[完]Spark安装学习实践

【参考】http://dblab.xmu.edu.cn/blog/804-2/

一、前提

• 安装Hadoop2.6.0以上；
• 安装JAVA JDK 1.7以上。

二、下载Spark

官方网站：http://spark.apache.org/downloads.html
1. 选择版本：Spark 1.6.2
2. 选择包类型：Pre-build with user-provided Hadoop [can use with most Hadoop distributions]
3. 选择下载类型：Select Apache Mirror
4. 下载Spark：点击接下来的链接，即可下载

三、安装Spark

假设Spark下载到当前用户的HOME目录下。

# 解压缩
sudo tar -zxf spark-1.6.2-bin-without-hadoop -C /usr/local/
cd /usr/local
sudo mv ./spark-1.6.2-bin-without-hadoop/ ./spark
# 修改权限
sudo chown -R hadoop:hadoop ./spark

配置Spark，修改配置文件spark-env.sh。

cd /usr/local/spark/conf
cp spark-env.sh.template spark-env.sh
vim spark-env.sh

添加配置信息。

export SPARK_DIST_CLASSPATH=$(/usr/local/hadoop/bin/hadoop classpath)

配置完成，无需Hadoop那样运行启动命令，可直接使用。使用示例程序，验证Spark是否安装成功。

cd /usr/local/spark
bin/run-example SparkPi
# 2>&1,将所有信息都输出到stdout中
bin/run-example SparkPi 2>&1 | grep "Pi is"

示例程序结果：

hadoop@ubuntu:/usr/local/spark$ bin/run-example SparkPi 2>&1 | grep "Pi is"
Pi is roughly 3.14576

四、使用Spark Shell编写代码

• 启动Spark Shell，会自动创建为sc的spark context对象和名为sqlContext的sql context对象。

cd /usr/local/spark
bin/spark-shell

运行spark shell后结果：

......
16/09/14 05:18:32 INFO repl.SparkILoop: Created spark context..
Spark context available as sc.
16/09/14 05:18:33 INFO repl.SparkILoop: Created sql context..
SQL context available as sqlContext.
scala> 

• 加载text文件，spark创建sc，可加载本地文件和HDFS文件创建RDD。

scala> val textFile = sc.textFile("file:///usr/local/spark/README.md")

• 简单的RDD操作

# 获取RDD文件textFile的第一行内容
scala> textFile.first()
# 获取RDD文件textFile所有项的计数
scala> textFile.count()
# 抽取含有"Spark"的行，返回一个新的RDD
scala> val lineWithSpark = textFile.filter(line => line.contains("Spark"))
# 统计新的RDD的行数
scala> lineWithSpark.count()
# 通过组合RDD操作，实现简易MapReduce操作
scala> textFile.map(line => line.split(" ").size).reduce((a, b) => if (a>b) a else b)

• 退出Spark Shell，输入exit，或者Ctrl+C，即可退出Spark Shell

五、Scala应用程序编程

• Scala编写的程序需要使用sbt进行编译打包
• Java程序使用Maven编译打包
• Python程序则通过spark-submit直接提交

5-1 安装sbt

sbt是Spark用来对Scala程序进行打包的工具。

• 下载地址：https://repo.typesafe.com/typesafe/ivy-releases/org.scala-sbt/sbt-launch/0.13.11/sbt-launch.jar
• 安装在/usr/local/sbt目录下：

sudo mkdir /usr/local/sbt
sudo chown -R hadoop:hadoop /usr/local/sbt
cd /usr/local/sbt
cp ~/sbt-launch.jar .
# 创建sbt脚本
vim ./sbt

• 脚本sbt中，添加下面内容：

#!/bin/bash
SBT_OPTS="-Xms512M -Xmx1536M -Xss1M -XX:+CMSClassUnloadingEnabled -XX:MaxPermSize=256M"
java $SBT_OPTS -jar `dirname $0`/sbt-launch.jar "$@"

• 为脚本添加可执行权限

chmod u+x ./sbt

• 检查sbt是否可用，确保电脑处于联网状态，首次运行会出现“Getting org.scala-sbt sbt 0.13.11 …”的下载信息。

./sbt sbt-version

　　出现如下结果，表示安装成功

......
    [SUCCESSFUL ] org.fusesource.jansi#jansi;1.4!jansi.jar (6739ms)
:: retrieving :: org.scala-sbt#boot-scala
    confs: [default]
    5 artifacts copied, 0 already retrieved (24494kB/222ms)
[info] Set current project to sbt (in build file:/usr/local/sbt/)
[info] 0.13.11

5-2 Scala应用程序代码

• 创建一个文件夹 sparkapp 作为应用程序根目录，在目录下创建一个名为 SimpleApp.scala 的文件。

cd ~
mkdir sparkapp
mkdir -p ./sparkapp/src/main/scala      # 创建所需的文件夹结构
vim ./sparkapp/src/main/scala/SimpleApp.scala

• 在SimpleApp.scala文件中，编写Scala应用程序代码

import org.apache.spark.SparkContext
import org.apache.spark.SparkContext._
import org.apache.spark.SparkConf

object SimpleApp {
        def main(args:Array[String]) {
                val logFile = "file:///usr/local/spark/README.md"
                val conf = new SparkConf().setAppName("Simple Application")
                val sc = new SparkContext(conf)
                val logData = sc.textFile(logFile, 2).cache()
                val numAs = logData.filter(line => line.contains("a")).count()
                val numBs = logData.filter(line => line.contains("b")).count()
                println("Lines with a: %s, Lines with b: %s".format(numAs, numBs))
        }
}

　　该程序用于计算/usr/local/spark/README.md中含有“a”的行数和含有“b”的行数。程序依赖于Spark API，需要使用sbt进行编译打包。

• ~/sparkapp中新建文件simple.sbt（vim ./sparkapp/simple.sbt），添加下面内容，声明改程序的信息以及与Spark的依赖关系

name := "Simple Project"

version := "1.0"

scalaVersion := "2.10.5"

libraryDependencies += "org.apache.spark" %% "spark-core" % "1.6.2"

　　在上面的配置信息中，scalaVersion用来指定scala的版本，sparkcore用来指定spark的版本，这两个版本信息都可以在之前的启动 Spark shell 的过程中，从如下的屏幕的显示信息中找到。

......
Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _\ \/ _ \/ _ `/ __/  '_/
   /___/ .__/\_,_/_/ /_/\_\   version 1.6.2
      /_/

Using Scala version 2.10.5 (OpenJDK Client VM, Java 1.7.0_111)
......

5-3 使用sbt打包Scala程序

• 检查应用程序的目录结构

cd ~/sparkapp
find .

　　文件结构应如下所示：

.
./simple.sbt
./src
./src/main
./src/main/scala
./src/main/scala/SimpleApp.scala

• 将应用程序打包成JAR（首次运行需要下载依赖包），生成的JAR包位置为~/sparkapp/target/scala-2.10/simple-project_2.10-1.0.jar

hadoop@ubuntu:~/sparkapp$ /usr/local/sbt/sbt package
......
[info] Packaging /home/hadoop/sparkapp/target/scala-2.10/simple-project_2.10-1.0.jar ...
[info] Done packaging.
[success] Total time: 7 s, completed Sep 17, 2016 11:31:28 PM

• 通过spark-submit运行程序

# 显示完整信息
hadoop@ubuntu:~/sparkapp$ /usr/local/spark/bin/spark-submit --class "SimpleApp" ~/sparkapp/target/scala-2.10/simple-project_2.10-1.0.jar 
16/09/17 23:50:00 INFO spark.SparkContext: Running Spark version 1.6.2
16/09/17 23:50:01 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
......
16/09/17 23:50:12 INFO scheduler.TaskSchedulerImpl: Removed TaskSet 1.0, whose tasks have all completed, from pool 
Lines with a: 58, Lines with b: 26
16/09/17 23:50:12 INFO spark.SparkContext: Invoking stop() from shutdown hook
......
16/09/17 23:50:13 INFO remote.RemoteActorRefProvider$RemotingTerminator: Remote daemon shut down; proceeding with flushing remote transports.
# 显示所需要的信息
hadoop@ubuntu:~/sparkapp$ /usr/local/spark/bin/spark-submit --class "SimpleApp" ~/sparkapp/target/scala-2.10/simple-project_2.10-1.0.jar 2>&1 | grep "Lines with a:"
Lines with a: 58, Lines with b: 26

六、Java独立应用编程

6-1 安装Maven

• 下载
  
  • 去官网下载 https://maven.apache.org/download.cgi#Files
  • wget http://apache.fayea.com/maven/maven-3/3.3.9/binaries/apache-maven-3.3.9-bin.zip
• 安装

sudo unzip apache-maven-3.3.9-bin.zip -d /usr/local
cd /usr/local
sudo mv apache-maven-3.3.9/ maven
/usr/local$ sudo chown -R hadoop:hadoop maven/

6-2 Java应用程序代码

• 进入HOME目录，创建相关目录，建立SimpleApp.java文件

cd ~
mkdir -p sparkapp2/src/main/java
vim sparkapp2/src/main/java/Simple.java

• SimpleApp.java文件中添加如下代码：

import org.apache.spark.api.java.*;
import org.apache.spark.api.java.function.Function;

public class SimpleApp {
        public static void main(String[] args){
                String logFile = "file:///usr/local/spark/README.md";
                JavaSparkContext sc = new JavaSparkContext("local", "Simple App", "file:///usr/local/spark/", 
                    new String[]{"target/simple-project-1.0.jar"});
                JavaRDD logData = sc.textFile(logFile).cache();
                long numAs = logData.filter(new Function() {
                        public Boolean call(String s) {
                                return s.contains("a");
                        }
                }).count();

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

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

• 该程序依赖Spark Java API，需要通过Maven进行编译打包。在./sparkapp2中新建文件pom.xml（vim ~/sparkapp2/pom.xml），添加下面内容，声明该程序信息以及与Spark的依赖关系：

<project>
        <groupId>edu.berkeleygroupId>
        <artifactId>simple-projectartifactId>
        <modelVersion>4.0.0modelVersion>
        <name>Simple Projectname>
        <packaging>jarpackaging>
        <version>1.0version>
        <repositories>
                <repository>
                        <id>Akka repositoryid>
                        <url>http://repo.akka.io/releasesurl>
                repository>
        repositories>
        <dependencies>
                <dependency>
                        <groupId>org.apache.sparkgroupId>
                        <artifactId>spark-core_2.11artifactId>
                        <version>2.0.0-previewversion>
                dependency>
        dependencies>
project>

6-3 使用maven打包java程序

• 检查应用程序文件结构

hadoop@ubuntu:~/sparkapp2$ find
.
./src
./src/main
./src/main/java
./src/main/java/Simple.java
./pom.xml

• 将应用程序打包成JAR文件（首次运行需要下载依赖包，需要联网，消耗一定的时间）：

hadoop@ubuntu:~/sparkapp2$ /usr/local/maven/bin/mvn package
......
[INFO] Building jar: /home/hadoop/sparkapp2/target/simple-project-1.0.jar
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 32.926 s
[INFO] Finished at: 2016-09-18T18:59:14-07:00
[INFO] Final Memory: 26M/63M
[INFO] ------------------------------------------------------------------------

• 通过spark-submit运行程序

hadoop@ubuntu:~/sparkapp2$ /usr/local/spark/bin/spark-submit --class "SimpleApp" ~/sparkapp2/target/simple-project-1.0.jar
......
hadoop@ubuntu:~/sparkapp2$ /usr/local/spark/bin/spark-submit --class "SimpleApp" ~/sparkapp2/target/simple-project-1.0.jar 2>&1 | grep "Lines with a"
Lines with a: 58, Lines with b: 26