sqoop 是apache下用於RDBMS和HDFS互相導數據的工具。
本文檔是sqoop的使用實例，實現從mysql到hdfs互導數據，以及從Mysql導數據到HBase。
下載：
http://www.apache.org/dyn/closer.cgi/sqoop/
[zhouhh@Hadoop48 ~]$ wget http://labs.renren.com/apache-mirror/sqoop/1.4.1-incubating/sqoop-1.4.1-incubating__hadoop-1.0.0.tar.gz
最新用戶手冊
http://sqoop.apache.org/docs/1.4.1-incubating/SqoopUserGuide.html
一、從HBase庫中直接導出到mysql中？
一開始我想從HBase庫中直接導出到mysql中。
在mysql中創建一個庫和表
mysql> create database toplists; Query OK, 1 row affected (0.06 sec) mysql> use toplists Database changed mysql> create table t1(id int not null primary key, name varchar(255),value int); Query OK, 0 rows affected (0.10 sec) hbase(main):011:0> scan 't1' ROW COLUMN+CELL 1001 column=info:count, timestamp=1340265059531, value=724988 1009 column=info:count, timestamp=1340265059533, value=108051 ... total column=info:count, timestamp=1340265059534, value=833039 total_user_count column=info:, timestamp=1340266656307, value=154516 11 row(s) in 0.0420 seconds [zhouhh@Hadoop48 ~]$ sqoop list-tables --connect jdbc:mysql://localhost/toplists --username root java.lang.RuntimeException: Could not load db driver class: com.mysql.jdbc.Driver at org.apache.sqoop.manager.SqlManager.makeConnection(SqlManager.java:657) at org.apache.sqoop.manager.GenericJdbcManager.getConnection(GenericJdbcManager.java:52) at org.apache.sqoop.manager.SqlManager.execute(SqlManager.java:473) at org.apache.sqoop.manager.SqlManager.execute(SqlManager.java:496) at org.apache.sqoop.manager.SqlManager.getColumnTypesForRawQuery(SqlManager.java:194) at org.apache.sqoop.manager.SqlManager.getColumnTypes(SqlManager.java:178) at org.apache.sqoop.manager.ConnManager.getColumnTypes(ConnManager.java:114) at org.apache.sqoop.orm.ClassWriter.getColumnTypes(ClassWriter.java:1235) at org.apache.sqoop.orm.ClassWriter.generate(ClassWriter.java:1060) at org.apache.sqoop.tool.CodeGenTool.generateORM(CodeGenTool.java:82) at org.apache.sqoop.tool.ExportTool.exportTable(ExportTool.java:64) at org.apache.sqoop.tool.ExportTool.run(ExportTool.java:97) at org.apache.sqoop.Sqoop.run(Sqoop.java:145) at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:65) at org.apache.sqoop.Sqoop.runSqoop(Sqoop.java:181) at org.apache.sqoop.Sqoop.runTool(Sqoop.java:220) at org.apache.sqoop.Sqoop.runTool(Sqoop.java:229) at org.apache.sqoop.Sqoop.main(Sqoop.java:238) at com.cloudera.sqoop.Sqoop.main(Sqoop.java:57)
需下載 MySQL JDBC Connector 庫，並將其複製到$SQOOP_HOME/lib
下載mysql jdbc連接庫
地址：
http://www.mysql.com/downloads/connector/j/
[zhouhh@Hadoop48 ~]$ wget http://dev.mysql.com/get/Downloads/Connector-J/mysql-connector-java-5.1.21.tar.gz/from/http://cdn.mysql.com/ [zhouhh@Hadoop48 mysql-connector-java-5.1.21]$ cp mysql-connector-java-5.1.21-bin.jar ../sqoop/lib/. [zhouhh@Hadoop48 ~]$ sqoop list-tables --connect jdbc:mysql://localhost/toplists --username root t1 [zhouhh@Hadoop48 ~]$ sqoop-export --connect jdbc:mysql://localhost/toplists --username root --table t1 --export-dir /hbase java.io.IOException: com.mysql.jdbc.exceptions.jdbc4.CommunicationsException: Communications link failure The last packet sent successfully to the server was 0 milliseconds ago. The driver has not received any packets from the server. at org.apache.sqoop.mapreduce.ExportOutputFormat.getRecordWriter(ExportOutputFormat.java:79) at org.apache.hadoop.mapred.MapTask$NewDirectOutputCollector.<init>(MapTask.java:628) at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:753) at org.apache.hadoop.mapred.MapTask.run(MapTask.java:370) at org.apache.hadoop.mapred.Child$4.run(Child.java:255) at java.security.AccessController.doPrivileged(Native Method) at javax.security.auth.Subject.doAs(Subject.java:415) at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1121) at org.apache.hadoop.mapred.Child.main(Child.java:249) Caused by: com.mysql.jdbc.exceptions.jdbc4.CommunicationsException: Communications link failure
這是可能由jdbc版本引起的，換成5.1.18
[zhouhh@Hadoop48 ~]$ sqoop-export --connect jdbc:mysql://localhost:3306/toplists --username root --table t1 --export-dir /hbase Error initializing attempt_201206271529_0006_r_000000_0: org.apache.hadoop.util.DiskChecker$DiskErrorException: Could not find any valid local directory for ttprivate/taskTracker/zhouhh/jobcache/job_201206271529_0006/jobToken at org.apache.hadoop.fs.LocalDirAllocator$AllocatorPerContext.getLocalPathForWrite(LocalDirAllocator.java:381) at org.apache.hadoop.fs.LocalDirAllocator.getLocalPathForWrite(LocalDirAllocator.java:146) at org.apache.hadoop.fs.LocalDirAllocator.getLocalPathForWrite(LocalDirAllocator.java:127) at org.apache.hadoop.mapred.TaskTracker.localizeJobTokenFile(TaskTracker.java:4271) at org.apache.hadoop.mapred.TaskTracker.initializeJob(TaskTracker.java:1177) at org.apache.hadoop.mapred.TaskTracker.localizeJob(TaskTracker.java:1118) at org.apache.hadoop.mapred.TaskTracker$5.run(TaskTracker.java:2430) at java.lang.Thread.run(Thread.java:722)
DiskErrorException ，定位半天，發現是另一臺機器的空間滿了，在mapreduce運行時會引起該異常。
[zhouhh@Hadoop46 ~]$ df Filesystem 1K-blocks Used Available Use% Mounted on /dev/sda3 28337624 26877184 0 100% /
[zhouhh@Hadoop48 ~]$ sqoop-export --connect jdbc:mysql://192.168.10.48:3306/toplists --username root --table t1 --export-dir /hbase Caused by: java.sql.SQLException: null, message from server: "Host 'Hadoop47' is not allowed to connect to this MySQL server"
這是權限問題，設置授權：
mysql> GRANT ALL PRIVILEGES ON *.* TO '%'@'%';#允許所有用戶查看和修改databaseName數據庫模式的內容，否則別的IP連不上本MYSQL Query OK, 0 rows affected (0.06 sec)
這是測試，所以權限沒有限制。實際工作環境需謹慎授權。
[zhouhh@Hadoop48 ~]$ sqoop-export --connect jdbc:mysql://192.168.10.48:3306/toplists --username root --table t1 --export-dir /hbase Note: /tmp/sqoop-zhouhh/compile/fa1d1c042030b0ec8537c7a4cd02aab3/t1.java uses or overrides a deprecated API. Note: Recompile with -Xlint:deprecation for details. java.lang.NumberFormatException: For input string: "7" at java.lang.NumberFormatException.forInputString(NumberFormatException.java:65) at java.lang.Integer.parseInt(Integer.java:481) at java.lang.Integer.valueOf(Integer.java:582) at t1.__loadFromFields(t1.java:218) at t1.parse(t1.java:170) at org.apache.sqoop.mapreduce.TextExportMapper.map(TextExportMapper.java:77) at org.apache.sqoop.mapreduce.TextExportMapper.map(TextExportMapper.java:36) at org.apache.hadoop.mapreduce.Mapper.run(Mapper.java:144) at org.apache.sqoop.mapreduce.AutoProgressMapper.run(AutoProgressMapper.java:183) at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:764) at org.apache.hadoop.mapred.MapTask.run(MapTask.java:370) at org.apache.hadoop.mapred.Child$4.run(Child.java:255) at java.security.AccessController.doPrivileged(Native Method) at javax.security.auth.Subject.doAs(Subject.java:415) at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1121) at org.apache.hadoop.mapred.Child.main(Child.java:249)
這是由於/hbase是hbase的庫表，根本不是可以導的格式，所以報錯。
[zhouhh@Hadoop48 ~]$ sqoop-export --connect jdbc:mysql://192.168.10.48:3306/toplists --username root --table t1 --export-dir /hbase/t1 [zhouhh@Hadoop48 ~]$ sqoop-export --verbose --connect jdbc:mysql://192.168.10.48:3306/toplists --username root --table t1 --update-key id --input-fields-terminated-by '\t' --export-dir /hbase/t1 Note: /tmp/sqoop-zhouhh/compile/8ce6556eb13b3000550a9c864eaa6820/t1.java uses or overrides a deprecated API. Note: Recompile with -Xlint:deprecation for details. [zhouhh@Hadoop48 ~]$
但將導出目錄指到/hbase/t1表中，導出不會報錯，而mysql中沒有數據。後面才瞭解到，sqoop沒有直接從hbase中將表導出到mysql的辦法。必須先將hbase導出成平面文件，或者導出到hive中，纔可以用sqoop將數據導出到mysql。
二、從mysql中導到hdfs。
創建mysql表，將其導入到hdfs
mysql> create table test(id int not null primary key auto_increment,name varchar(64) not null,price decimal(10,2), cdate date,version int,comment varchar(255)); Query OK, 0 rows affected (0.10 sec) mysql> insert into test values(null,'iphone',3900.00,'2012-7-18',1,'8g'); Query OK, 1 row affected (0.04 sec) mysql> insert into test values(null,'ipad',3200.00,'2012-7-16',2,'16g'); Query OK, 1 row affected (0.00 sec) mysql> select * from test; +----+--------+---------+------------+---------+---------+ | id | name | price | cdate | version | comment | +----+--------+---------+------------+---------+---------+ | 1 | iphone | 3900.00 | 2012-07-18 | 1 | 8g | | 2 | ipad | 3200.00 | 2012-07-16 | 2 | 16g | +----+--------+---------+------------+---------+---------+ 2 rows in set (0.00 sec)
導入：
[zhouhh@Hadoop48 ~]$ sqoop import --connect jdbc:mysql://Hadoop48/toplists --table test -m 1 java.lang.RuntimeException: com.mysql.jdbc.exceptions.jdbc4.MySQLSyntaxErrorException: Access denied for user ''@'Hadoop48' to database 'toplists' at org.apache.sqoop.manager.CatalogQueryManager.getColumnNames(CatalogQueryManager.java:162)
給空用戶授權
mysql> GRANT ALL PRIVILEGES ON *.* TO ''@'%';
[zhouhh@Hadoop48 ~]$ sqoop import --connect jdbc:mysql://Hadoop48/toplists --username root --table test -m 1 12/07/18 11:10:16 INFO manager.MySQLManager: Preparing to use a MySQL streaming resultset. 12/07/18 11:10:16 INFO tool.CodeGenTool: Beginning code generation 12/07/18 11:10:16 INFO manager.SqlManager: Executing SQL statement: SELECT t.* FROM `index_mapping` AS t LIMIT 1 12/07/18 11:10:16 INFO orm.CompilationManager: HADOOP_HOME is /home/zhoulei/hadoop-1.0.0/libexec/.. 注: /tmp/sqoop-zhoulei/compile/2b04bdabb7043e4f75b215d72f65388e/index_mapping.java使用或覆蓋了已過時的 API。 注: 有關詳細信息, 請使用 -Xlint:deprecation 重新編譯。 12/07/18 11:10:18 INFO orm.CompilationManager: Writing jar file: /tmp/sqoop-zhoulei/compile/2b04bdabb7043e4f75b215d72f65388e/index_mapping.jar 12/07/18 11:10:18 WARN manager.MySQLManager: It looks like you are importing from mysql. 12/07/18 11:10:18 WARN manager.MySQLManager: This transfer can be faster! Use the --direct 12/07/18 11:10:18 WARN manager.MySQLManager: option to exercise a MySQL-specific fast path. 12/07/18 11:10:18 INFO manager.MySQLManager: Setting zero DATETIME behavior to convertToNull (mysql) 12/07/18 11:10:25 INFO mapreduce.ImportJobBase: Beginning import of index_mapping 12/07/18 11:10:26 INFO mapred.JobClient: Running job: job_201207101344_0519 12/07/18 11:10:27 INFO mapred.JobClient: map 0% reduce 0% 12/07/18 11:10:40 INFO mapred.JobClient: map 100% reduce 0% 12/07/18 11:10:45 INFO mapred.JobClient: Job complete: job_201207101344_0519 12/07/18 11:10:45 INFO mapred.JobClient: Counters: 18 12/07/18 11:10:45 INFO mapred.JobClient: Job Counters 12/07/18 11:10:45 INFO mapred.JobClient: SLOTS_MILLIS_MAPS=12083 12/07/18 11:10:45 INFO mapred.JobClient: Total time spent by all reduces waiting after reserving slots (ms)=0 12/07/18 11:10:45 INFO mapred.JobClient: Total time spent by all maps waiting after reserving slots (ms)=0 12/07/18 11:10:45 INFO mapred.JobClient: Launched map tasks=1 12/07/18 11:10:45 INFO mapred.JobClient: SLOTS_MILLIS_REDUCES=0 12/07/18 11:10:45 INFO mapred.JobClient: File Output Format Counters 12/07/18 11:10:45 INFO mapred.JobClient: Bytes Written=28 12/07/18 11:10:45 INFO mapred.JobClient: FileSystemCounters 12/07/18 11:10:45 INFO mapred.JobClient: HDFS_BYTES_READ=87 12/07/18 11:10:45 INFO mapred.JobClient: FILE_BYTES_WRITTEN=30396 12/07/18 11:10:45 INFO mapred.JobClient: HDFS_BYTES_WRITTEN=28 12/07/18 11:10:45 INFO mapred.JobClient: File Input Format Counters 12/07/18 11:10:45 INFO mapred.JobClient: Bytes Read=0 12/07/18 11:10:45 INFO mapred.JobClient: Map-Reduce Framework 12/07/18 11:10:45 INFO mapred.JobClient: Map input records=2 12/07/18 11:10:45 INFO mapred.JobClient: Physical memory (bytes) snapshot=79167488 12/07/18 11:10:45 INFO mapred.JobClient: Spilled Records=0 12/07/18 11:10:45 INFO mapred.JobClient: CPU time spent (ms)=340 12/07/18 11:10:45 INFO mapred.JobClient: Total committed heap usage (bytes)=56623104 12/07/18 11:10:45 INFO mapred.JobClient: Virtual memory (bytes) snapshot=955785216 12/07/18 11:10:45 INFO mapred.JobClient: Map output records=2 12/07/18 11:10:45 INFO mapred.JobClient: SPLIT_RAW_BYTES=87 12/07/18 11:10:45 INFO mapreduce.ImportJobBase: Transferred 28 bytes in 20.2612 seconds (1.382 bytes/sec) 12/07/18 11:10:45 INFO mapreduce.ImportJobBase: Retrieved 2 records.
檢查數據是否導入
[zhouhh@Hadoop48 ~]$ fs -cat /user/zhouhh/test/part-m-00000 1,iphone,3900.00,2012-07-18,1,8g 2,ipad,3200.00,2012-07-16,2,16g
或
[zhouhh@Hadoop48 ~]$ fs -cat test/part-m-00000 1,iphone,3900.00,2012-07-18,1,8g 2,ipad,3200.00,2012-07-16,2,16g
三、從hdfs導出到mysql
清空表
mysql> delete from test; Query OK, 2 rows affected (0.00 sec) mysql> select * from test; Empty set (0.00 sec)
導出
[zhouhh@Hadoop48 ~]$ sqoop-export --connect jdbc:mysql://192.168.10.48:3306/toplists --username root --table test --export-dir test Note: /tmp/sqoop-zhouhh/compile/7adaaa7ffe5f49ed9d794b1be8a9a983/test.java uses or overrides a deprecated API. Note: Recompile with -Xlint:deprecation for details.
導出時，–connect,–table, –export-dir是必須設置的。其中toplists是庫名，–table是該庫下的表名。 –export-dir是要導出的HDFS平面文件位置。如果不是絕對路徑，指/user/username/datadir
檢查mysql表
mysql> select * from test; +----+--------+---------+------------+---------+---------+ | id | name | price | cdate | version | comment | +----+--------+---------+------------+---------+---------+ | 1 | iphone | 3900.00 | 2012-07-18 | 1 | 8g | | 2 | ipad | 3200.00 | 2012-07-16 | 2 | 16g | +----+--------+---------+------------+---------+---------+ 2 rows in set (0.00 sec)
可見導出成功。
四、不執行mapreduce，但生成導入代碼
[zhouhh@Hadoop48 ~]$ sqoop codegen --connect jdbc:mysql://192.168.10.48:3306/toplists --username root --table test --class-name Mycodegen Note: /tmp/sqoop-zhouhh/compile/104b871487669b89dcd5b9b2c61f905f/Mycodegen.java uses or overrides a deprecated API. Note: Recompile with -Xlint:deprecation for details. [zhouhh@Hadoop48 ~]$ sqoop help codegen usage: sqoop codegen [GENERIC-ARGS] [TOOL-ARGS]
sqoop導入時，可以加選擇語句，以過濾和綜合多表，用–query.也可以只加條件，用–where。這樣可以不必每次導入整張表。 如 –where ‘id > 1000′
示例，採用join選擇多表數據：
sqoop import –query ‘select a.*,b.* from a join b on (a.id == b.id) where $conditions’ -m 1 –target-dir /usr/foo/joinresults
五、將mysql表導入到HBase
雖然目前，sqoop沒有將HBase直接導入mysql的辦法，但將mysql直接導入HBase是可以的。需指定–hbase-table，用–hbase-create-table來自動在HBase中創建表。–column-family指定列族名。–hbase-row-key指定rowkey對應的mysql的鍵。
[zhouhh@Hadoop48 ~]$ sqoop import –connect jdbc:mysql://Hadoop48/toplists –table test –hbase-table a –column-family name –hbase-row-key id –hbase-create-table –username ‘root’
檢查hbase被導入的表：
hbase(main):002:0> scan 'a' ROW COLUMN+CELL 1 column=name:cdate, timestamp=1342601695952, value=2012-07-18 1 column=name:comment, timestamp=1342601695952, value=8g 1 column=name:name, timestamp=1342601695952, value=iphone 1 column=name:price, timestamp=1342601695952, value=3900.00 1 column=name:version, timestamp=1342601695952, value=1 2 column=name:cdate, timestamp=1342601695952, value=2012-07-16 2 column=name:comment, timestamp=1342601695952, value=16g 2 column=name:name, timestamp=1342601695952, value=ipad 2 column=name:price, timestamp=1342601695952, value=3200.00 2 column=name:version, timestamp=1342601695952, value=2 2 row(s) in 0.2370 seconds
關於導入的一致性：建議停止mysql表的寫入再導入到HDFS或HIVE，否則，mapreduce可能會丟失新增的數據。
關於效率：mysql直接模式(–direct)導入的方式效率高。但不支持大對象數據，類型爲CLOB或BLOB的列。用JDBC效率較低，但有專用API可以支持CLOB及BLOB。
六、從HBase導出數據到Mysql
目前沒有直接的導出命令。但有兩個方法可以將HBase數據導出到mysql。
其一，將HBase導出成HDFS平面文件，再導出到mysql.
其二，將HBase數據導出到HIVE，再導出到mysql，參見後續文章《從hbase(hive)將數據導出到mysql》
