一、概述
程序运行环境很重要,本次测试基于:
hadoop-2.6.5
spark-1.6.2
hbase-1.2.4
zookeeper-3.4.6
jdk-1.8
废话不多说了,直接上需求
Andy column=baseINFO:age, value=21
Andy column=baseINFO:gender, value=0
Andy column=baseINFO:telphone_number, value=110110110
Tom column=baseINFO:age, value=18
Tom column=baseINFO:gender, value=1
Tom column=baseINFO:telphone_number, value=120120120
如上表所示,将之用spark进行分组,达到这样的效果:
[Andy,(21,0,110110110)]
[Tom,(18,1,120120120)]
需求比较简单,主要是熟悉一下程序运行过程
二、具体代码
package com.union.bigdata.spark.hbase;import org.apache.hadoop.hbase.HBaseConfiguration;import org.apache.hadoop.hbase.mapreduce.TableSplit;import org.apache.hadoop.hbase.util.Base64;import org.apache.hadoop.hbase.util.Bytes;import org.apache.spark.api.java.JavaSparkContext;import org.apache.spark.api.java.JavaRDD;import org.apache.spark.SparkConf;import org.apache.spark.api.java.function.Function;import org.apache.spark.api.java.function.Function2;import org.apache.hadoop.conf.Configuration;import org.apache.hadoop.hbase.client.Scan;import org.apache.hadoop.hbase.client.Result;import org.apache.hadoop.hbase.io.ImmutableBytesWritable;import org.apache.hadoop.hbase.mapreduce.TableInputFormat;import org.apache.spark.api.java.JavaPairRDD;import org.apache.hadoop.hbase.protobuf.ProtobufUtil;import org.apache.hadoop.hbase.protobuf.generated.ClientProtos;import org.apache.spark.api.java.function.PairFunction;import scala.Tuple10;import scala.Tuple2;import java.io.IOException;import java.util.ArrayList;import java.util.List;public class ReadHbase { private static String appName = "ReadTable"; public static void main(String[] args) { SparkConf sparkConf = new SparkConf(); //we can also run it at local:"local[3]" the number 3 means 3 threads sparkConf.setMaster("spark://master:7077").setAppName(appName); JavaSparkContext jsc = new JavaSparkContext(sparkConf); Configuration conf = HBaseConfiguration.create(); conf.set("hbase.zookeeper.quorum", "master"); conf.set("hbase.zookeeper.property.clientPort", "2181"); Scan scan = new Scan(); scan.addFamily(Bytes.toBytes("baseINFO")); scan.addColumn(Bytes.toBytes("baseINFO"), Bytes.toBytes("telphone_number")); scan.addColumn(Bytes.toBytes("baseINFO"), Bytes.toBytes("age")); scan.addColumn(Bytes.toBytes("baseINFO"), Bytes.toBytes("gender")); String scanToString = ""; try { ClientProtos.Scan proto = ProtobufUtil.toScan(scan); scanToString = Base64.encodeBytes(proto.toByteArray()); } catch (IOException io) { System.out.println(io); } for (int i = 0; i < 2; i++) { try { String tableName = "VIPUSER"; conf.set(TableInputFormat.INPUT_TABLE, tableName); conf.set(TableInputFormat.SCAN, scanToString); //get the Result of query from the Table of Hbase JavaPairRDD<ImmutableBytesWritable, Result> hBaseRDD = jsc.newAPIHadoopRDD(conf, TableInputFormat.class, ImmutableBytesWritable.class, Result.class); //group by row key like : [(Andy,110,21,0),(Tom,120,18,1)] JavaPairRDD<String, List<Integer>> art_scores = hBaseRDD.mapToPair( new PairFunction<Tuple2<ImmutableBytesWritable, Result>, String, List<Integer>>() { @Override public Tuple2<String, List<Integer>> call(Tuple2<ImmutableBytesWritable, Result> results) { List<Integer> list = new ArrayList<Integer>(); byte[] telphone_number = results._2().getValue(Bytes.toBytes("baseINFO"), Bytes.toBytes("telphone_number")); byte[] age = results._2().getValue(Bytes.toBytes("baseINFO"), Bytes.toBytes("age")); byte[] gender = results._2().getValue(Bytes.toBytes("baseINFO"), Bytes.toBytes("gender")); //the type of storage at Hbase is Byte Array,so we must let it be normal like Int,String and so on list.add(Integer.parseInt(Bytes.toString(telphone_number))); list.add(Integer.parseInt(Bytes.toString(age))); list.add(Integer.parseInt(Bytes.toString(gender))); return new Tuple2<String, List<Integer>>(Bytes.toString(results._1().get()), list); } } ); //switch to Cartesian product JavaPairRDD<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>> cart = art_scores.cartesian(art_scores); //use Row Key to delete the repetition from the last step "Cartesian product" JavaPairRDD<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>> cart2 = cart.filter( new Function<Tuple2<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>>, Boolean>() { public Boolean call(Tuple2<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>> tuple2Tuple2Tuple2) throws Exception { return tuple2Tuple2Tuple2._1()._1().compareTo(tuple2Tuple2Tuple2._2()._1()) < 0; } } ); System.out.println("Create the List 'collect'..."); //get the result we need List<Tuple2<Tuple2<String, List<Integer>>, Tuple2<String, List<Integer>>>> collect = cart2.collect(); System.out.println("Done.."); System.out.println(collect.size() > i ? collect.get(i):"STOP"); if (collect.size() > i ) break; } catch (Exception e) { System.out.println(e); } } } }
三、程序运行过程分析
1、spark自检以及Driver和excutor的启动过程
实例化一个SparkContext(若在spark2.x下,这里初始化的是一个SparkSession对象),这时候启动SecurityManager线程去检查用户权限,OK之后创建sparkDriver线程,spark底层远程通信模块(akka框架实现)启动并监听sparkDriver,之后由sparkEnv对象来注册BlockManagerMaster线程,由它的实现类对象去监测运行资源
2、zookeeper与Hbase的自检和启动
第一步顺利完成之后由sparkContext对象去实例去启动程序访问Hbase的入口,触发之后zookeeper完成自己的一系列自检活动,包括用户权限、操作系统、数据目录等,一切OK之后初始化客户端连接对象,之后由Hbase的ClientCnxn对象来建立与master的完整连接
3、spark job 的运行
程序开始调用spark的action类方法,比如这里调用了collect,会触发job的执行,这个流程网上资料很详细,无非就是DAGScheduler搞的一大堆事情,连带着出现一大堆线程,比如TaskSetManager、TaskScheduler等等,最后完成job,返回结果集
4、结束程序
正确返回结果集之后,sparkContext利用反射调用stop()方法,这之后也会触发一系列的stop操作,主要线程有这些:BlockManager,ShutdownHookManager,后面还有释放actor的操作等等,最后一切结束,临时数据和目录会被删除,资源会被释放
免责声明:本站发布的内容(图片、视频和文字)以原创、转载和分享为主,文章观点不代表本网站立场,如果涉及侵权请联系站长邮箱:is@yisu.com进行举报,并提供相关证据,一经查实,将立刻删除涉嫌侵权内容。