King's Uncle

Local Execution
在该模式下，在一个含有多个线程的进程中执行task。

$SPARK_HOME/bin/spark-submit --class org.apress.prospark.TranslateApp --master local[n]
./target/scala-2.10/FirstApp-assembly-1.0.jar <app_name> <book_path> <output_path> <language>

Standalone Cluster 在该模式下，Driver可以在提交job的机器上执行 ``` shell $SPARK_HOME/bin/spark-submit –class org.apress.prospark.TranslateApp –master ./target/scala-2.10/FirstApp-assembly-1.0.jar

``` ![standalone_cluster1](/assets/img/standalone cluster1.png) 也可以在集群中运行Driver ``` shell $SPARK_HOME/bin/spark-submit ¨Cclass org.apress.prospark.TranslateApp --master --deploy-mode cluster ./target/scala-2.10/FirstApp-assembly-1.0.jar ``` ![standalone_cluster2](/assets/img/standalone cluster2.png) 这两种方式的job都在work上执行，只是Driver的执行位置不同而已。 * Yarn 和standalone模式一样，Driver可以在Client执行 ``` shell $SPARK_HOME/bin/spark-submit --class org.apress.prospark.TranslateApp --master yarn --deploy-mode client ./target/scala-2.10/FirstApp-assembly-1.0.jar ``` 也可以在集群中执行： ``` shell $SPARK_HOME/bin/spark-submit --class org.apress.prospark.TranslateApp --master yarn --deploy-mode cluster ./target/scala-2.10/FirstApp-assembly-1.0.jar ```

Kuncle

God I pray to prosper thee.

Spark Execution