《Hadoop The Definitive Guide》ch05 Developing a MapReduce Application

1. 介绍

MapReduce应用开发包含特定的流程。首先，编写map和reduce函数，最好能进行单元测试以保证它们能如期运行。然后写一个驱动程序来运行作业，可以使用数据集中的少量数据从IDE运行，看它是否能够正常运行。

2. GenericOptionsParser, Tool和ToolRunner

[ate: /local/nomad2/hadoop/tomwhite-hadoop-book-32dae01 ]
>> hadoop ConfigurationPrinter -conf conf/hadoop-localhost.xml |grep mapred.job.tracker=
mapred.job.tracker=localhost:8021

3. 问题：老是抱怨找不到class?
解决办法:
1. stop-all.sh
2. rm -rf /tmp/hadoop-nomad2/*
3. hadoop namenode -format
4. start-all.sh
5. jps 确认datanode进程起来
6. 重新运行程序，注意这里的jar文件是在HADOOP_CLASSPATH中的，而不是在hdfs中。

[ate: /local/nomad2/hadoop/tomwhite-hadoop-book-32dae01 ]
>> hadoop jar ch05.jar v3.MaxTemperatureDriver input/ncdc/all max-temp
12/07/03 01:33:40 INFO mapred.FileInputFormat: Total input paths to process : 2
12/07/03 01:33:40 INFO mapred.JobClient: Running job: job_201207030133_0001
12/07/03 01:33:41 INFO mapred.JobClient:  map 0% reduce 0%
12/07/03 01:33:55 INFO mapred.JobClient:  map 100% reduce 0%
12/07/03 01:34:07 INFO mapred.JobClient:  map 100% reduce 100%
12/07/03 01:34:12 INFO mapred.JobClient: Job complete: job_201207030133_0001
12/07/03 01:34:12 INFO mapred.JobClient: Counters: 26
12/07/03 01:34:12 INFO mapred.JobClient:   Job Counters 
12/07/03 01:34:12 INFO mapred.JobClient:     Launched reduce tasks=1
12/07/03 01:34:12 INFO mapred.JobClient:     SLOTS_MILLIS_MAPS=16347
12/07/03 01:34:12 INFO mapred.JobClient:     Total time spent by all reduces waiting after reserving slots (ms)=0
12/07/03 01:34:12 INFO mapred.JobClient:     Total time spent by all maps waiting after reserving slots (ms)=0
12/07/03 01:34:12 INFO mapred.JobClient:     Launched map tasks=2
12/07/03 01:34:12 INFO mapred.JobClient:     Data-local map tasks=2
12/07/03 01:34:12 INFO mapred.JobClient:     SLOTS_MILLIS_REDUCES=10004
12/07/03 01:34:12 INFO mapred.JobClient:   File Input Format Counters 
12/07/03 01:34:12 INFO mapred.JobClient:     Bytes Read=147972
12/07/03 01:34:12 INFO mapred.JobClient:   File Output Format Counters 
12/07/03 01:34:12 INFO mapred.JobClient:     Bytes Written=18
12/07/03 01:34:12 INFO mapred.JobClient:   FileSystemCounters
12/07/03 01:34:12 INFO mapred.JobClient:     FILE_BYTES_READ=28
12/07/03 01:34:12 INFO mapred.JobClient:     HDFS_BYTES_READ=148184
12/07/03 01:34:12 INFO mapred.JobClient:     FILE_BYTES_WRITTEN=62923
12/07/03 01:34:12 INFO mapred.JobClient:     HDFS_BYTES_WRITTEN=18
12/07/03 01:34:12 INFO mapred.JobClient:   Map-Reduce Framework
12/07/03 01:34:12 INFO mapred.JobClient:     Map output materialized bytes=34
12/07/03 01:34:12 INFO mapred.JobClient:     Map input records=13130
12/07/03 01:34:12 INFO mapred.JobClient:     Reduce shuffle bytes=34
12/07/03 01:34:12 INFO mapred.JobClient:     Spilled Records=4
12/07/03 01:34:12 INFO mapred.JobClient:     Map output bytes=118161
12/07/03 01:34:12 INFO mapred.JobClient:     Map input bytes=1777168
12/07/03 01:34:12 INFO mapred.JobClient:     Combine input records=13129
12/07/03 01:34:12 INFO mapred.JobClient:     SPLIT_RAW_BYTES=212
12/07/03 01:34:12 INFO mapred.JobClient:     Reduce input records=2
12/07/03 01:34:12 INFO mapred.JobClient:     Reduce input groups=2
12/07/03 01:34:12 INFO mapred.JobClient:     Combine output records=2
12/07/03 01:34:12 INFO mapred.JobClient:     Reduce output records=2
12/07/03 01:34:12 INFO mapred.JobClient:     Map output records=13129

4. MapReduce web用户界面

http://server:50030

Job的详细信息，