Merge remote-tracking branch 'upstream/master' into mllib-stats-api-check

Author: jkbradley

.rat-excludes

@@ -25,6 +25,7 @@ log4j-defaults.properties
 bootstrap-tooltip.js
 jquery-1.11.1.min.js
 sorttable.js
+.*avsc
 .*txt
 .*json
 .*data

README.md

@@ -115,6 +115,15 @@ If your project is built with Maven, add this to your POM file's `<dependencies>
     </dependency>
 
 
+## A Note About Thrift JDBC server and CLI for Spark SQL
+
+Spark SQL supports Thrift JDBC server and CLI.
+See sql-programming-guide.md for more information about those features.
+You can use those features by setting `-Phive-thriftserver` when building Spark as follows.
+
+    $ sbt/sbt -Phive-thriftserver assembly
+
+
 ## Configuration
 
 Please refer to the [Configuration guide](http://spark.apache.org/docs/latest/configuration.html)

bin/spark-shell.cmd

@@ -19,4 +19,4 @@ rem
 
 set SPARK_HOME=%~dp0..
 
-cmd /V /E /C %SPARK_HOME%\bin\spark-submit.cmd spark-shell --class org.apache.spark.repl.Main %*
+cmd /V /E /C %SPARK_HOME%\bin\spark-submit.cmd --class org.apache.spark.repl.Main %* spark-shell

bin/spark-sql

@@ -65,30 +65,30 @@ while (($#)); do
   case $1 in
     -d | --define | --database | -f | -h | --hiveconf | --hivevar | -i | -p)
       ensure_arg_number $# 2
-      CLI_ARGS+=($1); shift
-      CLI_ARGS+=($1); shift
+      CLI_ARGS+=("$1"); shift
+      CLI_ARGS+=("$1"); shift
       ;;
 
     -e)
       ensure_arg_number $# 2
-      CLI_ARGS+=($1); shift
-      CLI_ARGS+=(\"$1\"); shift
+      CLI_ARGS+=("$1"); shift
+      CLI_ARGS+=("$1"); shift
       ;;
 
     -s | --silent)
-      CLI_ARGS+=($1); shift
+      CLI_ARGS+=("$1"); shift
       ;;
 
     -v | --verbose)
       # Both SparkSubmit and SparkSQLCLIDriver recognizes -v | --verbose
-      CLI_ARGS+=($1)
-      SUBMISSION_ARGS+=($1); shift
+      CLI_ARGS+=("$1")
+      SUBMISSION_ARGS+=("$1"); shift
       ;;
 
     *)
-      SUBMISSION_ARGS+=($1); shift
+      SUBMISSION_ARGS+=("$1"); shift
       ;;
   esac
 done
 
-eval exec "$FWDIR"/bin/spark-submit --class $CLASS ${SUBMISSION_ARGS[*]} spark-internal ${CLI_ARGS[*]}
+exec "$FWDIR"/bin/spark-submit --class $CLASS "${SUBMISSION_ARGS[@]}" spark-internal "${CLI_ARGS[@]}"

core/src/main/java/org/apache/spark/network/netty/FileClient.java

@@ -66,10 +66,15 @@ private[spark] class ContextCleaner(sc: SparkContext) extends Logging {
 
   /**
    * Whether the cleaning thread will block on cleanup tasks.
-   * This is set to true only for tests.
+   *
+   * Due to SPARK-3015, this is set to true by default. This is intended to be only a temporary
+   * workaround for the issue, which is ultimately caused by the way the BlockManager actors
+   * issue inter-dependent blocking Akka messages to each other at high frequencies. This happens,
+   * for instance, when the driver performs a GC and cleans up all broadcast blocks that are no
+   * longer in scope.
    */
   private val blockOnCleanupTasks = sc.conf.getBoolean(
-    "spark.cleaner.referenceTracking.blocking", false)
+    "spark.cleaner.referenceTracking.blocking", true)
 
   @volatile private var stopped = false
 
@@ -174,9 +179,6 @@ private[spark] class ContextCleaner(sc: SparkContext) extends Logging {
   private def blockManagerMaster = sc.env.blockManager.master
   private def broadcastManager = sc.env.broadcastManager
   private def mapOutputTrackerMaster = sc.env.mapOutputTracker.asInstanceOf[MapOutputTrackerMaster]
-
-  // Used for testing. These methods explicitly blocks until cleanup is completed
-  // to ensure that more reliable testing.
 }
 
 private object ContextCleaner {