Spark集群框架的搭建与入门

[root@hop01 opt]# tar -zxvf scala-2.12.2.tgz
[root@hop01 opt]# mv scala-2.12.2 scala2.12

[root@hop01 opt]# vim /etc/profile

export SCALA_HOME=/opt/scala2.12
export PATH=$PATH:$SCALA_HOME/bin

[root@hop01 opt]# source /etc/profile

[root@hop01 opt]# scala -version

[root@hop01 opt]# tar -zxvf spark-2.1.1-bin-hadoop2.7.tgz
[root@hop01 opt]# mv spark-2.1.1-bin-hadoop2.7 spark2.1

[root@hop01 opt]# vim /etc/profile

export SPARK_HOME=/opt/spark2.1
export PATH=$PATH:$SPARK_HOME/bin

[root@hop01 opt]# source /etc/profile

[root@hop01 opt]# spark-shell

[root@hop01 opt]# cd /opt/spark2.1/conf/
[root@hop01 conf]# cp slaves.template slaves
[root@hop01 conf]# vim slaves

hop01
hop02
hop03

[root@hop01 conf]# cp spark-env.sh.template spark-env.sh
[root@hop01 conf]# vim spark-env.sh

export JAVA_HOME=/opt/jdk1.8
export SCALA_HOME=/opt/scala2.12
export SPARK_MASTER_IP=hop01
export SPARK_LOCAL_IP=安装节点IP
export SPARK_WORKER_MEMORY=1g
export HADOOP_CONF_DIR=/opt/hadoop2.7/etc/hadoop

启动：/opt/spark2.1/sbin/start-all.sh
停止：/opt/spark2.1/sbin/stop-all.sh

http://hop01:8080/

[root@hop01 spark2.1]# cd /opt/spark2.1/
[root@hop01 spark2.1]# bin/spark-submit --class org.apache.spark.examples.SparkPi --master local examples/jars/spark-examples_2.11-2.1.1.jar

运行结果：Pi is roughly 3.1455357276786384

<dependency>
    <groupId>org.apache.spark</groupId>
    <artifactId>spark-core_2.11</artifactId>
    <version>2.1.1</version>
</dependency>

<plugin>
    <groupId>net.alchim31.maven</groupId>
    <artifactId>scala-maven-plugin</artifactId>
    <version>3.2.2</version>
    <executions>
        <execution>
            <goals>
                <goal>compile</goal>
                <goal>testCompile</goal>
            </goals>
        </execution>
    </executions>
</plugin>

@RestController
public class WordWeb implements Serializable {

    @GetMapping("/word/web")
    public String getWeb (){
        // 1、创建Spark的配置对象
        SparkConf sparkConf = new SparkConf().setAppName("LocalCount")
                                             .setMaster("local[*]");

        // 2、创建SparkContext对象
        JavaSparkContext sc = new JavaSparkContext(sparkConf);
        sc.setLogLevel("WARN");

        // 3、读取测试文件
        JavaRDD lineRdd = sc.textFile("/var/spark/test/word.txt");

        // 4、行内容进行切分
        JavaRDD wordsRdd = lineRdd.flatMap(new FlatMapFunction() {
            @Override
            public Iterator call(Object obj) throws Exception {
                String value = String.valueOf(obj);
                String[] words = value.split(",");
                return Arrays.asList(words).iterator();
            }
        });

        // 5、切分的单词进行标注
        JavaPairRDD wordAndOneRdd = wordsRdd.mapToPair(new PairFunction() {
            @Override
            public Tuple2 call(Object obj) throws Exception {
                //将单词进行标记：
                return new Tuple2(String.valueOf(obj), 1);
            }
        });

        // 6、统计单词出现次数
        JavaPairRDD wordAndCountRdd = wordAndOneRdd.reduceByKey(new Function2() {
            @Override
            public Object call(Object obj1, Object obj2) throws Exception {
                return Integer.parseInt(obj1.toString()) + Integer.parseInt(obj2.toString());
            }
        });

        // 7、排序
        JavaPairRDD sortedRdd = wordAndCountRdd.sortByKey();
        List<Tuple2> finalResult = sortedRdd.collect();

        // 8、结果打印
        for (Tuple2 tuple2 : finalResult) {
            System.out.println(tuple2._1 + " ===> " + tuple2._2);
        }

        // 9、保存统计结果
        sortedRdd.saveAsTextFile("/var/spark/output");
        sc.stop();
        return "success" ;
    }
}