hive鏈接hbase外部表，向hbase中insert數據時報錯

繼上篇博客，hive鏈接hbase外部表成功後，能夠正常的查詢hbase的數據了。可是向hbase插入數據卻報錯了。

Error: java.lang.RuntimeException: java.lang.NoSuchMethodError: org.apache.hadoop.hbase.client.Put.setDurability(Lorg/apache/hadoop/hbase/client/Durability;)V
        at org.apache.hadoop.hive.ql.exec.mr.ExecMapper.map(ExecMapper.java:168)
        at org.apache.hadoop.mapred.MapRunner.run(MapRunner.java:54)
        at org.apache.hadoop.mapred.MapTask.runOldMapper(MapTask.java:453)
        at org.apache.hadoop.mapred.MapTask.run(MapTask.java:343)
        at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:164)
        at java.security.AccessController.doPrivileged(Native Method)
        at javax.security.auth.Subject.doAs(Subject.java:415)
        at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1657)
        at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:158)
Caused by: java.lang.NoSuchMethodError: org.apache.hadoop.hbase.client.Put.setDurability(Lorg/apache/hadoop/hbase/client/Durability;)V
        at org.apache.hadoop.hive.hbase.HiveHBaseTableOutputFormat$MyRecordWriter.write(HiveHBaseTableOutputFormat.java:142)
        at org.apache.hadoop.hive.hbase.HiveHBaseTableOutputFormat$MyRecordWriter.write(HiveHBaseTableOutputFormat.java:117)
        at org.apache.hadoop.hive.ql.io.HivePassThroughRecordWriter.write(HivePassThroughRecordWriter.java:40)
        at org.apache.hadoop.hive.ql.exec.FileSinkOperator.process(FileSinkOperator.java:743)
        at org.apache.hadoop.hive.ql.exec.Operator.forward(Operator.java:837)
        at org.apache.hadoop.hive.ql.exec.SelectOperator.process(SelectOperator.java:97)
        at org.apache.hadoop.hive.ql.exec.Operator.forward(Operator.java:837)
        at org.apache.hadoop.hive.ql.exec.TableScanOperator.process(TableScanOperator.java:115)
        at org.apache.hadoop.hive.ql.exec.MapOperator$MapOpCtx.forward(MapOperator.java:169)
        at org.apache.hadoop.hive.ql.exec.MapOperator.process(MapOperator.java:561)
        at org.apache.hadoop.hive.ql.exec.mr.ExecMapper.map(ExecMapper.java:159)
        ... 8 more

在網上查找問題，發現MapR解決過這個問題，參考頁面http://doc.mapr.com/display/components/Hive+Release+Notes;jsessionid=73C03B3BB0D8547A19E6CCEF80010D30#HiveReleaseNotes-Hive1.2.1-1601ReleaseNotes

的Hive 1.2.1-1601 Release Notes的說明，fe18d11的commit的說明中和我當前錯誤一致的。可是這個MapR公司對hive版本的一個patch，我用的是apache的版本，改到MapR的hive版本不現實啊，嘗試按照他補丁包中的updated，進行更新我本身的hive的jar包和配置，發現又引出新的問題。這種方式不行啊，得換條路了。

查看hive官方的發佈版本，當前有兩個版本apache-hive-1.2.1-bin.tar.gz和apache-hive-2.0.0-bin.tar.gz 。我當前是1.2.1的，我能夠升級到2.0.0試一下。

下載2.0.0版本並安裝，主要是修改了hive-site.xml文件（執行cp hive-default.xml.template hive-site.xml）。同時在hive/lib目錄下引入hbase的jar包：

guava-14.0.1.jar 
protobuf-java-2.5.0.jar 
hbase-client-1.1.1.jar 
hbase-common-1.1.1.jar 
zookeeper-3.4.6.jar 
hbase-server-1.1.1.jar

hive-site.xml

<property>
    <name>hive.exec.scratchdir</name>
    <value>/tmp/hive</value>
    <description>HDFS root scratch dir for Hive jobs which gets created with write all (733) permission. For each connecting user, an HDFS scratch dir: ${hive.exec.scratchdir}/&lt;username&gt; is created, with ${hive.scratch.dir.permission}.</description>
  </property>
  <property>
    <name>hive.exec.local.scratchdir</name>
    <value>/data/hive/logs</value>
    <description>Local scratch space for Hive jobs</description>
  </property>
  <property>
    <name>hive.downloaded.resources.dir</name>
    <value>/tmp/hive/temp0_resources</value>
    <description>Temporary local directory for added resources in the remote file system.</description>
  </property>
  
  ...
  
  <property>
    <name>javax.jdo.option.ConnectionPassword</name>
    <value>password</value>
    <description>password to use against metastore database</description>
  </property>
  <property>
    <name>javax.jdo.option.ConnectionURL</name>
    <value>jdbc:mysql://localhost:3306/hive_db?createDatabaseIfNotExist=true</value>
    <description>JDBC connect string for a JDBC metastore</description>
  </property>
  
  <property>
    <name>javax.jdo.option.ConnectionDriverName</name>
    <value>com.mysql.jdbc.Driver</value>
    <description>Driver class name for a JDBC metastore</description>
  </property>
  
  <property>
    <name>javax.jdo.option.ConnectionUserName</name>
    <value>username</value>
    <description>Username to use against metastore database</description>
  </property>
  
  <property>
    <name>hive.session.id</name>
    <value>temp0</value>
    <description/>
  </property>
  
  <property>
    <name>hive.aux.jars.path</name>
    <value>file:///data/hive/lib/guava-14.0.1.jar,file:///data/hive/lib/protobuf-java-2.5.0.jar,file:///data/hive/lib/hbase-client-1.1.1.jar,file:///data/hive/lib/hbase-common-1.1.1.jar,file:///data/hive/lib/zookeeper-3.4.6.jar,file:///data/hive/lib/hbase-server-1.1.1.jar</value>
    <description>The location of the plugin jars that contain implementations of user defined functions and serdes.</description>
  </property>
  
  <property>
    <name>hive.querylog.location</name>
    <value>/data/hive/logs</value>
    <description>Location of Hive run time structured log file</description>
  </property>
  
  <property>
    <name>hive.zookeeper.quorum</name>
    <value>slave1,slave2,master,slave4,slave5,slave6,slave7</value>
    <description>
      List of ZooKeeper servers to talk to. This is needed for: 
      1. Read/write locks - when hive.lock.manager is set to 
      org.apache.hadoop.hive.ql.lockmgr.zookeeper.ZooKeeperHiveLockManager, 
      2. When HiveServer2 supports service discovery via Zookeeper.
      3. For delegation token storage if zookeeper store is used, if
      hive.cluster.delegation.token.store.zookeeper.connectString is not set
    </description>
  </property>
  <property>
    <name>hbase.zookeeper.quorum</name>
    <value>slave1,slave2,master,slave4,slave5,slave6,slave7</value>
  </property>
  
  <property>
    <name>hive.server2.logging.operation.log.location</name>
    <value>/data/hive/logs/operation_logs</value>
    <description>Top level directory where operation logs are stored if logging functionality is enabled</description>
  </property>

老外真能寫，一個配置文件快5000行了，都改蒙圈了。

修改配置文件後，啓動metaStore的後臺進程，執行hive就可進入hive的命令行了，執行insert into table hbase的表，執行成功。歐耶！