spark-submit提交任务到集群

1.参数选取

当我们的代码写完，打好jar，就可以通过bin/spark-submit 提交到集群，命令如下：

./bin/spark-submit \  
 --class <main-class> 
 --master <master-url> \ 
 --deploy-mode <deploy-mode> \ 
 --conf <key>=<value> \  
     ... # other options  
 <application-jar> \  
 [application-arguments]

一般情况下使用上面这几个参数就够用了

• --class: The entry point for your application (e.g. org.apache.spark.examples.SparkPi)

• --master: The master URL for the cluster (e.g. spark://23.195.26.187:7077)

• --deploy-mode: Whether to deploy your driver on the worker nodes (cluster) or locally as an external client (client) (default: client) †

• --conf: Arbitrary Spark configuration property in key=value format. For values that contain spaces wrap “key=value” in quotes (as shown).

• application-jar: Path to a bundled jar including your application and all dependencies. The URL must be globally visible inside of your cluster, for instance, an hdfs:// path or a file:// path that is present on all nodes.

• application-arguments: Arguments passed to the main method of your main class, if any

  对于不同的集群管理，对spark-submit的提交列举几个简单的例子
  

# Run application locally on 8 cores

 ./bin/spark-submit \
 --class org.apache.spark.examples.SparkPi \
--master local[8] \ 
  /path/to/examples.jar \
100

 # Run on a Spark standalone cluster in client deploy mode
 
./bin/spark-submit \
 --class org.apache.spark.examples.SparkPi \
--master spark://207.184.161.138:7077 \
--executor-memory 20G \ 
--total-executor-cores 100 \
/path/to/examples.jar \ 
 1000 
 
 # Run on a Spark standalone cluster in cluster deploy mode with supervise 
 # make sure that the driver is automatically restarted if it fails with non-zero exit code
 
 ./bin/spark-submit \
 --class org.apache.spark.examples.SparkPi \
 --master spark://207.184.161.138:7077 \ 
 --deploy-mode cluster
--supervise 
 --executor-memory 20G \
--total-executor-cores 100 \
 /path/to/examples.jar \ 
   1000
   
# Run on a YARN cluster export HADOOP_CONF_DIR=XXX

./bin/spark-submit \ 
 --class org.apache.spark.examples.SparkPi \
--master yarn-cluster \  # can also be `yarn-client` for client mode 
 --executor-memory 20G \ 
  --num-executors 50 \ 
  /path/to/examples.jar \ 
 1000 
 
 # Run a Python application on a Spark standalone cluster
 
./bin/spark-submit \ 
  --master spark://207.184.161.138:7077 \ 
 examples/src/main/python/pi.py \ 
1000

2.具体提交步骤

代码实现一个简单的统计

public class SimpleSample {
	public static void main(String[] args) {
		String logFile = "/home/bigdata/spark-1.5.1/README.md"; 
		SparkConf conf = new SparkConf().setAppName("Simple Application");
		JavaSparkContext sc = new JavaSparkContext(conf);
		JavaRDD<String> logData = sc.textFile(logFile).cache();

		long numAs = logData.filter(new Function<String, Boolean>() {
			public Boolean call(String s) {
				return s.contains("a");
			}
		}).count();

		long numBs = logData.filter(new Function<String, Boolean>() {
			public Boolean call(String s) {
				return s.contains("b");
			}
		}).count();

		System.out.println("Lines with a: " + numAs + ", lines with b: " + numBs);
	}

}

打成jar

上传命令

./bin/spark-submit --class cs.spark.SimpleSample --master spark://spark1:7077 /home/jar/spark-test-0.0.1-SNAPSHOT.jar