[SPARK-6406] Launch Spark using assembly jar instead of a separate launcher jar

Author: Nishkam Ravi <nravi@cloudera.com>
Author: nishkamravi2 <nishkamravi@gmail.com>
Author: nravi <nravi@c1704.halxg.cloudera.com>

Closes #5085 from nishkamravi2/master_nravi and squashes the following commits:

bad4349 [nishkamravi2] Update Main.java
36a6f87 [Nishkam Ravi] Minor changes and bug fixes
b7f4ae7 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
4a45d6a [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
458af39 [Nishkam Ravi] Locate the jar using getLocation, obviates the need to pass assembly path as an argument
d9658d6 [Nishkam Ravi] Changes for SPARK-6406
ccdc334 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
3faa7a4 [Nishkam Ravi] Launcher library changes (SPARK-6406)
345206a [Nishkam Ravi] spark-class merge Merge branch 'master_nravi' of https://github.com/nishkamravi2/spark into master_nravi
ac58975 [Nishkam Ravi] spark-class changes
06bfeb0 [nishkamravi2] Update spark-class
35af990 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
32c3ab3 [nishkamravi2] Update AbstractCommandBuilder.java
4bd4489 [nishkamravi2] Update AbstractCommandBuilder.java
746f35b [Nishkam Ravi] "hadoop" string in the assembly name should not be mandatory (everywhere else in spark we mandate spark-assembly*hadoop*.jar)
bfe96e0 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
ee902fa [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
d453197 [nishkamravi2] Update NewHadoopRDD.scala
6f41a1d [nishkamravi2] Update NewHadoopRDD.scala
0ce2c32 [nishkamravi2] Update HadoopRDD.scala
f7e33c2 [Nishkam Ravi] Merge branch 'master_nravi' of https://github.com/nishkamravi2/spark into master_nravi
ba1eb8b [Nishkam Ravi] Try-catch block around the two occurrences of removeShutDownHook. Deletion of semi-redundant occurrences of expensive operation inShutDown.
71d0e17 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
494d8c0 [nishkamravi2] Update DiskBlockManager.scala
3c5ddba [nishkamravi2] Update DiskBlockManager.scala
f0d12de [Nishkam Ravi] Workaround for IllegalStateException caused by recent changes to BlockManager.stop
79ea8b4 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
b446edc [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
5c9a4cb [nishkamravi2] Update TaskSetManagerSuite.scala
535295a [nishkamravi2] Update TaskSetManager.scala
3e1b616 [Nishkam Ravi] Modify test for maxResultSize
9f6583e [Nishkam Ravi] Changes to maxResultSize code (improve error message and add condition to check if maxResultSize > 0)
5f8f9ed [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
636a9ff [nishkamravi2] Update YarnAllocator.scala
8f76c8b [Nishkam Ravi] Doc change for yarn memory overhead
35daa64 [Nishkam Ravi] Slight change in the doc for yarn memory overhead
5ac2ec1 [Nishkam Ravi] Remove out
dac1047 [Nishkam Ravi] Additional documentation for yarn memory overhead issue
42c2c3d [Nishkam Ravi] Additional changes for yarn memory overhead issue
362da5e [Nishkam Ravi] Additional changes for yarn memory overhead
c726bd9 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
f00fa31 [Nishkam Ravi] Improving logging for AM memoryOverhead
1cf2d1e [nishkamravi2] Update YarnAllocator.scala
ebcde10 [Nishkam Ravi] Modify default YARN memory_overhead-- from an additive constant to a multiplier (redone to resolve merge conflicts)
2e69f11 [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark into master_nravi
efd688a [Nishkam Ravi] Merge branch 'master' of https://github.com/apache/spark
2b630f9 [nravi] Accept memory input as "30g", "512M" instead of an int value, to be consistent with rest of Spark
3bf8fad [nravi] Merge branch 'master' of https://github.com/apache/spark
5423a03 [nravi] Merge branch 'master' of https://github.com/apache/spark
eb663ca [nravi] Merge branch 'master' of https://github.com/apache/spark
df2aeb1 [nravi] Improved fix for ConcurrentModificationIssue (Spark-1097, Hadoop-10456)
6b840f0 [nravi] Undo the fix for SPARK-1758 (the problem is fixed)
5108700 [nravi] Fix in Spark for the Concurrent thread modification issue (SPARK-1097, HADOOP-10456)
681b36f [nravi] Fix for SPARK-1758: failing test org.apache.spark.JavaAPISuite.wholeTextFiles

4 files changed, 69 insertions, 125 deletions

diff --git a/bin/spark-class b/bin/spark-class
index e29b234afa..c03946d92e 100755
--- a/bin/spark-class
+++ b/bin/spark-class
@@ -40,35 +40,46 @@ else
   fi
 fi
 
-# Look for the launcher. In non-release mode, add the compiled classes directly to the classpath
-# instead of looking for a jar file.
-SPARK_LAUNCHER_CP=
-if [ -f $SPARK_HOME/RELEASE ]; then
-  LAUNCHER_DIR="$SPARK_HOME/lib"
-  num_jars="$(ls -1 "$LAUNCHER_DIR" | grep "^spark-launcher.*\.jar$" | wc -l)"
-  if [ "$num_jars" -eq "0" -a -z "$SPARK_LAUNCHER_CP" ]; then
-    echo "Failed to find Spark launcher in $LAUNCHER_DIR." 1>&2
-    echo "You need to build Spark before running this program." 1>&2
-    exit 1
-  fi
+# Find assembly jar
+SPARK_ASSEMBLY_JAR=
+if [ -f "$SPARK_HOME/RELEASE" ]; then
+  ASSEMBLY_DIR="$SPARK_HOME/lib"
+else
+  ASSEMBLY_DIR="$SPARK_HOME/assembly/target/scala-$SPARK_SCALA_VERSION"
+fi
 
-  LAUNCHER_JARS="$(ls -1 "$LAUNCHER_DIR" | grep "^spark-launcher.*\.jar$" || true)"
-  if [ "$num_jars" -gt "1" ]; then
-    echo "Found multiple Spark launcher jars in $LAUNCHER_DIR:" 1>&2
-    echo "$LAUNCHER_JARS" 1>&2
-    echo "Please remove all but one jar." 1>&2
-    exit 1
-  fi
+num_jars="$(ls -1 "$ASSEMBLY_DIR" | grep "^spark-assembly.*hadoop.*\.jar$" | wc -l)"
+if [ "$num_jars" -eq "0" -a -z "$SPARK_ASSEMBLY_JAR" ]; then
+  echo "Failed to find Spark assembly in $ASSEMBLY_DIR." 1>&2
+  echo "You need to build Spark before running this program." 1>&2
+  exit 1
+fi
+ASSEMBLY_JARS="$(ls -1 "$ASSEMBLY_DIR" | grep "^spark-assembly.*hadoop.*\.jar$" || true)"
+if [ "$num_jars" -gt "1" ]; then
+  echo "Found multiple Spark assembly jars in $ASSEMBLY_DIR:" 1>&2
+  echo "$ASSEMBLY_JARS" 1>&2
+  echo "Please remove all but one jar." 1>&2
+  exit 1
+fi
 
-  SPARK_LAUNCHER_CP="${LAUNCHER_DIR}/${LAUNCHER_JARS}"
+SPARK_ASSEMBLY_JAR="${ASSEMBLY_DIR}/${ASSEMBLY_JARS}"
+
+# Verify that versions of java used to build the jars and run Spark are compatible
+if [ -n "$JAVA_HOME" ]; then
+  JAR_CMD="$JAVA_HOME/bin/jar"
 else
-  LAUNCHER_DIR="$SPARK_HOME/launcher/target/scala-$SPARK_SCALA_VERSION"
-  if [ ! -d "$LAUNCHER_DIR/classes" ]; then
-    echo "Failed to find Spark launcher classes in $LAUNCHER_DIR." 1>&2
-    echo "You need to build Spark before running this program." 1>&2
+  JAR_CMD="jar"
+fi
+
+if [ $(command -v "$JAR_CMD") ] ; then
+  jar_error_check=$("$JAR_CMD" -tf "$SPARK_ASSEMBLY_JAR" nonexistent/class/path 2>&1)
+  if [[ "$jar_error_check" =~ "invalid CEN header" ]]; then
+    echo "Loading Spark jar with '$JAR_CMD' failed. " 1>&2
+    echo "This is likely because Spark was compiled with Java 7 and run " 1>&2
+    echo "with Java 6. (see SPARK-1703). Please use Java 7 to run Spark " 1>&2
+    echo "or build Spark with Java 6." 1>&2
     exit 1
   fi
-  SPARK_LAUNCHER_CP="$LAUNCHER_DIR/classes"
 fi
 
 # The launcher library will print arguments separated by a NULL character, to allow arguments with
@@ -77,7 +88,7 @@ fi
 CMD=()
 while IFS= read -d '' -r ARG; do
   CMD+=("$ARG")
-done < <("$RUNNER" -cp "$SPARK_LAUNCHER_CP" org.apache.spark.launcher.Main "$@")
+done < <("$RUNNER" -cp "$SPARK_ASSEMBLY_JAR" org.apache.spark.launcher.Main "$@")
 
 if [ "${CMD[0]}" = "usage" ]; then
   "${CMD[@]}"
diff --git a/bin/spark-class2.cmd b/bin/spark-class2.cmd
index 37d22215a0..4ce727bc99 100644
--- a/bin/spark-class2.cmd
+++ b/bin/spark-class2.cmd
@@ -29,31 +29,20 @@ if "x%1"=="x" (
   exit /b 1
 )
 
-set LAUNCHER_CP=0
-if exist %SPARK_HOME%\RELEASE goto find_release_launcher
+rem Find assembly jar
+set SPARK_ASSEMBLY_JAR=0
 
-rem Look for the Spark launcher in both Scala build directories. The launcher doesn't use Scala so
-rem it doesn't really matter which one is picked up. Add the compiled classes directly to the
-rem classpath instead of looking for a jar file, since it's very common for people using sbt to use
-rem the "assembly" target instead of "package".
-set LAUNCHER_CLASSES=%SPARK_HOME%\launcher\target\scala-2.10\classes
-if exist %LAUNCHER_CLASSES% (
-  set LAUNCHER_CP=%LAUNCHER_CLASSES%
+if exist "%SPARK_HOME%\RELEASE" (
+  set ASSEMBLY_DIR=%SPARK_HOME%\lib
+) else (
+  set ASSEMBLY_DIR=%SPARK_HOME%\assembly\target\scala-%SPARK_SCALA_VERSION%
 )
-set LAUNCHER_CLASSES=%SPARK_HOME%\launcher\target\scala-2.11\classes
-if exist %LAUNCHER_CLASSES% (
-  set LAUNCHER_CP=%LAUNCHER_CLASSES%
-)
-goto check_launcher
 
-:find_release_launcher
-for %%d in (%SPARK_HOME%\lib\spark-launcher*.jar) do (
-  set LAUNCHER_CP=%%d
+for %%d in (%ASSEMBLY_DIR%\spark-assembly*hadoop*.jar) do (
+  set SPARK_ASSEMBLY_JAR=%%d
 )
-
-:check_launcher
-if "%LAUNCHER_CP%"=="0" (
-  echo Failed to find Spark launcher JAR.
+if "%SPARK_ASSEMBLY_JAR%"=="0" (
+  echo Failed to find Spark assembly JAR.
   echo You need to build Spark before running this program.
   exit /b 1
 )
@@ -64,7 +53,7 @@ if not "x%JAVA_HOME%"=="x" set RUNNER=%JAVA_HOME%\bin\java
 
 rem The launcher library prints the command to be executed in a single line suitable for being
 rem executed by the batch interpreter. So read all the output of the launcher into a variable.
-for /f "tokens=*" %%i in ('cmd /C ""%RUNNER%" -cp %LAUNCHER_CP% org.apache.spark.launcher.Main %*"') do (
+for /f "tokens=*" %%i in ('cmd /C ""%RUNNER%" -cp %SPARK_ASSEMBLY_JAR% org.apache.spark.launcher.Main %*"') do (
   set SPARK_CMD=%%i
 )
 %SPARK_CMD%
diff --git a/launcher/src/main/java/org/apache/spark/launcher/AbstractCommandBuilder.java b/launcher/src/main/java/org/apache/spark/launcher/AbstractCommandBuilder.java
index 2da5f72787..d8279145d8 100644
--- a/launcher/src/main/java/org/apache/spark/launcher/AbstractCommandBuilder.java
+++ b/launcher/src/main/java/org/apache/spark/launcher/AbstractCommandBuilder.java
@@ -86,10 +86,14 @@ abstract class AbstractCommandBuilder {
    */
   List<String> buildJavaCommand(String extraClassPath) throws IOException {
     List<String> cmd = new ArrayList<String>();
-    if (javaHome == null) {
-      cmd.add(join(File.separator, System.getProperty("java.home"), "bin", "java"));
-    } else {
+    String envJavaHome;
+
+    if (javaHome != null) {
       cmd.add(join(File.separator, javaHome, "bin", "java"));
+    } else if ((envJavaHome = System.getenv("JAVA_HOME")) != null) {
+        cmd.add(join(File.separator, envJavaHome, "bin", "java"));
+    } else {
+        cmd.add(join(File.separator, System.getProperty("java.home"), "bin", "java"));
     }
 
     // Load extra JAVA_OPTS from conf/java-opts, if it exists.
@@ -182,59 +186,25 @@ abstract class AbstractCommandBuilder {
       addToClassPath(cp, String.format("%s/core/target/jars/*", sparkHome));
     }
 
-    String assembly = findAssembly();
+    final String assembly = AbstractCommandBuilder.class.getProtectionDomain().getCodeSource().
+	getLocation().getPath();
     addToClassPath(cp, assembly);
 
-    // When Hive support is needed, Datanucleus jars must be included on the classpath. Datanucleus
-    // jars do not work if only included in the uber jar as plugin.xml metadata is lost. Both sbt
-    // and maven will populate "lib_managed/jars/" with the datanucleus jars when Spark is built
-    // with Hive, so first check if the datanucleus jars exist, and then ensure the current Spark
-    // assembly is built for Hive, before actually populating the CLASSPATH with the jars.
-    //
-    // This block also serves as a check for SPARK-1703, when the assembly jar is built with
-    // Java 7 and ends up with too many files, causing issues with other JDK versions.
-    boolean needsDataNucleus = false;
-    JarFile assemblyJar = null;
-    try {
-      assemblyJar = new JarFile(assembly);
-      needsDataNucleus = assemblyJar.getEntry("org/apache/hadoop/hive/ql/exec/") != null;
-    } catch (IOException ioe) {
-      if (ioe.getMessage().indexOf("invalid CEN header") >= 0) {
-        System.err.println(
-          "Loading Spark jar failed.\n" +
-          "This is likely because Spark was compiled with Java 7 and run\n" +
-          "with Java 6 (see SPARK-1703). Please use Java 7 to run Spark\n" +
-          "or build Spark with Java 6.");
-        System.exit(1);
-      } else {
-        throw ioe;
-      }
-    } finally {
-      if (assemblyJar != null) {
-        try {
-          assemblyJar.close();
-        } catch (IOException e) {
-          // Ignore.
-        }
-      }
+    // Datanucleus jars must be included on the classpath. Datanucleus jars do not work if only 
+    // included in the uber jar as plugin.xml metadata is lost. Both sbt and maven will populate 
+    // "lib_managed/jars/" with the datanucleus jars when Spark is built with Hive
+    File libdir;
+    if (new File(sparkHome, "RELEASE").isFile()) {
+      libdir = new File(sparkHome, "lib");
+    } else {
+      libdir = new File(sparkHome, "lib_managed/jars");
     }
 
-    if (needsDataNucleus) {
-      System.err.println("Spark assembly has been built with Hive, including Datanucleus jars " +
-        "in classpath.");
-      File libdir;
-      if (new File(sparkHome, "RELEASE").isFile()) {
-        libdir = new File(sparkHome, "lib");
-      } else {
-        libdir = new File(sparkHome, "lib_managed/jars");
-      }
-
-      checkState(libdir.isDirectory(), "Library directory '%s' does not exist.",
-        libdir.getAbsolutePath());
-      for (File jar : libdir.listFiles()) {
-        if (jar.getName().startsWith("datanucleus-")) {
-          addToClassPath(cp, jar.getAbsolutePath());
-        }
+    checkState(libdir.isDirectory(), "Library directory '%s' does not exist.",
+      libdir.getAbsolutePath());
+    for (File jar : libdir.listFiles()) {
+      if (jar.getName().startsWith("datanucleus-")) {
+        addToClassPath(cp, jar.getAbsolutePath());
       }
     }
 
@@ -270,7 +240,6 @@ abstract class AbstractCommandBuilder {
     if (scala != null) {
       return scala;
     }
-
     String sparkHome = getSparkHome();
     File scala210 = new File(sparkHome, "assembly/target/scala-2.10");
     File scala211 = new File(sparkHome, "assembly/target/scala-2.11");
@@ -330,30 +299,6 @@ abstract class AbstractCommandBuilder {
     return firstNonEmpty(childEnv.get(key), System.getenv(key));
   }
 
-  private String findAssembly() {
-    String sparkHome = getSparkHome();
-    File libdir;
-    if (new File(sparkHome, "RELEASE").isFile()) {
-      libdir = new File(sparkHome, "lib");
-      checkState(libdir.isDirectory(), "Library directory '%s' does not exist.",
-          libdir.getAbsolutePath());
-    } else {
-      libdir = new File(sparkHome, String.format("assembly/target/scala-%s", getScalaVersion()));
-    }
-
-    final Pattern re = Pattern.compile("spark-assembly.*hadoop.*\\.jar");
-    FileFilter filter = new FileFilter() {
-      @Override
-      public boolean accept(File file) {
-        return file.isFile() && re.matcher(file.getName()).matches();
-      }
-    };
-    File[] assemblies = libdir.listFiles(filter);
-    checkState(assemblies != null && assemblies.length > 0, "No assemblies found in '%s'.", libdir);
-    checkState(assemblies.length == 1, "Multiple assemblies found in '%s'.", libdir);
-    return assemblies[0].getAbsolutePath();
-  }
-
   private String getConfDir() {
     String confDir = getenv("SPARK_CONF_DIR");
     return confDir != null ? confDir : join(File.separator, getSparkHome(), "conf");
diff --git a/make-distribution.sh b/make-distribution.sh
index 9ed1abfe8c..738a9c4d69 100755
--- a/make-distribution.sh
+++ b/make-distribution.sh
@@ -199,7 +199,6 @@ echo "Build flags: $@" >> "$DISTDIR/RELEASE"
 # Copy jars
 cp "$SPARK_HOME"/assembly/target/scala*/*assembly*hadoop*.jar "$DISTDIR/lib/"
 cp "$SPARK_HOME"/examples/target/scala*/spark-examples*.jar "$DISTDIR/lib/"
-cp "$SPARK_HOME"/launcher/target/spark-launcher_$SCALA_VERSION-$VERSION.jar "$DISTDIR/lib/"
 # This will fail if the -Pyarn profile is not provided
 # In this case, silence the error and ignore the return code of this command
 cp "$SPARK_HOME"/network/yarn/target/scala*/spark-*-yarn-shuffle.jar "$DISTDIR/lib/" &> /dev/null || :