Hadoop 中利用 mapreduce 读写 mysql 数据

有时候我们在项目中会遇到输入结果集很大，但是输出结果很小，比如一些 pv、uv 数据，然后为了实时查询的需求，或者一些 OLAP 的需求，我们需要 mapreduce 与 mysql 进行数据的交互，而这些特性正是 hbase 或者 hive 目前亟待改进的地方。

好了言归正传，简单的说说背景、原理以及需要注意的地方：

1、为了方便 MapReduce 直接访问关系型数据库（Mysql,Oracle），Hadoop提供了DBInputFormat和DBOutputFormat两个类。通过DBInputFormat类把数据库表数据读入到HDFS，根据DBOutputFormat类把MapReduce产生的结果集导入到数据库表中。

2、由于0.20版本对DBInputFormat和DBOutputFormat支持不是很好，该例用了0.19版本来说明这两个类的用法。

至少在我的 0.20.203 中的 org.apache.hadoop.mapreduce.lib 下是没见到 db 包，所以本文也是以老版的 API 来为例说明的。

3、运行MapReduce时候报错：java.io.IOException: com.mysql.jdbc.Driver，一般是由于程序找不到mysql驱动包。解决方法是让每个tasktracker运行MapReduce程序时都可以找到该驱动包。

添加包有两种方式：

（1）在每个节点下的${HADOOP_HOME}/lib下添加该包。重启集群，一般是比较原始的方法。

（2）a)把包传到集群上： hadoop fs -put mysql-connector-java-5.1.0- bin.jar /hdfsPath/

       b)在mr程序提交job前，添加语句：DistributedCache.addFileToClassPath(new Path(“/hdfsPath/mysql- connector-java- 5.1.0-bin.jar”), conf);

（3）虽然API用的是0.19的，但是使用0.20的API一样可用，只是会提示方法已过时而已。

4、测试数据：

01	CREATE TABLE `t` (

02	`id` int DEFAULT NULL,

03	`name` varchar(10) DEFAULT NULL

04	) ENGINE=InnoDB DEFAULT CHARSET=utf8;

05

06	CREATE TABLE `t2` (

07	`id` int DEFAULT NULL,

08	`name` varchar(10) DEFAULT NULL

09	) ENGINE=InnoDB DEFAULT CHARSET=utf8;

10

11	insert into t values (1,"june"),(2,"decli"),(3,"hello"),

12	(4,"june"),(5,"decli"),(6,"hello"),(7,"june"),

13	(8,"decli"),(9,"hello"),(10,"june"),

14	(11,"june"),(12,"decli"),(13,"hello");

5、代码：

001	import java.io.DataInput;

002	import java.io.DataOutput;

003	import java.io.IOException;

004	import java.sql.PreparedStatement;

005	import java.sql.ResultSet;

006	import java.sql.SQLException;

007	import java.util.Iterator;

008

009	import org.apache.hadoop.filecache.DistributedCache;

010	import org.apache.hadoop.fs.Path;

011	import org.apache.hadoop.io.LongWritable;

012	import org.apache.hadoop.io.Text;

013	import org.apache.hadoop.io.Writable;

014	import org.apache.hadoop.mapred.JobClient;

015	import org.apache.hadoop.mapred.JobConf;

016	import org.apache.hadoop.mapred.MapReduceBase;

017	import org.apache.hadoop.mapred.Mapper;

018	import org.apache.hadoop.mapred.OutputCollector;

019	import org.apache.hadoop.mapred.Reducer;

020	import org.apache.hadoop.mapred.Reporter;

021	import org.apache.hadoop.mapred.lib.IdentityReducer;

022	import org.apache.hadoop.mapred.lib.db.DBConfiguration;

023	import org.apache.hadoop.mapred.lib.db.DBInputFormat;

024	import org.apache.hadoop.mapred.lib.db.DBOutputFormat;

025	import org.apache.hadoop.mapred.lib.db.DBWritable;

026

027

/**

028	* Function: 测试 mr 与 mysql 的数据交互，此测试用例将一个表中的数据复制到另一张表中

029	*           实际当中，可能只需要从 mysql 读，或者写到 mysql 中。

030	* date: 2013-7-29 上午2:34:04 <br/>

031	* @author june

032

*/

033	public class Mysql2Mr {

034	// DROP TABLE IF EXISTS `hadoop`.`studentinfo`;

035	// CREATE TABLE studentinfo (

036	// id INTEGER NOT NULL PRIMARY KEY,

037	// name VARCHAR(32) NOT NULL);

038

039	public static class StudentinfoRecord implements Writable, DBWritable {

040

int id;

041	String name;

042

043	public StudentinfoRecord() {

044

045

}

046

047	public void readFields(DataInput in) throws IOException {

048	this.id = in.readInt();

049	this.name = Text.readString(in);

050

}

051

052	public String toString() {

053	return new String(this.id + " " + this.name);

054

}

055

056

@Override

057	public void write(PreparedStatement stmt) throws SQLException {

058	stmt.setInt(1, this.id);

059	stmt.setString(2, this.name);

060

}

061

062

@Override

063	public void readFields(ResultSet result) throws SQLException {

064	this.id = result.getInt(1);

065	this.name = result.getString(2);

066

}

067

068

@Override

069	public void write(DataOutput out) throws IOException {

070	out.writeInt(this.id);

071	Text.writeString(out, this.name);

072

}

073

}

074

075	// 记住此处是静态内部类，要不然你自己实现无参构造器，或者等着抛异常：

076	// Caused by: java.lang.NoSuchMethodException: DBInputMapper.<init>()

077	// http://stackoverflow.com/questions/7154125/custom-mapreduce-input-format-cant-find-constructor

078	// 网上脑残式的转帖，没见到一个写对的。。。

079	public static class DBInputMapper extends MapReduceBase implements

080	Mapper<LongWritable, StudentinfoRecord, LongWritable, Text> {

081	public void map(LongWritable key, StudentinfoRecord value,

082	OutputCollector<LongWritable, Text> collector, Reporter reporter) throws IOException {

083	collector.collect(new LongWritable(value.id), new Text(value.toString()));

084

}

085

}

086

087	public static class MyReducer extends MapReduceBase implements

088	Reducer<LongWritable, Text, StudentinfoRecord, Text> {

089

@Override

090	public void reduce(LongWritable key, Iterator<Text> values,

091	OutputCollector<StudentinfoRecord, Text> output, Reporter reporter) throws IOException {

092	String[] splits = values.next().toString().split(" ");

093	StudentinfoRecord r = new StudentinfoRecord();

094	r.id = Integer.parseInt(splits[0]);

095	r.name = splits[1];

096	output.collect(r, new Text(r.name));

097

}

098

}

099

100	public static void main(String[] args) throws IOException {

101	JobConf conf = new JobConf(Mysql2Mr.class);

102	DistributedCache.addFileToClassPath(new Path("/tmp/mysql-connector-java-5.0.8-bin.jar"), conf);

103

104	conf.setMapOutputKeyClass(LongWritable.class);

105	conf.setMapOutputValueClass(Text.class);

106	conf.setOutputKeyClass(LongWritable.class);

107	conf.setOutputValueClass(Text.class);

108

109	conf.setOutputFormat(DBOutputFormat.class);

110	conf.setInputFormat(DBInputFormat.class);

111	// // mysql to hdfs

112	// conf.setReducerClass(IdentityReducer.class);

113	// Path outPath = new Path("/tmp/1");

114	// FileSystem.get(conf).delete(outPath, true);

115	// FileOutputFormat.setOutputPath(conf, outPath);

116

117	DBConfiguration.configureDB(conf, "com.mysql.jdbc.Driver", "jdbc:mysql://192.168.1.101:3306/test",

118	"root", "root");

119	String[] fields = { "id", "name" };

120	// 从 t 表读数据

121	DBInputFormat.setInput(conf, StudentinfoRecord.class, "t", null, "id", fields);

122	// mapreduce 将数据输出到 t2 表

123	DBOutputFormat.setOutput(conf, "t2", "id", "name");

124	// conf.setMapperClass(org.apache.hadoop.mapred.lib.IdentityMapper.class);

125	conf.setMapperClass(DBInputMapper.class);

126	conf.setReducerClass(MyReducer.class);

127

128	JobClient.runJob(conf);

129

}

130

}

6、结果：

执行两次后，你可以看到mysql结果：

01	mysql> select * from t2;

02	+------+-------+

03	| id   | name  |

04	+------+-------+

05	|    1 | june  |

06	|    2 | decli |

07	|    3 | hello |

08	|    4 | june  |

09	|    5 | decli |

10	|    6 | hello |

11	|    7 | june  |

12	|    8 | decli |

13	|    9 | hello |

14	|   10 | june  |

15	|   11 | june  |

16	|   12 | decli |

17	|   13 | hello |

18	|    1 | june  |

19	|    2 | decli |

20	|    3 | hello |

21	|    4 | june  |

22	|    5 | decli |

23	|    6 | hello |

24	|    7 | june  |

25	|    8 | decli |

26	|    9 | hello |

27	|   10 | june  |

28	|   11 | june  |

29	|   12 | decli |

30	|   13 | hello |

31	+------+-------+

32	26 rows in set (0.00 sec)

33

34

mysql>

7、日志：

01	13/07/29 02:33:03 WARN mapred.JobClient: Use GenericOptionsParser for parsing the arguments. Applications should implement Tool for the same.

02	13/07/29 02:33:03 INFO filecache.TrackerDistributedCacheManager: Creating mysql-connector-java-5.0.8-bin.jar in /tmp/hadoop-june/mapred/local/archive/-8943686319031389138_-1232673160_640840668/192.168.1.101/tmp-work--8372797484204470322 with rwxr-xr-x

03	13/07/29 02:33:03 INFO filecache.TrackerDistributedCacheManager: Cached hdfs://192.168.1.101:9000/tmp/mysql-connector-java-5.0.8-bin.jar as /tmp/hadoop-june/mapred/local/archive/-8943686319031389138_-1232673160_640840668/192.168.1.101/tmp/mysql-connector-java-5.0.8-bin.jar

04	13/07/29 02:33:03 INFO filecache.TrackerDistributedCacheManager: Cached hdfs://192.168.1.101:9000/tmp/mysql-connector-java-5.0.8-bin.jar as /tmp/hadoop-june/mapred/local/archive/-8943686319031389138_-1232673160_640840668/192.168.1.101/tmp/mysql-connector-java-5.0.8-bin.jar

05	13/07/29 02:33:03 INFO mapred.JobClient: Running job: job_local_0001

06	13/07/29 02:33:03 INFO mapred.MapTask: numReduceTasks: 1

07	13/07/29 02:33:03 INFO mapred.MapTask: io.sort.mb = 100

08	13/07/29 02:33:03 INFO mapred.MapTask: data buffer = 79691776/99614720

09	13/07/29 02:33:03 INFO mapred.MapTask: record buffer = 262144/327680

10	13/07/29 02:33:03 INFO mapred.MapTask: Starting flush of map output

11	13/07/29 02:33:03 INFO mapred.MapTask: Finished spill 0

12	13/07/29 02:33:03 INFO mapred.Task: Task:attempt_local_0001_m_000000_0 is done. And is in the process of commiting

13	13/07/29 02:33:04 INFO mapred.JobClient:  map 0% reduce 0%

14	13/07/29 02:33:06 INFO mapred.LocalJobRunner:

15	13/07/29 02:33:06 INFO mapred.Task: Task 'attempt_local_0001_m_000000_0' done.

16	13/07/29 02:33:06 INFO mapred.LocalJobRunner:

17	13/07/29 02:33:06 INFO mapred.Merger: Merging 1 sorted segments

18	13/07/29 02:33:06 INFO mapred.Merger: Down to the last merge-pass, with 1 segments left of total size: 235 bytes

19	13/07/29 02:33:06 INFO mapred.LocalJobRunner:

20	13/07/29 02:33:06 INFO mapred.Task: Task:attempt_local_0001_r_000000_0 is done. And is in the process of commiting

21	13/07/29 02:33:07 INFO mapred.JobClient:  map 100% reduce 0%

22	13/07/29 02:33:09 INFO mapred.LocalJobRunner: reduce > reduce

23	13/07/29 02:33:09 INFO mapred.Task: Task 'attempt_local_0001_r_000000_0' done.

24	13/07/29 02:33:09 WARN mapred.FileOutputCommitter: Output path is null in cleanup

25	13/07/29 02:33:10 INFO mapred.JobClient:  map 100% reduce 100%

26	13/07/29 02:33:10 INFO mapred.JobClient: Job complete: job_local_0001

27	13/07/29 02:33:10 INFO mapred.JobClient: Counters: 18

28	13/07/29 02:33:10 INFO mapred.JobClient:   File Input Format Counters

29	13/07/29 02:33:10 INFO mapred.JobClient:     Bytes Read=0

30	13/07/29 02:33:10 INFO mapred.JobClient:   File Output Format Counters

31	13/07/29 02:33:10 INFO mapred.JobClient:     Bytes Written=0

32	13/07/29 02:33:10 INFO mapred.JobClient:   FileSystemCounters

33	13/07/29 02:33:10 INFO mapred.JobClient:     FILE_BYTES_READ=1211691

34	13/07/29 02:33:10 INFO mapred.JobClient:     HDFS_BYTES_READ=1081704

35	13/07/29 02:33:10 INFO mapred.JobClient:     FILE_BYTES_WRITTEN=2392844

36	13/07/29 02:33:10 INFO mapred.JobClient:   Map-Reduce Framework

37	13/07/29 02:33:10 INFO mapred.JobClient:     Map output materialized bytes=239

38	13/07/29 02:33:10 INFO mapred.JobClient:     Map input records=13

39	13/07/29 02:33:10 INFO mapred.JobClient:     Reduce shuffle bytes=0

40	13/07/29 02:33:10 INFO mapred.JobClient:     Spilled Records=26

41	13/07/29 02:33:10 INFO mapred.JobClient:     Map output bytes=207

42	13/07/29 02:33:10 INFO mapred.JobClient:     Map input bytes=13

43	13/07/29 02:33:10 INFO mapred.JobClient:     SPLIT_RAW_BYTES=75

44	13/07/29 02:33:10 INFO mapred.JobClient:     Combine input records=0

45	13/07/29 02:33:10 INFO mapred.JobClient:     Reduce input records=13

46	13/07/29 02:33:10 INFO mapred.JobClient:     Reduce input groups=13

47	13/07/29 02:33:10 INFO mapred.JobClient:     Combine output records=0

48	13/07/29 02:33:10 INFO mapred.JobClient:     Reduce output records=13

49	13/07/29 02:33:10 INFO mapred.JobClient:     Map output records=13

8、REF：

新版 API 写法：

http://superlxw1234.iteye.com/blog/1880712

老版：

http://blog.csdn.net/dajuezhao/article/details/5799371

http://www.zhengmenbb.com/archives/583.htm