hdfs 跨集群数据迁移

distcp

#!/bin/bash

DB=$1
#获取hive表定义
ret=$(hive -e "use ${DB};show tables;"|grep -v _es|grep -v _hb|grep -v importinfo)

for tem in $ret;
do
        echo $tem
        if [ "$tem" != "tab_name" ];then
                hive -e "use ${DB};show create table $tem" >> /data1/job/secha.sh
                echo -e ';c' >> /data1/job/secha.sh
        fi
done

#迁移hive的表数据
ret=$(hive -e "use ${DB};show tables;"|grep -v _es|grep -v _hb|grep -v importinfo)

for tem in $ret;
do
        echo $tem
        if [ "$tem" != "tab_name" ];then
    		hadoop distcp hdfs://master:9000/user/hive/warehouse/${DB}.db/$tem hdfs://192.168.0.21:8020/user/hive/warehouse/${DB}.db/$tem
    	fi
done