Loading

hadoop-16-sqoop导入oracle数据

  sqoop = sql + hadoop

1,系统方面的准备

scp ./ojdbc6.jar root@server-1:/usr/hdp/2.6.0.3-8/sqoop/lib/
scp ./ojdbc6.jar root@server-2:/usr/hdp/2.6.0.3-8/sqoop/lib/

cd /usr/hdp/2.6.0.3-8/sqoop/lib/
chmod 755 ojdbc6.jar

学习了:https://www.cnblogs.com/jxhd1/p/6528574.html
su root
# usermod -a -G hadoop user001 '把user001用户添加到hadoop组中
# usermod -a -G hdfs user001 '把user001用户添加到hdfs组中

2,sqoop导入到hdfs中
学习了:http://blog.csdn.net/yu616568/article/details/45030739

'可行,识别--connection-manager,主要是-m 2进行并行导入
sqoop import --connect jdbc:Oracle:thin:@127.0.0.1:1521:ora --connection-manager org.apache.sqoop.manager.OracleManager --username admin --password admin -m 2 --split-by id --query 'SELECT t.* FROM tbl01 t where $CONDITIONS' --target-dir '/user/user001/tbl01'

'方法2 ,用--driver指定了类,-m 1 进行导入了
sqoop import --connect jdbc:Oracle:thin:@127.0.0.1:1521:ora --username admin --password admin -m 1 --driver oracle.jdbc.driver.OracleDriver --query 'SELECT * FROM tbl02 t where 1=1 and $CONDITIONS' --target-dir '/user/user001/tbl02'

‘方法3 , 查看源码之后org.apache.sqoop.manager.OracleManager.OracleManager(SqoopOptions),

'里面:    } else if (scheme.startsWith("jdbc:oracle:")) {, 注意自己的连接串写的不规范

sqoop import --connect jdbc:oracle:thin:@127.0.0.1:1521:ddxt11 --username admin --password admin   -m 2 --split-by id --query 'SELECT t.* FROM tbl01  t where $CONDITIONS' --target-dir '/user/user001/tbl01'

 ‘方法4, 表名(TBL01)、列名(--split-by ID)都要大写,不大写就报错给你看;

'主要是生成的sql语句:SELECT t.* FROM "TBL01" t WHERE 1=0,SELECT MIN("NAME1"), MAX("NAME1") FROM "TBL01"

sqoop import --connect jdbc:oracle:thin:@127:1521:ddxt11 --username admin  --password admin -m 2 --split-by ID --table TBL01 --target-dir '/user/user001/tbl01'

 

3,数据库表从hdfs中导入到hive中

user001用户登录:

>hive

>show databases;

>create database admin;

>use admin;

>create table tbl01
(
id bigint comment '父id',
pid bigint comment '父id',
name1 VARCHAR(16) comment 'a',
name2 VARCHAR(64) comment 'b',
)
ROW FORMAT DELIMITED FIELDS TERMINATED BY ','
stored as textfile;


在hive中执行:
load data inpath '/user/user001/tbl01/part-m-00000' into table admin.tbl01;

在user001用户中执行:

必须先把表tbl01创建出来;
hive -e "load data inpath '/user/user001/tbl01/part-m-00000' into table admin.tbl01;"
或者
hive -f '/websoftware/tbl01_imp.sql'

不会报错了;需要把user001添加到hdfs组下面;
因为hadoop fs -ls /apps/hive/warehouse 输出:
drwxrwxrwx - user001 hdfs 0 2017-11-09 10:00 /apps/hive/warehouse/admin.db
drwxrwxrwx - user001 hdfs 0 2017-11-02 15:14 /apps/hive/warehouse/hive_test.db
hive存储文件的地方,需要使用hdfs组权限;


user001用户下面:直接从hdfs下面move到hive下面;
方法2,方法1是在hive下面使用load data 方法;
hadoop fs -mv /user/user001/d_zmzd/part-m-00000 /apps/hive/warehouse/admin.db/tbl02

 

posted @ 2017-11-10 11:22  stono  阅读(410)  评论(0编辑  收藏  举报