Issue submitting Spark Job from code when Spark Job is a Python program.

[sehunley]

When submitting a Java or Scala program, everything works fine. When submitting a python program, it's gets to the ACCEPTED state and then stalls. It eventually times out, but it's not getting picked up to run. Is this interface just for Java/Scala programs/jobs or should it be able to submit PySpark/Python jobs as well?

I am trying to invoke the pi.py sample program that comes with Spark 1.6.0.

Below is the java program that I am testing with. I'm new to Spark so apologies for any "newbie" errors.

import org.apache.spark.SparkConf;
import org.apache.spark.deploy.yarn.Client;
import org.apache.spark.deploy.yarn.ClientArguments;
import org.apache.hadoop.conf.Configuration;
// import org.apache.log4j.Logger;

/**

• This class submits a SparkPi to a YARN from a Java client (as opposed

• to submitting a Spark job from a shell command line using spark-submit).

• To accomplish submitting a Spark job from a Java client, we use

• the org.apache.spark.deploy.yarn.Client class described below:

• Usage: org.apache.spark.deploy.yarn.Client [options]
  Options:
  --jar JAR_PATH Path to your application's JAR file (required in yarn-cluster mode)
  --class CLASS_NAME Name of your application's main class (required)
  --primary-py-file A main Python file
  --arg ARG Argument to be passed to your application's main class.
  Multiple invocations are possible, each will be passed in order.
  --num-executors NUM Number of executors to start (Default: 2)
  --executor-cores NUM Number of cores per executor (Default: 1).
  --driver-memory MEM Memory for driver (e.g. 1000M, 2G) (Default: 512 Mb)
  --driver-cores NUM Number of cores used by the driver (Default: 1).
  --executor-memory MEM Memory per executor (e.g. 1000M, 2G) (Default: 1G)
  --name NAME The name of your application (Default: Spark)
  --queue QUEUE The hadoop queue to use for allocation requests (Default: 'default')
  --addJars jars Comma separated list of local jars that want SparkContext.addJar to work with.
  --py-files PY_FILES Comma-separated list of .zip, .egg, or .py files to place on the PYTHONPATH for Python apps.
  --files files Comma separated list of files to be distributed with the job.
  --archives archives Comma separated list of archives to be distributed with the job.

  How to call this program example:

  export SPARK_HOME="/Users/mparsian/spark-1.6.0"
  java -DSPARK_HOME="$SPARK_HOME" org.dataalgorithms.client.SubmitSparkPiToYARNFromJavaCode 10
  */
  public class SubmitSparkPiToYARNFromJavaCode {

  public static void main(String[] args) throws Exception {
  long startTime = System.currentTimeMillis();

  // this is passed to SparkPi program
  //THE_LOGGER.info("Slices Passed=" + args[0]);
  String slices = args[0];  
  // String slices = "10";
  //
  // String SPARK_HOME = System.getProperty("SPARK_HOME");
  String SPARK_HOME = "/opt/spark/spark-1.6.0";
  // THE_LOGGER.info("SPARK_HOME=" + SPARK_HOME);
  
  //
  pi(SPARK_HOME, slices); // ... the code being measured ... 
  //
  long elapsedTime = System.currentTimeMillis() - startTime;
  // THE_LOGGER.info("elapsedTime (millis)=" + elapsedTime);
  

  }

  static void pi(String SPARK_HOME, String slices) throws Exception {
  //
  String[] args = new String[]{
  "--name",
  "Submit-SparkPi-To-Yarn",
  //
  "--driver-memory",
  "512MB",
  //
  "--jar",
  SPARK_HOME + "/examples/target/spark-examples_2.11-1.6.0.jar",
  //
  "--class",
  "org.apache.spark.examples.JavaSparkPi",

      // argument 1 to my Spark program
      "--arg",
      slices,
  
      // argument 2 to my Spark program (helper argument to create a proper JavaSparkContext object)
      "--arg",
      "yarn-cluster"
  };
  
  Configuration config = new Configuration();
  //
  System.setProperty("SPARK_YARN_MODE", "true");
  //
  SparkConf sparkConf = new SparkConf();
  ClientArguments clientArgs = new ClientArguments(args, sparkConf);
  Client client = new Client(clientArgs, config, sparkConf);
  
  client.run();
  // done!
  

  }
  }

Thanks,

-Scott

[mahmoudparsian]

Hello Scott,

Thank you very much for your email and introduction.
I am not an expert in python, but I want to take a look at your python program, please share it so that I can look at it.

Thank you again!
best regards,
Mahmoud

On Feb 15, 2016, at 11:23 AM, sehunley notifications@github.com wrote:

When submitting a Java or Scala program, everything works fine. When submitting a python program, it's gets to the ACCEPTED state and then stalls. It eventually times out, but it's not getting picked up to run. Is this interface just for Java/Scala programs/jobs or should it be able to submit PySpark/Python jobs as well?

Thanks,

-Scott

—
Reply to this email directly or view it on GitHub #4.

[sehunley]

Hello Mahmoud,
My Python program is pretty simple, it's a sample program that came with Spark 1.6.0. Our Data Scientists are doing their Machine Learning scripts in python, hence the need to call python programatically. Here is the program:

from future import print_function

import sys
from random import random
from operator import add

from pyspark import SparkContext

if name == "main":
"""
Usage: pi [partitions]
"""
sc = SparkContext(appName="PythonPi")
partitions = int(sys.argv[1]) if len(sys.argv) > 1 else 2
n = 100000 * partitions

def f(_):
    x = random() * 2 - 1
    y = random() * 2 - 1
    return 1 if x ** 2 + y ** 2 < 1 else 0

count = sc.parallelize(range(1, n + 1), partitions).map(f).reduce(add)
print("Pi is roughly %f" % (4.0 * count / n))

sc.stop()

Thanks in advance for any help that you can give. Here is a link to the actual output that happens when I run the code and try to submit the program above: [http://stackoverflow.com/questions/35373367/issue-submitting-a-python-application-to-yarn-from-java-code]

-Scott