Hadoop 中利用 mapreduce 读写 mysql 数据

转载 2013年12月04日 17:54:46

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 `namevarchar(10) DEFAULT NULL
04 ) ENGINE=InnoDB DEFAULT CHARSET=utf8;
05  
06 CREATE TABLE `t2` (
07 `id` int DEFAULT NULL,
08 `namevarchar(10) DEFAULT NULL
09 ) ENGINE=InnoDB DEFAULT CHARSET=utf8;
10  
11 insert into 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(1this.id);
059             stmt.setString(2this.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