expdp&impdp备份恢复常用命令

• 备份前准备

创建备份用户

create user backup identified by backup#2018 ;

授予导入导出角色

grant connect,resource to backup;

grant exp_full_database,imp_full_database to backup;

创建逻辑目录

create or replace directory oradmp as '/oradmp';

授予备份用户逻辑目录读写权限

grant read,write on directory oradmp to backup;

• 备份数据库--EXPDP

备份全库

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_full.dmp logfile=`date +%Y%m%d`_full.log full=y

备份单个用户下的数据
expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log schemas=scott

备份多个用户下的数据

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log schemas=scott_1,scott_2

备份单个表

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log tables=scott.dept

备份多个表

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log tables=scott.dept,scott.emp

备份A开头的表

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log tables=scott.A%

备份某表前10行数据

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log tables=scott.dept query=scott.dept:"where rownum<5"

备份时排除单表

expdp backup/backup#2018 directory=oradmp schemas=scott dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log exclude=table:\"in\(\'DEPT\'\)\"

备份时排除多表

expdp backup/backup#2018 directory=oradmp schemas=scott dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log exclude=table:\"in\(\'DEPT\',\'EMP\'\)\"

备份时排除包含20的表

expdp backup/backup#2018 directory=oradmp schemas=scott dumpfile=`date +%Y%m%d`_scott.dmp logfile=`date +%Y%m%d`_scott.log exclude=table:\"like \'\%20\%\'\"

备份用户的数据表结构

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp schemas=scott content=metadata_only

备份用户的数据，不导出表结构

expdp backup/backup#2018 directory=oradmp dumpfile=`date +%Y%m%d`_scott.dmp schemas=scott content=data_only

多进程并行备份（加快备份速度） 注意并行数量一般要少于CPU的数量，并行数量决定了备份文件的分片数量，比如并行数量为8 就会产生8个分片集文件

expdp backup/backup#2018 compress=all full=y directory=oradmp dumpfile=`date +%Y%m%d`_full_%U.dmp logfile=`date +%Y%m%d`_full_%U.log parallel=8

备份排除统计信息

expdp backup/backup#2018 compress=all full=y dumpfile=`date +%Y%m%d`_full_%U.dmp logfile=`date +%Y%m%d`_full.log exclude=table_statistics

集群环境下限制备份文件分片集只在一个节点下

expdp backup/backup#2018 compress=all full=y directory=oradmp dumpfile=`date +%Y%m%d`_full.dmp logfile=`date +%Y%m%d`_full.log cluster=n

备份时只导出DB_LINK

expdp backup/backup#2018 compress=all full=y directory=oradmp dumpfile=`date +%Y%m%d`_full.dmp logfile=`date +%Y%m%d`_full.log include=db_link;

• 恢复数据库

全库恢复

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_full.dmp logfile=`date +%Y%m%d`_impdp.log full=y

恢复单个用户数据

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_scott.dmp logfile=`date +%Y%m%d`_impdp.log schemas=scott

恢复多个用户数据

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_scott.dmp logfile=`date +%Y%m%d`_impdp.log schemas=scott_1, scott_2

恢复时排除用户

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_scott.dmp logfile=`date +%Y%m%d`_impdp.log full=y exclude=schema:\"in\(\'SCOTT\'\)\"

恢复单个表

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_scott.dmp logfile=`date +%Y%m%d`_impdp.log tables=scott.dept

恢复多个表

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_scott.dmp logfile=`date +%Y%m%d`_impdp.log tables=scott.dept,scott.emp

恢复时排除表

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_scott.dmp logfile=`date +%Y%m%d`_impdp.log full=y exclude=table:\"in\(\'DEPT\'\)\"

恢复时排除统计信息

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_scott.dmp logfile=`date +%Y%m%d`_impdp.log full=y exclude=table_statistics

恢复时重命名表空间

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_full.dmp logfile=`date +%Y%m%d`_impdp.log full=y remap_tablespace=EXAMPLE:EXAMPLE_B

恢复时重命名用户名

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_full.dmp logfile=`date +%Y%m%d`_impdp.log schemas=hr remap_schema=hr:other_hr

恢复时重命名表名 （自动创建表other_hr）

impdp backup/backup#2018 directory=oradmp dumpfile=20180517_full.dmp logfile=`date +%Y%m%d`_impdp.log tables=scott.emp remap_table=scott.emp:scott.emp1

• 监控备份恢复进度

查看整体进度

select sid,serial#,start_time,opname,target,sofar,totalwork,round(sofar/totalwork,2) process from v$session_longops where sofar<>totalwork

查看对象创建进度

select count(*) from dba_objects where owner like 'scott';

创建数据导入进度
select sum(bytes)/1024/1024/1024,owner,tablespace_name from dba_segments where owner like 'scott' group by owner,tablespace_name

查看磁盘IO速率

iostat -dmx 1

查看表空间使用情况

select c.host_name HOSTNAME,
b.tablespace_name TABLESPACE_NAME,
a.total TOTAL_SIZE,
a.total - b.free USED_SIZE,
b.free FREE_SIZE,
to_char(round(100 - (b.free/a.total) * 100),'fm9999999990.00') USED_PERCENT
from (select tablespace_name, round(sum(bytes/1024/1024)) total
from dba_data_files
where tablespace_name not in ('UNDOTBS1', 'UNDOTBS2')
group by tablespace_name) a,
(select tablespace_name, round(sum(bytes/1024/1024)) free
from dba_free_space
where tablespace_name like 'USERS'
group by tablespace_name) b,
(select host_name from v$instance) c
where a.tablespace_name = b.tablespace_name