spark：spark-submit 提交任务及参数说明（yarn）

Spark：对于提交命令的理解：

https://blog.csdn.net/weixin_38750084/article/details/106973247

spark-submit 可以提交任务到 spark 集群执行，也可以提交到 hadoop 的 yarn 集群执行。

代码中配置：

util：

import org.apache.spark.serializer.KryoSerializer
import org.apache.spark.sql.SparkSession

object SparkContextUtil {
  /**
    * 封装创建sparkContext实例
    *
    * @param appName
    * @param params
    * @return
    */
  def createSparkContext(appName: String, params: Map[String, String] = Map.empty) = {
    // 入口
    val spark: SparkSession = SparkSession.builder()
      .appName(appName)
      .config("spark.sql.warehouse.dir", "/user/hive/warehouse")
      .master("local[*]")
      .config("spark.serializer",classOf[KryoSerializer].getName)
      .config("spark.debug.maxToStringFields", "100")
      .enableHiveSupport().getOrCreate()


    // 封装用户传递进来的参数
    params.foreach { case (key, value) => spark.conf.set(key, value) }

    spark
  }
}

使用：

object BusinessDataCombineErpJobs {

  Logger.getLogger("org").setLevel(Level.WARN)
  val logger = LoggerFactory.getLogger(BusinessDataCombineErpJobs.getClass.getSimpleName)


  def main(args: Array[String]): Unit = {

    val spark = SparkContextUtil.createSparkContext(TestSparkSql.getClass.getSimpleName)
    //返回基础sparkContext，用于创建RDD以及管理群集资源
    val sc = spark.sparkContext
    println("---数据处理开始---")
    test(spark)
    println("---数据处理结束---")
    spark.close()
  }
}

1. 例子

一个最简单的例子，部署 spark standalone 模式后，提交到本地执行。

./bin/spark-submit \
--master spark://localhost:7077 \
examples/src/main/python/pi.py

如果部署 hadoop，并且启动 yarn 后，spark 提交到 yarn 执行的例子如下。

注意，spark 必须编译成支持 yarn 模式，编译 spark 的命令为：

build/mvn -Pyarn -Phadoop-2.x -Dhadoop.version=2.x.x -DskipTests clean package

其中, 2.x 为 hadoop 的版本号。编译完成后，可执行下面的命令，提交任务到 hadoop yarn 集群执行。

./bin/spark-submit --class org.apache.spark.examples.SparkPi \
--master yarn \
--deploy-mode cluster \
--driver-memory 1g \
--executor-memory 1g \
--executor-cores 1 \
--queue thequeue \
examples/target/scala-2.11/jars/spark-examples*.jar 10

注意：后边的数字10是传入的一个参数

线上实操：

spark2-submit --class bi.tag.TSimilarTagsTable --master yarn-client --executor-memory 6G --num-executors 5 --executor-cores 2 /var/lib/hadoop-hdfs/seijing/ble/tag/spark-sql/pf-spark-master/pi/target/pi-1.0.1-SNAPSHOT.jar


spark2-submit --class resume.mlib.RcoAID \
--master yarn \
--deploy-mode client  \
--num-executors 4  \
--executor-memory 10G \
--executor-cores 3 \
--driver-memory 10g \
--conf "spark.executor.extraJavaOptions='-Xss512m'" \
--driver-java-options "-Xss512m" \
/var/lib/hadoop-hdfs/als_ecommend/reserver-1.0-SNAPSHOT.jar  $1 $2 >> /var/lib/hadoop-hdfs/als_ecommend/logs/log_spark_out_`date +\%Y\%m\%d`.log

注意：
（1）
$1 $2 是 上一层，执行这个脚本传进来的参数
如：
/bin/bash /root/combine.sh aa  bb
aa bb 就是传入的参数

（2）
最后打印出的日志格式为：
-rw-r--r-- 1 root root   2375 Feb 27 15:25 log_spark_out_20200227.log
-rw-r--r-- 1 root root 712272 Feb 28 17:03 log_spark_out_20200228.log
-rw-r--r-- 1 root root   2375 Mar  9 15:36 log_spark_out_20200309.log
-rw-r--r-- 1 root root 712463 Mar 10 20:24 log_spark_out_20200310.log
-rw-r--r-- 1 root root  10578 Mar 12 18:51 log_spark_out_20200312.log
-rw-r--r-- 1 root root 468018 Mar 13 10:06 log_spark_out_20200313.log
-rw-r--r-- 1 root root 712602 Mar 19 18:26 log_spark_out_20200319.log
只有print的，以及DF show 这样的日志才会存储到日志文件中。
logger打印的日志在控制台运行任务时可以看到，但是并不能存储到日志文件中。

2. spark-submit 详细参数说明

参数名	参数说明
--master	master 的地址，提交任务到哪里执行，例如 spark://host:port, yarn, local
--deploy-mode	在本地 (client) 启动 driver 或在 cluster 上启动，默认是 client
--class	应用程序的主类，仅针对 java 或 scala 应用
--name	应用程序的名称
--jars	用逗号分隔的本地 jar 包，设置后，这些 jar 将包含在 driver 和 executor 的 classpath 下
--packages	包含在driver 和executor 的 classpath 中的 jar 的 maven 坐标
--exclude-packages	为了避免冲突而指定不包含的 package
--repositories	远程 repository
--conf PROP=VALUE	指定 spark 配置属性的值，例如 -conf spark.executor.extraJavaOptions="-XX:MaxPermSize=256m"
--properties-file	加载的配置文件，默认为 conf/spark-defaults.conf
--driver-memory	Driver内存，默认 1G
--driver-java-options	传给 driver 的额外的 Java 选项
--driver-library-path	传给 driver 的额外的库路径
--driver-class-path	传给 driver 的额外的类路径
--driver-cores	Driver 的核数，默认是1。在 yarn 或者 standalone 下使用
--executor-memory	每个 executor 的内存，默认是1G
--total-executor-cores	所有 executor 总共的核数。仅仅在 mesos 或者 standalone 下使用
--num-executors	启动的 executor 数量。默认为2。在 yarn 下使用
--executor-core	每个 executor 的核数。在yarn或者standalone下使用

yarn-client模式跑任务无异常（代码配置中配置了.master("local[*]")）

脚本为：

spark2-submit \
--class bi.tag.BusinessDataCombineErpJobs \
--master yarn-client \
--executor-memory 3G \
--num-executors 5 \
--executor-cores 2 \
/var/business_data/p-1.0.1-SNAPSHOT.jar > /var/business_data/business_data.log

代码中去掉.master("local[*]")，任务依然可以跑成功。

但是代码中存在.master("local[*]")参数的情况下，我直接把脚本改为：

--master yarn \

--deploy-mode cluster \

报错了

spark2-submit \
--class bi.tag.BusinessDataCombineErpJobs \
--master yarn \
--deploy-mode cluster \
--driver-memory 1g \
--executor-memory 3g \
--executor-cores 2 \
/var/business_data/p-1.0.1-SNAPSHOT.jar 10

注意：数字10 是代码BusinessDataCombineErpJobs 中自定义的传入的一个参数

报错日志为：

azkaban：

28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 20/05/28 15:04:20 INFO yarn.Client: Application report for application_1583730534669_117324 (state: FAILED)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 20/05/28 15:04:20 INFO yarn.Client: 
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 client token: N/A
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 diagnostics: Application application_1583730534669_117324 failed 2 times due to AM Container for appattempt_1583730534669_117324_000002 exited with  exitCode: 13
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - For more detailed output, check application tracking page:http://pf-bigdata4:8088/proxy/application_1583730534669_117324/Then, click on links to logs of each attempt.
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Diagnostics: Exception from container-launch.
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Container id: container_e87_1583730534669_117324_02_000001
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Exit code: 13
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Stack trace: ExitCodeException exitCode=13: 
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.hadoop.util.Shell.runCommand(Shell.java:604)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.hadoop.util.Shell.run(Shell.java:507)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.hadoop.util.Shell$ShellCommandExecutor.execute(Shell.java:789)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.hadoop.yarn.server.nodemanager.DefaultContainerExecutor.launchContainer(DefaultContainerExecutor.java:213)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:302)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.hadoop.yarn.server.nodemanager.containermanager.launcher.ContainerLaunch.call(ContainerLaunch.java:82)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at java.lang.Thread.run(Thread.java:748)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Container exited with a non-zero exit code 13
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Failing this attempt. Failing the application.
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 ApplicationMaster host: N/A
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 ApplicationMaster RPC port: -1
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 queue: default
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 start time: 1590649410241
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 final status: FAILED
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 tracking URL: http://pf-bigdata4:8088/cluster/app/application_1583730534669_117324
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	 user: root
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Exception in thread "main" org.apache.spark.SparkException: Application application_1583730534669_117324 finished with failed status
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.spark.deploy.yarn.Client.run(Client.scala:1153)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.spark.deploy.yarn.YarnClusterApplication.start(Client.scala:1568)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:892)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:197)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:227)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:136)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 	at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 20/05/28 15:04:20 INFO util.ShutdownHookManager: Shutdown hook called
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 20/05/28 15:04:20 INFO util.ShutdownHookManager: Deleting directory /tmp/spark-eb1e1b60-ef09-4a58-8e5f-dc988411999e
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - 20/05/28 15:04:20 INFO util.ShutdownHookManager: Deleting directory /huayong/data/tmp/spark-dba79ec3-1f27-4da0-8e8e-5a98c31c156f
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Process completed unsuccessfully in 55 seconds.
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine ERROR - Job run failed!
java.lang.RuntimeException: azkaban.jobExecutor.utils.process.ProcessFailureException: Process exited with code 1
	at azkaban.jobExecutor.ProcessJob.run(ProcessJob.java:305)
	at azkaban.execapp.JobRunner.runJob(JobRunner.java:787)
	at azkaban.execapp.JobRunner.doRun(JobRunner.java:602)
	at azkaban.execapp.JobRunner.run(JobRunner.java:563)
	at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
	at java.util.concurrent.FutureTask.run(FutureTask.java:266)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
	at java.lang.Thread.run(Thread.java:748)
Caused by: azkaban.jobExecutor.utils.process.ProcessFailureException: Process exited with code 1
	at azkaban.jobExecutor.utils.process.AzkabanProcess.run(AzkabanProcess.java:125)
	at azkaban.jobExecutor.ProcessJob.run(ProcessJob.java:297)
	... 8 more
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine ERROR - azkaban.jobExecutor.utils.process.ProcessFailureException: Process exited with code 1 cause: azkaban.jobExecutor.utils.process.ProcessFailureException: Process exited with code 1
28-05-2020 15:04:20 CST bi_cal_business_data_table_combine INFO - Finishing job bi_cal_business_data_table_combine at 1590649460777 with status FAILED

yarn logs -applicationId application_1583730534669_117324命令查看日志为：

20/05/28 15:04:17 WARN lazy.LazyStruct: Extra bytes detected at the end of the row! Ignoring similar problems.
20/05/28 15:04:17 WARN lazy.LazyStruct: Extra bytes detected at the end of the row! Ignoring similar problems.
20/05/28 15:04:19 ERROR yarn.ApplicationMaster: Uncaught exception: 
java.lang.IllegalStateException: User did not initialize spark context!
	at org.apache.spark.deploy.yarn.ApplicationMaster.runDriver(ApplicationMaster.scala:467)
	at org.apache.spark.deploy.yarn.ApplicationMaster.org$apache$spark$deploy$yarn$ApplicationMaster$$runImpl(ApplicationMaster.scala:301)
	at org.apache.spark.deploy.yarn.ApplicationMaster$$anonfun$run$1.apply$mcV$sp(ApplicationMaster.scala:241)
	at org.apache.spark.deploy.yarn.ApplicationMaster$$anonfun$run$1.apply(ApplicationMaster.scala:241)
	at org.apache.spark.deploy.yarn.ApplicationMaster$$anonfun$run$1.apply(ApplicationMaster.scala:241)
	at org.apache.spark.deploy.yarn.ApplicationMaster$$anon$3.run(ApplicationMaster.scala:782)
	at java.security.AccessController.doPrivileged(Native Method)
	at javax.security.auth.Subject.doAs(Subject.java:422)
	at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1917)
	at org.apache.spark.deploy.yarn.ApplicationMaster.doAsUser(ApplicationMaster.scala:781)
	at org.apache.spark.deploy.yarn.ApplicationMaster.run(ApplicationMaster.scala:240)
	at org.apache.spark.deploy.yarn.ApplicationMaster$.main(ApplicationMaster.scala:806)
	at org.apache.spark.deploy.yarn.ApplicationMaster.main(ApplicationMaster.scala)

脚本最后一行的自定义的传参的参数 10去掉，依然上面的错。

但是代码中把.master("local[*]"去掉后，使用client和cluster模式，都可以跑成功。

总结：

1.代码中local[*]参数去掉后，两种模式都可以跑成功，不去掉，只能跑client模式

2.cluster模式是在集群跑任务，使用的是集群随机一台机器的资源，而client模式是在提交任务的这台机器上跑，使用的是这台机器的资源

3.没问题的脚本：

client：

spark2-submit \
--class bi.tag.BusinessDataCombineErpJobs \
--master yarn-client \
--driver-memory 1g \
--executor-memory 3g \
--executor-cores 2 \
/var/business_data/pi-1.0.1-SNAPSHOT-yarn-cluster.jar

cluster：

spark2-submit \
--class bi.tag.BusinessDataCombineErpJobs \
--master yarn \
--deploy-mode cluster \
--driver-memory 1g \
--executor-memory 3g \
--executor-cores 2 \
/var/business_data/pi-1.0.1-SNAPSHOT-yarn-cluster.jar

sparkstreaming的提交示例：

spark2-submit --master yarn-client --conf spark.driver.memory=2g --class com.tzb.sparkstreaming.prod.DataChangeStreaming --executor-memory 8G --num-executors 5 --executor-cores 2 /test/spark-test-jar-with-dependencies.jar >> /test/sparkstreaming_datachange.log

参考：

https://www.cnblogs.com/weiweifeng/p/8073553.html

spark：spark-submit 提交任务及参数说明（yarn）

代码中配置：

1. 例子

2. spark-submit 详细参数说明

继续阅读

ubuntu hadoop2.6.1，terminal下运行wordcount

Spark基础:Spark简介及特点,运行模式,安装Spark,Driver与Executor,Local模式,Standalone模式,Yarn模式,Mesos模式,WordCount案例,HA配置第1章 Spark概述第2章 Spark运行模式第3章案例实操

MapReduce(一)：入门级程序wordcount及其分析

hadoop操作遇到的问题问题一：输出文件已存在

Hadoop之运行wordcount

jdk1.7+Eclipse+Maven3.5+Hadoop2.7.3构建hadoop项目

Spark实现wordcount

Eclipse运行WordCount（详细版）相关连接Eclipse运行WordCount

hadoop 用MR实现join操作

Centos7 下 Hadoop 2.6.4 分布式集群环境搭建摘要集群准备安装JDK 安装 Hadoop 2.6.4 部署 slaver1-slaver4 启动 hadoop 集群成功了

MapReduce的几个企业级经典面试案例MapReduce的几个企业级经典面试案例

大数据排错SparkSpark集群启动时候，JAVA_HOME is not sethadoop集群，某台服务器jps无任何输出IDEAkafkahadoopspark sqlfile permissionsIDEA本地测试 - OutOfMemoryError: GC overhead limit exceededhdfs负载均衡

ubuntu14.04下安装hbse1.0.1.1

User Defined Hadoop DataType

Ambari介绍和架构原理

spark/scala关于【资源文件】加载方法概述外部文件加载方案测试资源文件打包入jar包中小结