get()) {
       logError("Waited 1 second without seeing runningThreads = 4 (it was " +
                 ThreadingSuiteState.runningThreads.get() + "); failing test")
       fail("One or more threads didn't see runningThreads = 4")
     }
-    throwable.foreach { t => throw improveStackTrace(t) }
   }
 
   test("set local properties in different thread") {
@@ -178,8 +178,8 @@ class ThreadingSuite extends SparkFunSuite with LocalSparkContext with Logging {
     threads.foreach(_.start())
 
     sem.acquire(5)
-    assert(sc.getLocalProperty("test") === null)
     throwable.foreach { t => throw improveStackTrace(t) }
+    assert(sc.getLocalProperty("test") === null)
   }
 
   test("set and get local properties in parent-children thread") {
@@ -207,9 +207,9 @@ class ThreadingSuite extends SparkFunSuite with LocalSparkContext with Logging {
     threads.foreach(_.start())
 
     sem.acquire(5)
+    throwable.foreach { t => throw improveStackTrace(t) }
     assert(sc.getLocalProperty("test") === "parent")
     assert(sc.getLocalProperty("Foo") === null)
-    throwable.foreach { t => throw improveStackTrace(t) }
   }
 
   test("mutation in parent local property does not affect child (SPARK-10563)") {

core/src/main/scala/org/apache/spark/annotation/Experimental.java renamed to core/src/main/java/org/apache/spark/annotation/Experimental.java

@@ -23,7 +23,7 @@ import scala.xml.Node
 
 import org.mockito.Mockito.{mock, when, RETURNS_SMART_NULLS}
 
-import org.apache.spark.{LocalSparkContext, SparkConf, SparkFunSuite, Success}
+import org.apache.spark._
 import org.apache.spark.executor.TaskMetrics
 import org.apache.spark.scheduler._
 import org.apache.spark.ui.jobs.{JobProgressListener, StagePage, StagesTab}
@@ -47,6 +47,14 @@ class StagePageSuite extends SparkFunSuite with LocalSparkContext {
     assert(html3.contains(targetString))
   }
 
+  test("SPARK-10543: peak execution memory should be per-task rather than cumulative") {
+    val unsafeConf = "spark.sql.unsafe.enabled"
+    val conf = new SparkConf(false).set(unsafeConf, "true")
+    val html = renderStagePage(conf).toString().toLowerCase
+    // verify min/25/50/75/max show task value not cumulative values
+    assert(html.contains("<td>10.0 b</td>" * 5))
+  }
+
   /**
    * Render a stage page started with the given conf and return the HTML.
    * This also runs a dummy stage to populate the page with useful content.
@@ -67,12 +75,19 @@ class StagePageSuite extends SparkFunSuite with LocalSparkContext {
 
     // Simulate a stage in job progress listener
     val stageInfo = new StageInfo(0, 0, "dummy", 1, Seq.empty, Seq.empty, "details")
-    val taskInfo = new TaskInfo(0, 0, 0, 0, "0", "localhost", TaskLocality.ANY, false)
-    jobListener.onStageSubmitted(SparkListenerStageSubmitted(stageInfo))
-    jobListener.onTaskStart(SparkListenerTaskStart(0, 0, taskInfo))
-    taskInfo.markSuccessful()
-    jobListener.onTaskEnd(
-      SparkListenerTaskEnd(0, 0, "result", Success, taskInfo, TaskMetrics.empty))
+    // Simulate two tasks to test PEAK_EXECUTION_MEMORY correctness
+    (1 to 2).foreach {
+      taskId =>
+        val taskInfo = new TaskInfo(taskId, taskId, 0, 0, "0", "localhost", TaskLocality.ANY, false)
+        val peakExecutionMemory = 10
+        taskInfo.accumulables += new AccumulableInfo(0, InternalAccumulator.PEAK_EXECUTION_MEMORY,
+          Some(peakExecutionMemory.toString), (peakExecutionMemory * taskId).toString, true)
+        jobListener.onStageSubmitted(SparkListenerStageSubmitted(stageInfo))
+        jobListener.onTaskStart(SparkListenerTaskStart(0, 0, taskInfo))
+        taskInfo.markSuccessful()
+        jobListener.onTaskEnd(
+          SparkListenerTaskEnd(0, 0, "result", Success, taskInfo, TaskMetrics.empty))
+    }
     jobListener.onStageCompleted(SparkListenerStageCompleted(stageInfo))
     page.render(request)
   }

core/src/main/scala/org/apache/spark/annotation/Private.java renamed to core/src/main/java/org/apache/spark/annotation/Private.java

@@ -35,7 +35,8 @@ object Main extends Logging {
   s.processArguments(List("-Yrepl-class-based",
     "-Yrepl-outdir", s"${outputDir.getAbsolutePath}",
     "-classpath", getAddedJars.mkString(File.pathSeparator)), true)
-  val classServer = new HttpServer(conf, outputDir, new SecurityManager(conf))
+  // the creation of SecurityManager has to be lazy so SPARK_YARN_MODE is set if needed
+  lazy val classServer = new HttpServer(conf, outputDir, new SecurityManager(conf))
   var sparkContext: SparkContext = _
   var sqlContext: SQLContext = _
   var interp = new SparkILoop // this is a public var because tests reset it.