Merge branch 'master' of https://github.com/apache/spark into dt-preprune

Conflicts:
	mllib/src/main/scala/org/apache/spark/mllib/tree/configuration/Strategy.scala

Author: qiping.lqp

core/src/main/scala/org/apache/spark/deploy/master/ApplicationInfo.scala

@@ -96,11 +96,13 @@ private[spark] class ApplicationInfo(
 
   def retryCount = _retryCount
 
-  def incrementRetryCount = {
+  def incrementRetryCount() = {
     _retryCount += 1
     _retryCount
   }
 
+  def resetRetryCount() = _retryCount = 0
+
   def markFinished(endState: ApplicationState.Value) {
     state = endState
     endTime = System.currentTimeMillis()

core/src/main/scala/org/apache/spark/deploy/master/Master.scala

@@ -296,28 +296,34 @@ private[spark] class Master(
       val execOption = idToApp.get(appId).flatMap(app => app.executors.get(execId))
       execOption match {
         case Some(exec) => {
+          val appInfo = idToApp(appId)
           exec.state = state
+          if (state == ExecutorState.RUNNING) { appInfo.resetRetryCount() }
           exec.application.driver ! ExecutorUpdated(execId, state, message, exitStatus)
           if (ExecutorState.isFinished(state)) {
-            val appInfo = idToApp(appId)
             // Remove this executor from the worker and app
-            logInfo("Removing executor " + exec.fullId + " because it is " + state)
+            logInfo(s"Removing executor ${exec.fullId} because it is $state")
             appInfo.removeExecutor(exec)
             exec.worker.removeExecutor(exec)
 
-            val normalExit = exitStatus.exists(_ == 0)
+            val normalExit = exitStatus == Some(0)
             // Only retry certain number of times so we don't go into an infinite loop.
-            if (!normalExit && appInfo.incrementRetryCount < ApplicationState.MAX_NUM_RETRY) {
-              schedule()
-            } else if (!normalExit) {
-              logError("Application %s with ID %s failed %d times, removing it".format(
-                appInfo.desc.name, appInfo.id, appInfo.retryCount))
-              removeApplication(appInfo, ApplicationState.FAILED)
+            if (!normalExit) {
+              if (appInfo.incrementRetryCount() < ApplicationState.MAX_NUM_RETRY) {
+                schedule()
+              } else {
+                val execs = appInfo.executors.values
+                if (!execs.exists(_.state == ExecutorState.RUNNING)) {
+                  logError(s"Application ${appInfo.desc.name} with ID ${appInfo.id} failed " +
+                    s"${appInfo.retryCount} times; removing it")
+                  removeApplication(appInfo, ApplicationState.FAILED)
+                }
+              }
             }
           }
         }
         case None =>
-          logWarning("Got status update for unknown executor " + appId + "/" + execId)
+          logWarning(s"Got status update for unknown executor $appId/$execId")
       }
     }
 

core/src/main/scala/org/apache/spark/deploy/worker/ExecutorRunner.scala

@@ -159,6 +159,8 @@ private[spark] class ExecutorRunner(
       Files.write(header, stderr, Charsets.UTF_8)
       stderrAppender = FileAppender(process.getErrorStream, stderr, conf)
 
+      state = ExecutorState.RUNNING
+      worker ! ExecutorStateChanged(appId, execId, state, None, None)
       // Wait for it to exit; executor may exit with code 0 (when driver instructs it to shutdown)
       // or with nonzero exit code
       val exitCode = process.waitFor()

core/src/main/scala/org/apache/spark/deploy/worker/Worker.scala

@@ -234,7 +234,7 @@ private[spark] class Worker(
         try {
           logInfo("Asked to launch executor %s/%d for %s".format(appId, execId, appDesc.name))
           val manager = new ExecutorRunner(appId, execId, appDesc, cores_, memory_,
-            self, workerId, host, sparkHome, workDir, akkaUrl, conf, ExecutorState.RUNNING)
+            self, workerId, host, sparkHome, workDir, akkaUrl, conf, ExecutorState.LOADING)
           executors(appId + "/" + execId) = manager
           manager.start()
           coresUsed += cores_

dev/run-tests

@@ -93,7 +93,7 @@ echo "========================================================================="
 # echo "q" is needed because sbt on encountering a build file with failure
 # (either resolution or compilation) prompts the user for input either q, r,
 # etc to quit or retry. This echo is there to make it not block.
-BUILD_MVN_PROFILE_ARGS="$SBT_MAVEN_PROFILES_ARGS -Phive -Phive-thriftserver "
+BUILD_MVN_PROFILE_ARGS="$SBT_MAVEN_PROFILES_ARGS -Phive "
 echo -e "q\n" | sbt/sbt $BUILD_MVN_PROFILE_ARGS clean package assembly/assembly | \
   grep -v -e "info.*Resolving" -e "warn.*Merging" -e "info.*Including"
 

docs/mllib-decision-tree.md

@@ -80,7 +80,7 @@ The ordered splits create "bins" and the maximum number of such
 bins can be specified using the `maxBins` parameter.
 
 Note that the number of bins cannot be greater than the number of instances `$N$` (a rare scenario
-since the default `maxBins` value is 100). The tree algorithm automatically reduces the number of
+since the default `maxBins` value is 32). The tree algorithm automatically reduces the number of
 bins if the condition is not satisfied.
 
 **Categorical features**
@@ -117,7 +117,7 @@ all nodes at each level of the tree. This could lead to high memory requirements
 of the tree, potentially leading to memory overflow errors. To alleviate this problem, a `maxMemoryInMB`
 training parameter specifies the maximum amount of memory at the workers (twice as much at the
 master) to be allocated to the histogram computation. The default value is conservatively chosen to
-be 128 MB to allow the decision algorithm to work in most scenarios. Once the memory requirements
+be 256 MB to allow the decision algorithm to work in most scenarios. Once the memory requirements
 for a level-wise computation cross the `maxMemoryInMB` threshold, the node training tasks at each
 subsequent level are split into smaller tasks.
 
@@ -167,7 +167,7 @@ val numClasses = 2
 val categoricalFeaturesInfo = Map[Int, Int]()
 val impurity = "gini"
 val maxDepth = 5
-val maxBins = 100
+val maxBins = 32
 
 val model = DecisionTree.trainClassifier(data, numClasses, categoricalFeaturesInfo, impurity,
   maxDepth, maxBins)
@@ -213,7 +213,7 @@ Integer numClasses = 2;
 HashMap<Integer, Integer> categoricalFeaturesInfo = new HashMap<Integer, Integer>();
 String impurity = "gini";
 Integer maxDepth = 5;
-Integer maxBins = 100;
+Integer maxBins = 32;
 
 // Train a DecisionTree model for classification.
 final DecisionTreeModel model = DecisionTree.trainClassifier(data, numClasses,
@@ -250,7 +250,7 @@ data = MLUtils.loadLibSVMFile(sc, 'data/mllib/sample_libsvm_data.txt').cache()
 # Train a DecisionTree model.
 #  Empty categoricalFeaturesInfo indicates all features are continuous.
 model = DecisionTree.trainClassifier(data, numClasses=2, categoricalFeaturesInfo={},
-                                     impurity='gini', maxDepth=5, maxBins=100)
+                                     impurity='gini', maxDepth=5, maxBins=32)
 
 # Evaluate model on training instances and compute training error
 predictions = model.predict(data.map(lambda x: x.features))
@@ -293,7 +293,7 @@ val data = MLUtils.loadLibSVMFile(sc, "data/mllib/sample_libsvm_data.txt").cache
 val categoricalFeaturesInfo = Map[Int, Int]()
 val impurity = "variance"
 val maxDepth = 5
-val maxBins = 100
+val maxBins = 32
 
 val model = DecisionTree.trainRegressor(data, categoricalFeaturesInfo, impurity,
   maxDepth, maxBins)
@@ -338,7 +338,7 @@ JavaSparkContext sc = new JavaSparkContext(sparkConf);
 HashMap<Integer, Integer> categoricalFeaturesInfo = new HashMap<Integer, Integer>();
 String impurity = "variance";
 Integer maxDepth = 5;
-Integer maxBins = 100;
+Integer maxBins = 32;
 
 // Train a DecisionTree model.
 final DecisionTreeModel model = DecisionTree.trainRegressor(data,
@@ -380,7 +380,7 @@ data = MLUtils.loadLibSVMFile(sc, 'data/mllib/sample_libsvm_data.txt').cache()
 # Train a DecisionTree model.
 #  Empty categoricalFeaturesInfo indicates all features are continuous.
 model = DecisionTree.trainRegressor(data, categoricalFeaturesInfo={},
-                                    impurity='variance', maxDepth=5, maxBins=100)
+                                    impurity='variance', maxDepth=5, maxBins=32)
 
 # Evaluate model on training instances and compute training error
 predictions = model.predict(data.map(lambda x: x.features))

examples/src/main/java/org/apache/spark/examples/mllib/JavaDecisionTree.java

@@ -63,7 +63,7 @@ public static void main(String[] args) {
     HashMap<Integer, Integer> categoricalFeaturesInfo = new HashMap<Integer, Integer>();
     String impurity = "gini";
     Integer maxDepth = 5;
-    Integer maxBins = 100;
+    Integer maxBins = 32;
 
     // Train a DecisionTree model for classification.
     final DecisionTreeModel model = DecisionTree.trainClassifier(data, numClasses,

examples/src/main/scala/org/apache/spark/examples/mllib/DecisionTreeRunner.scala

@@ -52,9 +52,9 @@ object DecisionTreeRunner {
       input: String = null,
       dataFormat: String = "libsvm",
       algo: Algo = Classification,
-      maxDepth: Int = 4,
+      maxDepth: Int = 5,
       impurity: ImpurityType = Gini,
-      maxBins: Int = 100,
+      maxBins: Int = 32,
       fracTest: Double = 0.2)
 
   def main(args: Array[String]) {

mllib/src/main/scala/org/apache/spark/mllib/tree/DecisionTree.scala

@@ -331,9 +331,9 @@ object DecisionTree extends Serializable with Logging {
    *                 Supported values: "gini" (recommended) or "entropy".
    * @param maxDepth Maximum depth of the tree.
    *                 E.g., depth 0 means 1 leaf node; depth 1 means 1 internal node + 2 leaf nodes.
-   *                  (suggested value: 4)
+   *                  (suggested value: 5)
    * @param maxBins maximum number of bins used for splitting features
-   *                 (suggested value: 100)
+   *                 (suggested value: 32)
    * @return DecisionTreeModel that can be used for prediction
    */
   def trainClassifier(
@@ -375,9 +375,9 @@ object DecisionTree extends Serializable with Logging {
    *                 Supported values: "variance".
    * @param maxDepth Maximum depth of the tree.
    *                 E.g., depth 0 means 1 leaf node; depth 1 means 1 internal node + 2 leaf nodes.
-   *                  (suggested value: 4)
+   *                  (suggested value: 5)
    * @param maxBins maximum number of bins used for splitting features
-   *                 (suggested value: 100)
+   *                 (suggested value: 32)
    * @return DecisionTreeModel that can be used for prediction
    */
   def trainRegressor(

mllib/src/main/scala/org/apache/spark/mllib/tree/configuration/Strategy.scala

@@ -56,20 +56,20 @@ import org.apache.spark.mllib.tree.configuration.QuantileStrategy._
  *                    If a split has less information gain than minInfoGain,
  *                    this split will not be considered as a valid split.
  * @param maxMemoryInMB Maximum memory in MB allocated to histogram aggregation. Default value is
- *                      128 MB.
+ *                      256 MB.
  */
 @Experimental
 class Strategy (
     val algo: Algo,
     val impurity: Impurity,
     val maxDepth: Int,
     val numClassesForClassification: Int = 2,
-    val maxBins: Int = 100,
+    val maxBins: Int = 32,
     val quantileCalculationStrategy: QuantileStrategy = Sort,
     val categoricalFeaturesInfo: Map[Int, Int] = Map[Int, Int](),
     val minInstancesPerNode: Int = 0,
     val minInfoGain: Double = 0.0,
-    val maxMemoryInMB: Int = 128) extends Serializable {
+    val maxMemoryInMB: Int = 256) extends Serializable {
 
   if (algo == Classification) {
     require(numClassesForClassification >= 2)