zoukankan      html  css  js  c++  java
  • MapReduce_counter

     1 package MapReduce;
     2 
     3 import java.io.IOException;
     4 import java.net.URI;
     5 import java.net.URISyntaxException;
     6 import java.util.StringTokenizer;
     7 import org.apache.hadoop.conf.Configuration;
     8 import org.apache.hadoop.fs.FileSystem;
     9 import org.apache.hadoop.fs.Path;
    10 import org.apache.hadoop.io.LongWritable;
    11 import org.apache.hadoop.io.Text;
    12 import org.apache.hadoop.mapreduce.Counter;
    13 import org.apache.hadoop.mapreduce.Job;
    14 import org.apache.hadoop.mapreduce.Mapper;
    15 import org.apache.hadoop.mapreduce.Reducer;
    16 import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
    17 import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
    18 
    19 /**
    20  * mapreduce中计数器的使用
    21  *
    22  */
    23 public class WordCountApp {
    24     private static final String INPUT_PATH = "hdfs://h201:9000/user/hadoop/input";
    25     private static final String OUTPUT_PATH = "hdfs://h201:9000/user/hadoop/output";
    26 
    27     public static class MyMapper extends Mapper<LongWritable, Text, Text, LongWritable> {
    28         protected void map(LongWritable key, Text value, Context context) throws IOException, InterruptedException {
    29             final String line = value.toString();
    30             StringTokenizer tokenizer = new StringTokenizer(line);//StringTokenizer是字符串分隔解析类型,按空格截取交给takenizer这个容器
    31             final Counter counter = context.getCounter("Sensitive", "hello");//计数器,前面是技术器名字,后面是给谁计数
    32             if (value.toString().contains("hello")) {
    33                 counter.increment(1L);   //当查询到包含hello的词语时,计数器加1
    34             }
    35             while(tokenizer.hasMoreTokens()) {
    36                 String target = tokenizer.nextToken();//分隔符前面的输出给target
    37                 if(target.equals("hello") || target.equals("jiejie")){
    38                 context.write(new Text(target), new LongWritable(1));
    39                 }
    40             }
    41         }
    42     }
    43     
    44     public static class MyReducer extends Reducer<Text, LongWritable, Text, LongWritable> {
    45         @Override
    46         protected void reduce(Text key, Iterable<LongWritable> value,
    47             Reducer<Text, LongWritable, Text, LongWritable>.Context context) throws IOException, InterruptedException {
    48             long times = 0l;
    49             while (value.iterator().hasNext()) {
    50                 times += value.iterator().next().get();//迭代器累加给time
    51             }
    52             //if(times > 3 ){ //输出计数大于3的选项
    53             context.write(key, new LongWritable(times));
    54             //}
    55         }
    56         
    57     }
    58     public static void main(String[] args) throws IOException, URISyntaxException, ClassNotFoundException, InterruptedException {
    59         Configuration conf = new Configuration();
    60         conf.set("mapred.jar","wcapp.jar");//申明jar名字为wcapp.jar        
    61         //我们可以在代码中进行设置来自定义 key/value 输出分隔符:在主函数中添加如下一行代码:
    62         conf.set("mapred.textoutputformat.separator", ";"); //此处以”;“作为分割符
    63         final FileSystem fileSystem = FileSystem.get(new URI(OUTPUT_PATH), conf);//读路径信息
    64         fileSystem.delete(new Path(OUTPUT_PATH), true);//删除路径信息 输出路径不能存在
    65     
    66         final Job job = new Job(conf, WordCountApp.class.getSimpleName());
    67         job.setJarByClass(WordCountApp.class);//启job任务
    68     
    69         FileInputFormat.setInputPaths(job, INPUT_PATH);//输入  区别 引入位置变量new Path(args[0])直接换成路径,好处:执行过程中不用再给路径。坏处:不够灵活。
    70         job.setMapperClass(MyMapper.class);
    71         job.setMapOutputKeyClass(Text.class);
    72         job.setMapOutputValueClass(LongWritable.class);
    73         job.setCombinerClass(MyReducer.class);
    74         job.setReducerClass(MyReducer.class);
    75         job.setOutputKeyClass(Text.class);
    76         job.setOutputValueClass(LongWritable.class);
    77         FileOutputFormat.setOutputPath(job, new Path(OUTPUT_PATH));//输出
    78         System.exit(job.waitForCompletion(true) ? 0 : 1);
    79     }
    80 }

    StringTokenizer是字符串分隔解析类型,属于:java.util包。

    1.StringTokenizer的构造函数
    StringTokenizer(String str):构造一个用来解析str的StringTokenizer对象。java默认的分隔符是“空格”、“制表符(‘ ’)”、“换行符(‘ ’)”、“回车符(‘ ’)”。
    StringTokenizer(String str,String delim):构造一个用来解析str的StringTokenizer对象,并提供一个指定的分隔符。
    StringTokenizer(String str,String delim,boolean returnDelims):构造一个用来解析str的StringTokenizer对象,并提供一个指定的分隔符,同时,指定是否返回分隔符。
    2.StringTokenizer的一些常用方法
    说明:
    1.所有方法均为public;
    2.书写格式:[修饰符] <返回类型><方法名([参数列表])>
    int countTokens():返回nextToken方法被调用的次数。
    boolean hasMoreTokens():返回是否还有分隔符。
    boolean hasMoreElements():返回是否还有分隔符。
    String nextToken():返回从当前位置到下一个分隔符的字符串。
    Object nextElement():返回从当前位置到下一个分隔符的字符串。
    String nextToken(String delim):与4类似,以指定的分隔符返回结果。

    [hadoop@h201 counter]$ /usr/jdk1.7.0_25/bin/javac WordCountApp.java
    Note: WordCountApp.java uses or overrides a deprecated API.
    Note: Recompile with -Xlint:deprecation for details.
    [hadoop@h201 counter]$ /usr/jdk1.7.0_25/bin/jar cvf wcapp.jar WordCountApp*class
    added manifest
    adding: WordCountApp.class(in = 2358) (out= 1191)(deflated 49%)
    adding: WordCountApp$MyMapper.class(in = 2019) (out= 885)(deflated 56%)
    adding: WordCountApp$MyReducer.class(in = 1655) (out= 691)(deflated 58%)
    [hadoop@h201 counter]$ hadoop jar wcapp.jar WordCountApp
    18/03/11 23:11:09 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
    18/03/11 23:11:10 INFO client.RMProxy: Connecting to ResourceManager at h201/192.168.121.132:8032
    18/03/11 23:11:10 WARN mapreduce.JobResourceUploader: Hadoop command-line option parsing not performed. Implement the Tool interface and execute your application with ToolRunner to remedy this.
    18/03/11 23:11:10 INFO input.FileInputFormat: Total input paths to process : 2
    18/03/11 23:11:11 INFO mapreduce.JobSubmitter: number of splits:2
    18/03/11 23:11:11 INFO Configuration.deprecation: mapred.jar is deprecated. Instead, use mapreduce.job.jar
    18/03/11 23:11:11 INFO Configuration.deprecation: mapred.textoutputformat.separator is deprecated. Instead, use mapreduce.output.textoutputformat.separator
    18/03/11 23:11:11 INFO mapreduce.JobSubmitter: Submitting tokens for job: job_1516635595760_0006
    18/03/11 23:11:11 INFO impl.YarnClientImpl: Submitted application application_1516635595760_0006
    18/03/11 23:11:11 INFO mapreduce.Job: The url to track the job: http://h201:8088/proxy/application_1516635595760_0006/
    18/03/11 23:11:11 INFO mapreduce.Job: Running job: job_1516635595760_0006
    18/03/11 23:11:20 INFO mapreduce.Job: Job job_1516635595760_0006 running in uber mode : false
    18/03/11 23:11:20 INFO mapreduce.Job:  map 0% reduce 0%
    18/03/11 23:11:26 INFO mapreduce.Job:  map 50% reduce 0%
    18/03/11 23:11:37 INFO mapreduce.Job:  map 100% reduce 0%
    18/03/11 23:11:38 INFO mapreduce.Job:  map 100% reduce 100%
    18/03/11 23:11:38 INFO mapreduce.Job: Job job_1516635595760_0006 completed successfully
    18/03/11 23:11:38 INFO mapreduce.Job: Counters: 50
            File System Counters
                    FILE: Number of bytes read=39
                    FILE: Number of bytes written=329603
                    FILE: Number of read operations=0
                    FILE: Number of large read operations=0
                    FILE: Number of write operations=0
                    HDFS: Number of bytes read=914
                    HDFS: Number of bytes written=19
                    HDFS: Number of read operations=9
                    HDFS: Number of large read operations=0
                    HDFS: Number of write operations=2
            Job Counters
                    Launched map tasks=2
                    Launched reduce tasks=1
                    Data-local map tasks=2
                    Total time spent by all maps in occupied slots (ms)=18964
                    Total time spent by all reduces in occupied slots (ms)=5647
                    Total time spent by all map tasks (ms)=18964
                    Total time spent by all reduce tasks (ms)=5647
                    Total vcore-seconds taken by all map tasks=18964
                    Total vcore-seconds taken by all reduce tasks=5647
                    Total megabyte-seconds taken by all map tasks=19419136
                    Total megabyte-seconds taken by all reduce tasks=5782528
            Map-Reduce Framework
                    Map input records=54
                    Map output records=35
                    Map output bytes=507
                    Map output materialized bytes=45
                    Input split bytes=227
                    Combine input records=35
                    Combine output records=2
                    Reduce input groups=2
                    Reduce shuffle bytes=45
                    Reduce input records=2
                    Reduce output records=2
                    Spilled Records=4
                    Shuffled Maps =2
                    Failed Shuffles=0
                    Merged Map outputs=2
                    GC time elapsed (ms)=584
                    CPU time spent (ms)=2380
                    Physical memory (bytes) snapshot=387678208
                    Virtual memory (bytes) snapshot=3221241856
                    Total committed heap usage (bytes)=257499136
            Sensitive
                    hello=18
            Shuffle Errors
                    BAD_ID=0
                    CONNECTION=0
                    IO_ERROR=0
                    WRONG_LENGTH=0
                    WRONG_MAP=0
                    WRONG_REDUCE=0
            File Input Format Counters
                    Bytes Read=687
            File Output Format Counters
                    Bytes Written=19

  • 相关阅读:
    atitit.  web组件化原理与设计
    Atitit.git的存储结构and 追踪
    Atitit.git的存储结构and 追踪
    atitit.atiHtmlUi web组件化方案与规范v1
    atitit.atiHtmlUi web组件化方案与规范v1
    Atitit.设计模式-----触发器模式 trigger  详解
    Atitit.设计模式-----触发器模式 trigger  详解
    Atitit.web ui  组件化 vs  mvc
    Atitit.web ui  组件化 vs  mvc
    Atitit..css的体系结构
  • 原文地址:https://www.cnblogs.com/jieran/p/8546957.html
Copyright © 2011-2022 走看看