spark/run

#!/bin/bash

SCALA_VERSION=2.9.2

# Figure out where the Scala framework is installed
FWDIR="$(cd `dirname $0`; pwd)"

# Export this as SPARK_HOME
export SPARK_HOME="$FWDIR"

# Load environment variables from conf/spark-env.sh, if it exists
if [ -e $FWDIR/conf/spark-env.sh ] ; then
  . $FWDIR/conf/spark-env.sh
fi

if [ -z "$1" ]; then
  echo "Usage: run <spark-class> [<args>]" >&2
  exit 1
fi

# If this is a standalone cluster daemon, reset SPARK_JAVA_OPTS and SPARK_MEM to reasonable
# values for that; it doesn't need a lot
if [ "$1" = "spark.deploy.master.Master" -o "$1" = "spark.deploy.worker.Worker" ]; then
  SPARK_MEM=${SPARK_DAEMON_MEMORY:-512m}
  SPARK_JAVA_OPTS=$SPARK_DAEMON_JAVA_OPTS   # Empty by default
fi


# Add java opts for master, worker, executor. The opts maybe null
case "$1" in
  'spark.deploy.master.Master')
    SPARK_JAVA_OPTS+=" $SPARK_MASTER_OPTS"
    ;;
  'spark.deploy.worker.Worker')
    SPARK_JAVA_OPTS+=" $SPARK_WORKER_OPTS"
    ;;
  'spark.executor.StandaloneExecutorBackend')
    SPARK_JAVA_OPTS+=" $SPARK_EXECUTOR_OPTS"
    ;;
  'spark.executor.MesosExecutorBackend')
    SPARK_JAVA_OPTS+=" $SPARK_EXECUTOR_OPTS"
    ;;
  'spark.repl.Main')
    SPARK_JAVA_OPTS+=" $SPARK_REPL_OPTS"
    ;;
esac

if [ "$SPARK_LAUNCH_WITH_SCALA" == "1" ]; then
  if [ `command -v scala` ]; then
    RUNNER="scala"
  else
    if [ -z "$SCALA_HOME" ]; then
      echo "SCALA_HOME is not set" >&2
      exit 1
    fi
    RUNNER="${SCALA_HOME}/bin/scala"
  fi
else
  if [ `command -v java` ]; then
    RUNNER="java"
  else
    if [ -z "$JAVA_HOME" ]; then
      echo "JAVA_HOME is not set" >&2
      exit 1
    fi
    RUNNER="${JAVA_HOME}/bin/java"
  fi
  if [ -z "$SCALA_LIBRARY_PATH" ]; then
    if [ -z "$SCALA_HOME" ]; then
      echo "SCALA_HOME is not set" >&2
      exit 1
    fi
    SCALA_LIBRARY_PATH="$SCALA_HOME/lib"
  fi
fi

# Figure out how much memory to use per executor and set it as an environment
# variable so that our process sees it and can report it to Mesos
if [ -z "$SPARK_MEM" ] ; then
  SPARK_MEM="512m"
fi
export SPARK_MEM

# Set JAVA_OPTS to be able to load native libraries and to set heap size
JAVA_OPTS="$SPARK_JAVA_OPTS"
JAVA_OPTS+=" -Djava.library.path=$SPARK_LIBRARY_PATH"
JAVA_OPTS+=" -Xms$SPARK_MEM -Xmx$SPARK_MEM"
# Load extra JAVA_OPTS from conf/java-opts, if it exists
if [ -e $FWDIR/conf/java-opts ] ; then
  JAVA_OPTS+=" `cat $FWDIR/conf/java-opts`"
fi
export JAVA_OPTS

CORE_DIR="$FWDIR/core"
REPL_DIR="$FWDIR/repl"
EXAMPLES_DIR="$FWDIR/examples"
BAGEL_DIR="$FWDIR/bagel"
STREAMING_DIR="$FWDIR/streaming"
PYSPARK_DIR="$FWDIR/python"

# Exit if the user hasn't compiled Spark
if [ ! -e "$REPL_DIR/target" ]; then
  echo "Failed to find Spark classes in $REPL_DIR/target" >&2
  echo "You need to compile Spark before running this program" >&2
  exit 1
fi

# Build up classpath
CLASSPATH="$SPARK_CLASSPATH"
CLASSPATH+=":$FWDIR/conf"
CLASSPATH+=":$CORE_DIR/target/scala-$SCALA_VERSION/classes"
if [ -n "$SPARK_TESTING" ] ; then
  CLASSPATH+=":$CORE_DIR/target/scala-$SCALA_VERSION/test-classes"
  CLASSPATH+=":$STREAMING_DIR/target/scala-$SCALA_VERSION/test-classes"
fi
CLASSPATH+=":$CORE_DIR/src/main/resources"
CLASSPATH+=":$REPL_DIR/target/scala-$SCALA_VERSION/classes"
CLASSPATH+=":$EXAMPLES_DIR/target/scala-$SCALA_VERSION/classes"
CLASSPATH+=":$STREAMING_DIR/target/scala-$SCALA_VERSION/classes"
CLASSPATH+=":$STREAMING_DIR/lib/org/apache/kafka/kafka/0.7.2-spark/*" # <-- our in-project Kafka Jar
if [ -e "$FWDIR/lib_managed" ]; then
  CLASSPATH+=":$FWDIR/lib_managed/jars/*"
  CLASSPATH+=":$FWDIR/lib_managed/bundles/*"
fi
CLASSPATH+=":$REPL_DIR/lib/*"
if [ -e repl-bin/target ]; then
  for jar in `find "repl-bin/target" -name 'spark-repl-*-shaded-hadoop*.jar'`; do
    CLASSPATH+=":$jar"
  done
fi
CLASSPATH+=":$BAGEL_DIR/target/scala-$SCALA_VERSION/classes"
for jar in `find $PYSPARK_DIR/lib -name '*jar'`; do
  CLASSPATH+=":$jar"
done
export CLASSPATH # Needed for spark-shell

# Figure out the JAR file that our examples were packaged into. This includes a bit of a hack
# to avoid the -sources and -doc packages that are built by publish-local.
if [ -e "$EXAMPLES_DIR/target/scala-$SCALA_VERSION/spark-examples"*[0-9T].jar ]; then
  # Use the JAR from the SBT build
  export SPARK_EXAMPLES_JAR=`ls "$EXAMPLES_DIR/target/scala-$SCALA_VERSION/spark-examples"*[0-9T].jar`
fi
if [ -e "$EXAMPLES_DIR/target/spark-examples-"*hadoop[12].jar ]; then
  # Use the JAR from the Maven build
  export SPARK_EXAMPLES_JAR=`ls "$EXAMPLES_DIR/target/spark-examples-"*hadoop[12].jar`
fi

# Figure out whether to run our class with java or with the scala launcher.
# In most cases, we'd prefer to execute our process with java because scala
# creates a shell script as the parent of its Java process, which makes it
# hard to kill the child with stuff like Process.destroy(). However, for
# the Spark shell, the wrapper is necessary to properly reset the terminal
# when we exit, so we allow it to set a variable to launch with scala.
if [ "$SPARK_LAUNCH_WITH_SCALA" == "1" ]; then
  EXTRA_ARGS=""     # Java options will be passed to scala as JAVA_OPTS
else
  CLASSPATH+=":$SCALA_LIBRARY_PATH/scala-library.jar"
  CLASSPATH+=":$SCALA_LIBRARY_PATH/scala-compiler.jar"
  CLASSPATH+=":$SCALA_LIBRARY_PATH/jline.jar"
  # The JVM doesn't read JAVA_OPTS by default so we need to pass it in
  EXTRA_ARGS="$JAVA_OPTS"
fi

exec "$RUNNER" -cp "$CLASSPATH" $EXTRA_ARGS "$@"