spark初探

首先解压scala，本次选用版本scala-2.11.1

[Hadoop@CentOS software]$ tar -xzvf scala-2.11.1.tgz

[hadoop@centos software]$ su -

# vi /etc/profile

添加如下内容：

SCALA_HOME=/home/hadoop/software/scala-2.11.1

PATH=$SCALA_HOME/bin

EXPORT SCALA_HOME

# source /etc/profile

# scala -version

Scala code runner version 2.11.1 -- Copyright 2002-2013, LAMP/EPFL

然后解压spark，本次选用版本spark-1.0.0-bin-hadoop1.tgz，这次用的是hadoop 1.0.4

[hadoop@centos software]$ tar -xzvf spark-1.0.0-bin-hadoop1.tgz

--------------------------------------分割线 --------------------------------------

CentOS 6.2(64位)下安装Spark0.8.0详细记录 http://www.linuxidc.com/Linux/2014-06/102583.htm

Spark简介及其在Ubuntu下的安装使用 http://www.linuxidc.com/Linux/2013-08/88606.htm

安装Spark集群(在CentOS上) http://www.linuxidc.com/Linux/2013-08/88599.htm

Hadoop vs Spark性能对比 http://www.linuxidc.com/Linux/2013-08/88597.htm

Spark安装与学习 http://www.linuxidc.com/Linux/2013-08/88596.htm

Spark 并行计算模型 http://www.linuxidc.com/Linux/2012-12/76490.htm

--------------------------------------分割线 --------------------------------------

进入到spark的conf目录下

[hadoop@centos conf]$ cp spark-env.sh.template spark-env.sh

[hadoop@centos conf]$ vi spark-env.sh

添加如下内容：

export SCALA_HOME=/home/hadoop/software/scala-2.11.1

export SPARK_MASTER_IP=centos.host1

export SPARK_WORKER_MEMORY=5G

export JAVA_HOME=/usr/software/jdk

启动

[hadoop@centos spark-1.0.0-bin-hadoop1]$ sbin/start-master.sh

可以通过 http://centos.host1:8080/ 看到对应界面

[hadoop@centos spark-1.0.0-bin-hadoop1]$ sbin/start-slaves.sh park://centos.host1:7077

可以通过 http://centos.host1:8081/ 看到对应界面

下面在spark上运行第一个例子：与Hadoop交互的WordCount

首先将word.txt文件上传到HDFS上，这里路径是 hdfs://centos.host1:9000/user/hadoop/data/wordcount/001/word.txt

进入交互模式

[hadoop@centos spark-1.0.0-bin-hadoop1]$ master=spark://centos.host1:7077 ./bin/spark-shell

scala>val file=sc.textFile("hdfs://centos.host1:9000/user/hadoop/data/wordcount/001/word.txt")

scala>val count=file.flatMap(line=>line.split(" ")).map(word=>(word,1)).reduceByKey(_+_)

scala>count.collect()

可以看到控制台有如下结果：

res0: Array[(String, Int)] = Array((hive,2), (zookeeper,1), (pig,1), (spark,1), (hadoop,4), (hbase,2))

同时也可以将结果保存到HDFS上

scala>count.saveAsTextFile("hdfs://centos.host1:9000/user/hadoop/data/wordcount/001/result.txt")

接下来再来看下如何运行Java版本的WordCount

这里需要用到一个jar文件：spark-assembly-1.0.0-hadoop1.0.4.jar

WordCount代码如下：

public class WordCount {

 private static final Pattern SPACE = Pattern.compile(" ");

 @SuppressWarnings("serial")
 public static void main(String[] args) throws Exception {
  if (args.length < 1) {
   System.err.println("Usage: JavaWordCount <file>");
   System.exit(1);
  }

  SparkConf sparkConf = new SparkConf().setAppName("JavaWordCount");
  JavaSparkContext ctx = new JavaSparkContext(sparkConf);
  JavaRDD<String> lines = ctx.textFile(args[0], 1);

  JavaRDD<String> words = lines.flatMap(new FlatMapFunction<String, String>() {
     @Override
     public Iterable<String> call(String s) {
      return Arrays.asList(SPACE.split(s));
     }
    });

  JavaPairRDD<String, Integer> ones = words.mapToPair(new PairFunction<String, String, Integer>() {
     @Override
     public Tuple2<String, Integer> call(String s) {
      return new Tuple2<String, Integer>(s, 1);
     }
    });

  JavaPairRDD<String, Integer> counts = ones.reduceByKey(new Function2<Integer, Integer, Integer>() {
     @Override
     public Integer call(Integer i1, Integer i2) {
      return i1 + i2;
     }
    });

  List<Tuple2<String, Integer>> output = counts.collect();
  for (Tuple2<?, ?> tuple : output) {
   System.out.println(tuple._1() + " : " + tuple._2());
  }

  ctx.stop();
 }
}