本文簡單記錄一下hive的安裝配置步驟
tar -xzvf hive-0.11.0-bin.tar.gz

在~home目錄中配置path變數將hive bin目錄加入path
export PATH=$PATH:/home/hadoop/hadoop-0.20.2/bin:/home/hadoop/hbase-0.92.1/bin:/home/hadoop/zookeeper-3.4.5/bin:/home/hadoop/hive-0.11.0-bin/bin

配置Hive
進入hive/conf目錄
依據hive-env.sh.template，建立hive-env.sh檔案
cp  hive-env.sh.template hive-env.sh
修改hive-env.sh
指定hive配置檔案的路徑
export HIVE_CONF_DIR=/home/hadoop/hive-0.11.0-bin/conf
指定Hadoop路徑
HADOOP_HOME=/home/hadoop/hadoop-0.20.2  

配置HADOOP_HOME後，Hive將資料倉庫檔案儲存在HDFS檔案系統；metastore儲存在Hive自帶的嵌入式資料庫Derby中。
在HDFS檔案系統中建立如下目錄
hadoop fs -mkdir /tmp
hadoop fs -mkdir /user/hive/warehouse
hadoop fs -chmod g+w /tmp
hadoop fs -chmod g+w /user/hive/warehouse

測試執行hive
hive> create table test(key string);
hive> show tables;

在本地檔案系統建立一個如下的文字檔案：/home/hadoop/hiveimpt/student.txt
C01,N0101,82
C01,N0102,59
C01,N0103,65
C02,N0201,81
C02,N0202,82
C02,N0203,79
C03,N0301,56
C03,N0302,92
C03,N0306,72
建立一個hive的表
create table student(classNo string, stuNo string, score int) row format delimited fields terminated by ',';
將資料load到表中
load data local inpath '/home/hadoop/hiveimpt/student.txt'overwrite into table student;

這個命令將student.txt檔案內容載入到表student中。這個載入操作將直接把student.txt檔案複製到hive的warehouse目錄中，這個目錄由hive.metastore.warehouse.dir配置項設定，預設值為/user/hive/warehouse。Overwrite選項將導致Hive事先刪除student目錄下所有的檔案。
Hive不會對student.txt做任何格式處理，因為Hive本身並不強調資料的儲存格式。
此例中，Hive將資料儲存於HDFS系統中。當然，Hive也可以將資料儲存於本地。

可以執行類似SQL的命令:
hive> select * from student;

hive> Select classNo,count(score) from student where score>=60 group by classNo;
Total MapReduce jobs = 1
Launching Job 1 out of 1
Number of reduce tasks not specified. Estimated from input data size: 1
In order to change the average load for a reducer (in bytes):
  set hive.exec.reducers.bytes.per.reducer=<number>
In order to limit the maximum number of reducers:
  set hive.exec.reducers.max=<number>
In order to set a constant number of reducers:
  set mapred.reduce.tasks=<number>
Starting Job = job_201403182323_0001, Tracking URL = http://hadoopsrv:50030/jobdetails.jsp?jobid=job_201403182323_0001
Kill Command = /home/hadoop/hadoop-0.20.2/bin/../bin/hadoop job  -kill job_201403182323_0001
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 1
2014-03-19 00:42:27,572 Stage-1 map = 0%,  reduce = 0%
2014-03-19 00:42:34,861 Stage-1 map = 100%,  reduce = 0%
2014-03-19 00:42:43,972 Stage-1 map = 100%,  reduce = 33%
2014-03-19 00:42:47,029 Stage-1 map = 100%,  reduce = 100%
Ended Job = job_201403182323_0001
MapReduce Jobs Launched:
Job 0: Map: 1  Reduce: 1   HDFS Read: 116 HDFS Write: 18 SUCCESS
Total MapReduce CPU Time Spent: 0 msec
OK
C01     2
C02     3
C03     2
Time taken: 38.263 seconds, Fetched: 3 row(s)

由此看見，HiveQL的使用和SQL及其類似。我們用到了group和count，其實在後臺Hive將這些操作都轉換成了MapReduce操作提交給Hadoop執行，並最終輸出結果。
用安裝在另外一臺機器的mysql儲存metastore：
在mysql上建立db和使用者
mysql>create database hive;
mysql> CREATE USER 'hive' IDENTIFIED BY 'hive';
mysql> GRANT ALL PRIVILEGES ON *.* TO 'hive'@'%' WITH GRANT OPTION;
mysql> flush privileges;
在hive的機器上驗證連線：
mysql -h 192.168.2.3 -u hive -phive

在hive的機器上配置hive-site.xml
    <property>
        <name>hive.metastore.local</name>
        <value>true</value>
    </property>
    <property>
        <name>javax.jdo.option.ConnectionURL</name>
        <value>jdbc:mysql://192.168.2.3:3306/hive?characterEncoding=UTF-8</value>
    </property>
    <property>
        <name>javax.jdo.option.ConnectionDriverName</name>
        <value>com.mysql.jdbc.Driver</value>
    </property>
    <property>
        <name>javax.jdo.option.ConnectionUserName</name>
        <value>hive</value>
    </property>
    <property>
        <name>javax.jdo.option.ConnectionPassword</name>
        <value>hive</value>
    </property>

常見異常：    
com.mysql.jdbc.exceptions.MySQLSyntaxErrorException: Specified key was toolong; max key length is nnn bytes
改資料庫編碼

alter database hive character set latin1;

http://www.cnblogs.com/linjiqin/archive/2013/03/04/2943025.html