Merge branch 'master' of github.com:apache/spark into dag-viz-streaming

core/src/main/resources/org/apache/spark/ui/static/spark-dag-viz.js

@@ -186,8 +186,9 @@ function renderDagVizForJob(svgContainer) {
     var stageId = metadata.attr("stage-id");
     var containerId = VizConstants.graphPrefix + stageId;
     // Link each graph to the corresponding stage page (TODO: handle stage attempts)
-    var stageLink = "/stages/stage/?id=" +
-      stageId.replace(VizConstants.stagePrefix, "") + "&attempt=0&expandDagViz=true";
+    var stageLink = $("#stage-" + stageId.replace(VizConstants.stagePrefix, "") + "-0")
+      .find("a")
+      .attr("href") + "&expandDagViz=true";
     var container = svgContainer
       .append("a")
       .attr("xlink:href", stageLink)

core/src/main/scala/org/apache/spark/SparkContext.scala

@@ -371,6 +371,14 @@ class SparkContext(config: SparkConf) extends Logging with ExecutorAllocationCli
       throw new SparkException("An application name must be set in your configuration")
     }
 
+    // System property spark.yarn.app.id must be set if user code ran by AM on a YARN cluster
+    // yarn-standalone is deprecated, but still supported
+    if ((master == "yarn-cluster" || master == "yarn-standalone") &&
+        !_conf.contains("spark.yarn.app.id")) {
+      throw new SparkException("Detected yarn-cluster mode, but isn't running on a cluster. " +
+        "Deployment to YARN is not supported directly by SparkContext. Please use spark-submit.")
+    }
+
     if (_conf.getBoolean("spark.logConf", false)) {
       logInfo("Spark configuration:\n" + _conf.toDebugString)
     }

core/src/main/scala/org/apache/spark/util/ClosureCleaner.scala

@@ -239,15 +239,6 @@ private[spark] object ClosureCleaner extends Logging {
     logDebug(s" + fields accessed by starting closure: " + accessedFields.size)
     accessedFields.foreach { f => logDebug("     " + f) }
 
-    val inInterpreter = {
-      try {
-        val interpClass = Class.forName("spark.repl.Main")
-        interpClass.getMethod("interp").invoke(null) != null
-      } catch {
-        case _: ClassNotFoundException => true
-      }
-    }
-
     // List of outer (class, object) pairs, ordered from outermost to innermost
     // Note that all outer objects but the outermost one (first one in this list) must be closures
     var outerPairs: List[(Class[_], AnyRef)] = (outerClasses zip outerObjects).reverse
@@ -274,7 +265,7 @@ private[spark] object ClosureCleaner extends Logging {
       // required fields from the original object. We need the parent here because the Java
       // language specification requires the first constructor parameter of any closure to be
       // its enclosing object.
-      val clone = instantiateClass(cls, parent, inInterpreter)
+      val clone = instantiateClass(cls, parent)
       for (fieldName <- accessedFields(cls)) {
         val field = cls.getDeclaredField(fieldName)
         field.setAccessible(true)
@@ -327,9 +318,8 @@ private[spark] object ClosureCleaner extends Logging {
 
   private def instantiateClass(
       cls: Class[_],
-      enclosingObject: AnyRef,
-      inInterpreter: Boolean): AnyRef = {
-    if (!inInterpreter) {
+      enclosingObject: AnyRef): AnyRef = {
+    if (!Utils.isInInterpreter) {
       // This is a bona fide closure class, whose constructor has no effects
       // other than to set its fields, so use its constructor
       val cons = cls.getConstructors()(0)

core/src/main/scala/org/apache/spark/util/Utils.scala

@@ -1795,6 +1795,20 @@ private[spark] object Utils extends Logging {
     }
   }
 
+  lazy val isInInterpreter: Boolean = {
+    try {
+      val interpClass = classForName("spark.repl.Main")
+      interpClass.getMethod("interp").invoke(null) != null
+    } catch {
+      // Returning true seems to be a mistake.
+      // Currently changing it to false causes tests failures in Streaming.
+      // For a more detailed discussion, please, refer to
+      // https://github.com/apache/spark/pull/5835#issuecomment-101042271 and subsequent comments.
+      // Addressing this changed is tracked as https://issues.apache.org/jira/browse/SPARK-7527
+      case _: ClassNotFoundException => true
+    }
+  }
+
   /**
    * Return a well-formed URI for the file described by a user input string.
    *

core/src/test/scala/org/apache/spark/ContextCleanerSuite.scala

@@ -165,6 +165,7 @@ class ContextCleanerSuite extends ContextCleanerSuiteBase {
     }
 
     // Test that GC causes RDD cleanup after dereferencing the RDD
+    // Note rdd is used after previous GC to avoid early collection by the JVM
     val postGCTester = new CleanerTester(sc, rddIds = Seq(rdd.id))
     rdd = null // Make RDD out of scope
     runGC()
@@ -181,9 +182,9 @@ class ContextCleanerSuite extends ContextCleanerSuiteBase {
     intercept[Exception] {
       preGCTester.assertCleanup()(timeout(1000 millis))
     }
+    rdd.count()  // Defeat early collection by the JVM
 
     // Test that GC causes shuffle cleanup after dereferencing the RDD
-    rdd.count()  // Defeat any early collection of rdd variable by the JVM
     val postGCTester = new CleanerTester(sc, shuffleIds = Seq(0))
     rdd = null  // Make RDD out of scope, so that corresponding shuffle goes out of scope
     runGC()
@@ -201,6 +202,7 @@ class ContextCleanerSuite extends ContextCleanerSuiteBase {
     }
 
     // Test that GC causes broadcast cleanup after dereferencing the broadcast variable
+    // Note broadcast is used after previous GC to avoid early collection by the JVM
     val postGCTester = new CleanerTester(sc, broadcastIds = Seq(broadcast.id))
     broadcast = null  // Make broadcast variable out of scope
     runGC()
@@ -226,7 +228,7 @@ class ContextCleanerSuite extends ContextCleanerSuiteBase {
 
     // the checkpoint is not cleaned by default (without the configuration set)
     var postGCTester = new CleanerTester(sc, Seq(rddId), Nil, Nil, Nil)
-    rdd = null // Make RDD out of scope
+    rdd = null // Make RDD out of scope, ok if collected earlier
     runGC()
     postGCTester.assertCleanup()
     assert(fs.exists(RDDCheckpointData.rddCheckpointDataPath(sc, rddId).get))
@@ -245,6 +247,9 @@ class ContextCleanerSuite extends ContextCleanerSuiteBase {
     // Confirm the checkpoint directory exists
     assert(fs.exists(RDDCheckpointData.rddCheckpointDataPath(sc, rddId).get))
 
+    // Reference rdd to defeat any early collection by the JVM
+    rdd.count()
+
     // Test that GC causes checkpoint data cleanup after dereferencing the RDD
     postGCTester = new CleanerTester(sc, Seq(rddId), Nil, Nil, Seq(rddId))
     rdd = null // Make RDD out of scope
@@ -352,6 +357,7 @@ class SortShuffleContextCleanerSuite extends ContextCleanerSuiteBase(classOf[Sor
     intercept[Exception] {
       preGCTester.assertCleanup()(timeout(1000 millis))
     }
+    rdd.count()  // Defeat early collection by the JVM
 
     // Test that GC causes shuffle cleanup after dereferencing the RDD
     val postGCTester = new CleanerTester(sc, shuffleIds = Seq(0))

mllib/src/main/scala/org/apache/spark/mllib/linalg/Matrices.scala

@@ -273,7 +273,8 @@ class DenseMatrix(
 
   override def copy: DenseMatrix = new DenseMatrix(numRows, numCols, values.clone())
 
-  private[mllib] def map(f: Double => Double) = new DenseMatrix(numRows, numCols, values.map(f))
+  private[mllib] def map(f: Double => Double) = new DenseMatrix(numRows, numCols, values.map(f),
+    isTransposed)
 
   private[mllib] def update(f: Double => Double): DenseMatrix = {
     val len = values.length
@@ -535,7 +536,7 @@ class SparseMatrix(
   }
 
   private[mllib] def map(f: Double => Double) =
-    new SparseMatrix(numRows, numCols, colPtrs, rowIndices, values.map(f))
+    new SparseMatrix(numRows, numCols, colPtrs, rowIndices, values.map(f), isTransposed)
 
   private[mllib] def update(f: Double => Double): SparseMatrix = {
     val len = values.length

python/pyspark/mllib/clustering.py

@@ -212,6 +212,9 @@ def predict(self, x):
         if isinstance(x, RDD):
             cluster_labels = self.predictSoft(x).map(lambda z: z.index(max(z)))
             return cluster_labels
+        else:
+            raise TypeError("x should be represented by an RDD, "
+                            "but got %s." % type(x))
 
     def predictSoft(self, x):
         """
@@ -225,6 +228,9 @@ def predictSoft(self, x):
             membership_matrix = callMLlibFunc("predictSoftGMM", x.map(_convert_to_vector),
                                               _convert_to_vector(self._weights), means, sigmas)
             return membership_matrix.map(lambda x: pyarray.array('d', x))
+        else:
+            raise TypeError("x should be represented by an RDD, "
+                            "but got %s." % type(x))
 
 
 class GaussianMixture(object):

sbin/start-master.sh

@@ -22,6 +22,8 @@
 sbin="`dirname "$0"`"
 sbin="`cd "$sbin"; pwd`"
 
+ORIGINAL_ARGS="$@"
+
 START_TACHYON=false
 
 while (( "$#" )); do
@@ -53,7 +55,9 @@ if [ "$SPARK_MASTER_WEBUI_PORT" = "" ]; then
   SPARK_MASTER_WEBUI_PORT=8080
 fi
 
-"$sbin"/spark-daemon.sh start org.apache.spark.deploy.master.Master 1 --ip $SPARK_MASTER_IP --port $SPARK_MASTER_PORT --webui-port $SPARK_MASTER_WEBUI_PORT
+"$sbin"/spark-daemon.sh start org.apache.spark.deploy.master.Master 1 \
+  --ip $SPARK_MASTER_IP --port $SPARK_MASTER_PORT --webui-port $SPARK_MASTER_WEBUI_PORT \
+  $ORIGINAL_ARGS
 
 if [ "$START_TACHYON" == "true" ]; then
   "$sbin"/../tachyon/bin/tachyon bootstrap-conf $SPARK_MASTER_IP

sql/core/src/main/scala/org/apache/spark/sql/SQLContext.scala

@@ -44,7 +44,7 @@ import org.apache.spark.sql.catalyst.ParserDialect
 import org.apache.spark.sql.execution.{Filter, _}
 import org.apache.spark.sql.jdbc.{JDBCPartition, JDBCPartitioningInfo, JDBCRelation}
 import org.apache.spark.sql.json._
-import org.apache.spark.sql.parquet.FSBasedParquetRelation
+import org.apache.spark.sql.parquet.ParquetRelation2
 import org.apache.spark.sql.sources._
 import org.apache.spark.sql.types._
 import org.apache.spark.util.Utils
@@ -610,7 +610,7 @@ class SQLContext(@transient val sparkContext: SparkContext)
     } else if (conf.parquetUseDataSourceApi) {
       val globbedPaths = paths.map(new Path(_)).flatMap(SparkHadoopUtil.get.globPath).toArray
       baseRelationToDataFrame(
-        new FSBasedParquetRelation(
+        new ParquetRelation2(
           globbedPaths.map(_.toString), None, None, Map.empty[String, String])(this))
     } else {
       DataFrame(this, parquet.ParquetRelation(
@@ -989,7 +989,7 @@ class SQLContext(@transient val sparkContext: SparkContext)
   def jdbc(url: String, table: String): DataFrame = {
     jdbc(url, table, JDBCRelation.columnPartition(null), new Properties())
   }
-  
+
   /**
    * :: Experimental ::
    * Construct a [[DataFrame]] representing the database table accessible via JDBC URL
@@ -1002,7 +1002,7 @@ class SQLContext(@transient val sparkContext: SparkContext)
   def jdbc(url: String, table: String, properties: Properties): DataFrame = {
     jdbc(url, table, JDBCRelation.columnPartition(null), properties)
   }
-  
+
   /**
    * :: Experimental ::
    * Construct a [[DataFrame]] representing the database table accessible via JDBC URL
@@ -1020,7 +1020,7 @@ class SQLContext(@transient val sparkContext: SparkContext)
   @Experimental
   def jdbc(
       url: String,
-      table: String,  
+      table: String,
       columnName: String,
       lowerBound: Long,
       upperBound: Long,
@@ -1056,7 +1056,7 @@ class SQLContext(@transient val sparkContext: SparkContext)
     val parts = JDBCRelation.columnPartition(partitioning)
     jdbc(url, table, parts, properties)
   }
-  
+
   /**
    * :: Experimental ::
    * Construct a [[DataFrame]] representing the database table accessible via JDBC URL
@@ -1093,7 +1093,7 @@ class SQLContext(@transient val sparkContext: SparkContext)
     }
     jdbc(url, table, parts, properties)
   }
-  
+
   private def jdbc(
       url: String,
       table: String,