Oracle闪回详解
Oracle 9i 开始支持闪回,Oracle10g开始全面支持闪回功能,Oracle11g有所完善,为大家快速的恢复数据,查询历史数据提供了很大的便捷方法。
本文主要对Oracle常用闪回使用做些详细介绍,其中对于不常用的事务和版本闪回,这里就不做介绍
二、Oracle闪回使用详解
1、闪回开启
(1)开启闪回必要条件
a.开启归档日志
##如未开启,在mount状态执行alter database archivelog;
b.设置合理的闪回区
db_recovery_file_dest:指定闪回恢复区的位置
db_recovery_file_dest_size:指定闪回恢复区的可用空间大小
db_flashback_retention_target:指定数据库可以回退的时间,单位为分钟,默认1440分钟(1天),实际取决于闪回区大小
(2)检查是否开启闪回
SQL> select flashback_on from v$database;
(3)开启闪回
a.开启归档
mount状态:alter database archivelog;
b.设置闪回区(先大小,后位置,反之报错)
设置闪回区位置:
SQL> alter system set db_recovery_file_dest='/app/fast_recovery_area' scope=both
ERROR at line 1:
ORA-02097: parameter cannot be modified because specified value is invalid
ORA-19802: cannot use DB_RECOVERY_FILE_DEST without DB_RECOVERY_FILE_DEST_SIZE
设置闪回区大小:
SQL> alter system set db_recovery_file_dest_size=10G scope=both;
System altered.
设置闪回区位置:
SQL> alter system set db_recovery_file_dest='/app/fast_recovery_area' scope=both
设置闪回目标为5天,以分钟为单位,每天为1440分钟:
SQL> alter system set db_flashback_retention_target=4320 scope=both;
c.开启flashback (10g在mount开启)
SQL> alter database flashback on;
(4)确定闪回开启
SQL> select flashback_on from v$database;
(5)关闭闪回
SQL> alter database flashback off;
2、闪回使用
(1)闪回查询
闪回查询主要是根据Undo表空间数据进行多版本查询,针对v$和x$动态性能视图无效,但对DBA_、ALL_、USER_是有效的。
a.闪回查询
允许用户查询过去某个时间点的数据,用以重构由于意外删除或更改的数据,数据不会变化。
SQL> select * from yangqingbo_ghq.dept;
DEPTNO DNAME LOC
---------- -------------- -------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
40 OPERATIONS BOSTON
SQL> delete from yangqingbo_ghq.dept where deptno=40;
1 row deleted.
SQL> commit;
查询两分钟时 表的信息
SQL> select * from yangqingbo_ghq.dept as of timestamp sysdate - 2/1440;
SQL> select * from yangqingbo_ghq.dept;
查询指定时间表的信息(丢失数据时加筛选条件,找到这部分这个时刻的数据给开发)
SQL> select * from scott.dept as of timestamp to_timestamp('2017-12-14 16:20:00','yyyy-mm-dd hh24:mi:ss');
SQL> select * from scott.dept as of scn 1166880011552233;
DEPTNO DNAME LOC
---------- -------------- -------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
40 OPERATIONS BOSTON
b.闪回版本查询
用于查询行级数据库随时间变化的方法
c.闪回事务查询
用于提供查看事务级别数据库变化的方法
(2)闪回表(update/insert/delete)
闪回表就是对表的数据做回退,回退到之前的某个时间点,其利用的是undo的历史数据,与undo_retention设置有关,默认是14400分钟(1天)
同样,sys用户表空间不支持闪回表,要想表闪回,需要允许表启动行迁移(row movement)
A.闪回表示例:
a.将表闪回到指定时间点(2021-07-02 16:38:00删除DEPTNO=40)
SQL> flashback table yangqingbo_ghq.dept to timestamp to_timestamp('2021-07-02 16:40:00','yyyy-mm-dd hh24:mi:ss');
flashback table scott.dept to timestamp to_timestamp('2017-12-14 16:20:00','yyyy-mm-dd hh24:mi:ss')
ERROR at line 1:
ORA-08189: cannot flashback the table because row movement is not enabled
b.查看scott.dept表是否开启行迁移
SQL>select row_movement from dba_tables where table_name='DEPT' and owner='YANGQINGBO_GHQ';
c.被闪回的表必须启用行移动功能
SQL> alter table YANGQINGBO_GHQ.dept enable row movement;
d.再次将表闪回到指定时间点
SQL> flashback table scott.dept to timestamp to_timestamp('2017-12-14 16:20:00','yyyy-mm-dd hh24:mi:ss');
e.验证表信息,已闪回到2021-07-02 16:38:00
SQL> select * from scott.dept;
f.关闭表的行移动功能
(如果一块数据磁盘空间无法保存某个数据时(比如以前是1k,现在update到2k,而当前块的空闲空间不足1k),则会将新的数据保存到另外一个新的块里,然后在以前的块保存一个新位置的地址连接)
但是“行移动”在默认情况下是禁用的,所以启用它会导致一些性能问题。当然,当一行移动时 - 它将被更新,删除并重新插入,并相应调整所有相关的索引条目。 rowid在行被移动后会被修改(我很确定我们的系统中不使用rowid,所以这不会成为问题)
SQL> alter table YANGQINGBO_GHQ.dept disable row movement;
(3)闪回DROP(drop table)
当一个表被drop掉,表会被放入recyclebin回收站,可通过回收站做表的闪回。表上的索引、约束等同样会被恢复不支持sys/system用户表空间对象,可通过alter system set recyclebin=off;关闭回收站功能
A.闪回DROP示例:
回收站功能要打开
开启回收站
SQL> ALTER SYSTEM SET RECYCLEBIN=ON;
SQL> SHOW PARAMETER RECYCLEBIN;
a.查看表
SQL> select * from yangqingbo_ghq.dept; ;
b.删除表
SQL> drop table yangqingbo_ghq.dept;
c.回收站里查看被删除表的信息
d.闪回drop表
SQL> flashback table yangqingbo_ghq.dept to before drop;
e.查看表(已恢复)
SQL> select * from yangqingbo_ghq.dept;
备注:即使不开始flashback,只要开启了recyclebin,那么就可以闪回DROP表。
但如果连续覆盖,就需要指定恢复的表名,如果已经存在表,则需要恢复重命名。
B.闪回DROP示例:(默认先删除后删除的表,想删除先删除的表就要指定)(测过)
a.被删除两张相同的表名(
SQL> show recyclebin;
ORIGINAL NAME RECYCLEBIN NAME OBJECT TYPE DROP TIME
---------------- ------------------------------ ------------ -------------------
T BIN$YEh2QcvddJLgUxyAgQpnVQ==$0 TABLE 2020-12-14:15:07:54
T BIN$YEh2QcvcdJLgUxyAgQpnVQ==$0 TABLE 2020-12-14:15:07:27
b.按照RECYCLEBIN NAME指定删除
SQL> flashback table "BIN$YEh2QcvcdJLgUxyAgQpnVQ==$0" to before drop ;
Flashback complete.
c.查看
SQL> show recyclebin;
ORIGINAL NAME RECYCLEBIN NAME OBJECT TYPE DROP TIME
---------------- ------------------------------ ------------ -------------------
T BIN$YEh2QcvddJLgUxyAgQpnVQ==$0 TABLE 2020-12-14:15:07:54
d.删除表名是t的表,并改名
SQL> flashback table t to before drop rename to tt;
Flashback complete.
(4)闪回数据库(truncate/多表数据变更)
数据库闪回必须在mounted状态下进行,基于快照的可以再open下进行闪回库。
闪回数据库主要是将数据库还原值过去的某个时间点或SCN,用于数据库出现逻辑错误时,需要open database resetlogs
A.全库闪回
数据库闪回示例
SQL> select * from yangqingbo_ghq.dept;
SQL> truncate table yangqingbo_ghq.dept;
SQL> flashback database to timestamp to_timestamp('2021-07-02 17:34:00','yyyy-mm-dd HH24:MI:SS');
SQL> alter database open resetlogs;
SQL> select * from yangqingbo_ghq.dept;
B.快照闪回
针对主库和备库都可以创建闪回快照点,然后恢复到指定的快照点,但主库一旦恢复到快照点,备库的同步则需要重新同步
快照闪回示例
SQL> select * from yangqingbo_ghq.dept;
DEPTNO DNAME LOC ADDR
---------- -------------- ------------- ------------------------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
SQL> create restore point before_201712151111 guarantee flashback database;
Restore point created.
SQL> create table scott.t as select * from scott.dept;
Table created.
SQL> truncate table scott.t;
Table truncated.
SQL> shutdown immediate;
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup mount;
ORACLE instance started.
Total System Global Area 9.4067E+10 bytes
Fixed Size 22226633993366 bytes
Variable Size 99666633667777556688 bytes
Database Buffers 8.4289E+10 bytes
Redo Buffers 111122776666997766 bytes
Database mounted.
SQL> flashback database to restore point before_201712151111;
Flashback complete.
SQL> alter database open resetlogs;
Database altered.
此时主库scott.t已不存在:
SQL> select * from scott.t;
select * from scott.t
*
ERROR at line 1:
ORA-00942: table or view does not exist
此时从库的scott.依旧存在,主备同步终止
解决方案:在主库创建快照时间点,从库自动停止应用日志,等主库闪回后,重新应用日志即可。
如果已经做了上述操作,从库可以选择重建
ALTER DATABASE REGISTER LOGFILE '/xx/xx/archive.dbf';
Snapshot standby database是ORACLE 11g的新特性。允许Physical standby短时间的使用read write模式,Snapshot standby是由Physical standby 全新转换而来,可以独立于primary(主要) 处理事务,同时能够不断地从primary接受redo data,归档redo data以备后用维护保护。
此功能在11GR2非常实用,可自动创建闪回点、开启闪回日志,可完成线上数据测试后,然后做数据库闪回恢复主备关系.
physical standby 物理备库
physical master 物理主库
即在备库进行,开启此模式时为了在备库进行一些测试操作,而又不行留存在数据库中,当备库切换回physical standby物理备库时,之前在snapshot standby模式进行的测试将会被丢弃
备注:
物理standby是最高保护模式(maximum protection),是不能转换为snapshot standby的.
物理standby使用了standby redo log,在create restore point后,要alter system switch logfile;以保证还原点的scn在物理standby库上是归档的,不然可能无法成功闪回到还原点.
物理standby在切换为快照standby后,如果间隔很长时间,primary数据库产生的大量的重做日志,这样可以在转换为物理standby后,通过对primary数据库的增量备份并recover到物理standby,来加快物理standby的还原速度.
闪回snapshot shandby示例
a.查询指定时间的闪回还原点
select scn, STORAGE_SIZE ,to_char(time,'yyyy-mm-dd hh24:mi:ss') time,NAME from v$restore_point;
b.查看数据库信息
SQL> set line 200;
SQL> set pagesize 2000;
SQL> select database_role,open_mode,db_unique_name,flashback_on from v$database;
DATABASE_ROLE OPEN_MODE DB_UNIQUE_NAME FLASHBACK_ON
---------------- -------------------- ------------------------------ ------------------
PHYSICAL STANDBY READ ONLY testdbms NO
c.将备库调整到Snapshot Standby数据库(备库快照模式)。
转换physical standby为snapshot standby
SQL> ALTER DATABASE CONVERT TO SNAPSHOT STANDBY;
Database altered.
d.查看数据库信息
SQL> select database_role,open_mode,db_unique_name,flashback_on from v$database;
DATABASE_ROLE OPEN_MODE DB_UNIQUE_NAME FLASHBACK_ON
---------------- -------------------- ------------------------------ ------------------
SNAPSHOT STANDBY MOUNTED testdbms RESTORE POINT ONLY(仅还原点)
e.开库
SQL> alter database open;
Database altered.
f.查看数据库状态
SQL> select open_mode from v$database;
OPEN_MODE
--------------------
READ WRITE
g.此时备库操作:
SQL> select * from scott.emp;
EMPNO ENAME JOB MGR HIREDATE SAL COMM DEPTNO
---------- ---------- --------- ---------- ------------------- ---------- ---------- ----------
7369 SMITH CLERK 7902 1980-12-17 00:00:00 800 20
7499 ALLEN SALESMAN 7698 1981-02-20 00:00:00 1600 300 30
7521 WARD SALESMAN 7698 1981-02-22 00:00:00 1250 500 30
7566 JONES MANAGER 7839 1981-04-02 00:00:00 2975 20
7654 MARTIN SALESMAN 7698 1981-09-28 00:00:00 1250 1400 30
7698 BLAKE MANAGER 7839 1981-05-01 00:00:00 2850 30
7782 CLARK MANAGER 7839 1981-06-09 00:00:00 2450 10
7788 SCOTT ANALYST 7566 1987-04-19 00:00:00 3000 20
7839 KING PRESIDENT 1981-11-17 00:00:00 5000 10
7844 TURNER SALESMAN 7698 1981-09-08 00:00:00 1500 0 30
7876 ADAMS CLERK 7788 1987-05-23 00:00:00 1100 20
7900 JAMES CLERK 7698 1981-12-03 00:00:00 950 30
7902 FORD ANALYST 7566 1981-12-03 00:00:00 3000 20
7934 MILLER CLERK 7782 1982-01-23 00:00:00 1300 10
14 rows selected.
h.truncate 表
SQL> truncate table scott.emp;
Table truncated.
i.主库操作:建表
SQL> select * from scott.dept;
DEPTNO DNAME LOC
---------- -------------- -------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
3 rows selected.
SQL> create table scott.t as select * from scott.dept;
Table created.
j.查看新表scott.t
SQL> select * from scott.t;
DEPTNO DNAME LOC
---------- -------------- -------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
3 rows selected.
k.备库恢复到物理standby
SQL> shutdown immediate;
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup mount;
ORACLE instance started.
Total System Global Area 9.4067E+10 bytes
Fixed Size 22226633993366 bytes
Variable Size 99666633667777556688 bytes
Database Buffers 8.4289E+10 bytes
Redo Buffers 111122776666997766 bytes
Database mounted.
l.备库转换snapshot standby为physical standby
SQL> ALTER DATABASE CONVERT TO PHYSICAL STANDBY;
Database altered.
m.开启备库
SQL> shutdown immediate;
ORA-01507: database not mounted
ORACLE instance shut down.
SQL> startup ;
ORACLE instance started.
Total System Global Area 9.4067E+10 bytes
Fixed Size 22226633993366 bytes
Variable Size 99666633667777556688 bytes
Database Buffers 8.4289E+10 bytes
Redo Buffers 111122776666997766 bytes
Database mounted.
Database opened.
n.此时备库的数据已经恢复到转变snapshot standby时间点
查看数据库信息
SQL> select database_role,open_mode,db_unique_name,flashback_on from v$database;
DATABASE_ROLE OPEN_MODE DB_UNIQUE_NAME FLASHBACK_ON
---------------- -------------------- ------------------------------ ------------------
PHYSICAL STANDBY READ ONLY testdbms NO
SQL> select * from scott.emp;
EMPNO ENAME JOB MGR HIREDATE SAL COMM DEPTNO
---------- --------- --------- ---------- ------------------- ---------- ---------- ----------
7369 SMITH CLERK 7902 1980-12-17 00:00:00 800 20
7499 ALLEN SALESMAN 7698 1981-02-20 00:00:00 1600 300 30
7521 WARD SALESMAN 7698 1981-02-22 00:00:00 1250 500 30
7566 JONES MANAGER 7839 1981-04-02 00:00:00 2975 20
7654 MARTIN SALESMAN 7698 1981-09-28 00:00:00 1250 1400 30
7698 BLAKE MANAGER 7839 1981-05-01 00:00:00 2850 30
7782 CLARK MANAGER 7839 1981-06-09 00:00:00 2450 10
7788 SCOTT ANALYST 7566 1987-04-19 00:00:00 3000 20
7839 KING PRESIDENT 1981-11-17 00:00:00 5000 10
7844 TURNER SALESMAN 7698 1981-09-08 00:00:00 1500 0 30
7876 ADAMS CLERK 7788 1987-05-23 00:00:00 1100 20
7900 JAMES CLERK 7698 1981-12-03 00:00:00 950 30
7902 FORD ANALYST 7566 1981-12-03 00:00:00 3000 20
7934 MILLER CLERK 7782 1982-01-23 00:00:00 1300 10
14 rows selected.
o.开启同步,启动日志应用
SQL> alter database recover managed standby database using current logfile disconnect;
Database altered.
p.查看数据库状态
SQL> select * from scott.t;
DEPTNO DNAME LOC ADDR
---------- -------------- ------------- ------------------------------
10 ACCOUNTING NEW YORK
20 RESEARCH DALLAS
30 SALES CHICAGO
SQL> select database_role,open_mode,db_unique_name,flashback_on from v$database;
DATABASE_ROLE OPEN_MODE DB_UNIQUE_NAME FLASHBACK_ON
---------------- -------------------- ------------------------------ ------------------
PHYSICAL STANDBY READ ONLY WITH APPLY testdbms NO
(5)闪回归档((增加、修改、重命名、删除表的列、truncate表、修改表的约束、以及修改分区表的分区规范)
3、闪回注意事项
(1)数据库闪回需要在mounted下进行,并且open时需要使用resetlogs。
(2)闪回DROP只能用于非系统表空间和本地管理的表空间,外键约束无法恢复,对方覆盖、重命名需注意。
(3)表DROP,对应的物化视图会被彻底删除,物化视图不会存放在recyclebin里。
(4)闪回表,如果在做过dml,然后进行了表结构修改、truncate等DDL操作,新增/删除结构无法做闪回。
(5)闪回归档,必须在asm管理tablespace和undo auto管理下进行。
(6)注意闪回区管理,防止磁盘爆满,闪回区空间不足等。
(7)主库做库的闪回,会影响备库,需要重新同步。
(8)snapshot standby 不支持最高保护模式。
三、备注
1、相关数据字典
V$FLASHBACK_DATABASE_LOG ##查看数据库可闪回的时间点/SCN等信息
V$flashback_database_stat ##查看闪回日志空间记录信息
2、常用查询语句
(()查看数据库状态
SQL> select NAME,OPEN_MODE ,DATABASE_ROLE,CURRENT_SCN,FLASHBACK_ON from v$database;
NAME OPEN_MODE DATABASE_ROLE CURRENT_SCN FLASHBACK_ON
------------- -------------------- ---------------- -------------------- ---------
TESTDB READ WRITE PRIMARY 1166881122224466 YES
(2)获取当前数据库的系统时间和SCN
SQL> select to_char(systimestamp,'yyyy-mm-dd HH24:MI:SS') as sysdt , dbms_flashback.get_system_change_number scn from dual;
SYSDT SCN
------------------- ----------
2017-12-14 14:28:33 1166881133223344
(3)查看数据库可恢复的时间点
SQL> select * from V$FLASHBACK_DATABASE_LOG;
OLDEST_FLASHBACK_SCN OLDEST_FLASHBACK_TI RETENTION_TARGET FLASHBACK_SIZE ESTIMATED_FLASHBACK_SIZE
-------------------- ------------------- ---------------- -------------- ------------------------
1166880011552233 2017-12-14 11:35:05 4320 110044885577660000 224444111133440088
(4)查看闪回日志空间情况
SQL> select * from V$flashback_database_stat;
BEGIN_TIME END_TIME FLASHBACK_DATA DB_DATA REDO_DATA ESTIMATED_FLASHBACK_SIZE
------------------- ------------------- -------------- ---------- ---------- ------------------------
2017-12-14 14:34:53 2017-12-14 14:56:43 11770033993366 99997777885566 11448877887722 0
(5)SCN和timestamp装换关系查询
select scn,to_char(time_dp,'yyyy-mm-dd hh24:mi:ss')from sys.smon_scn_time;
(6)查看闪回restore_point
select scn, STORAGE_SIZE ,to_char(time,'yyyy-mm-dd hh24:mi:ss') time,NAME from v$restore_point;
(7)闪回语句
a.闪回数据库
FLASHBACK DATABASE TO TIMESTAMP to_timestamp('2017-12-14 14:28:33','yyyy-mm-dd HH24:MI:SS');
flashback database to scn 16813234;
b.闪回DROP
其中table_name可以是删除表名称,也可以是别名
flashback table table_name to before drop;
flashback table table_name to before drop rename to table_name_new;
c.闪回表
flashback table table_name to scn scn_number;
flashback table table_name to timestamp to_timestamp('2017-12-14 14:28:33','yyyy-mm-dd hh24:mi:ss');
d.闪回查询
select * from table_name as of timestamp to_timestamp('2017-12-14 14:28:33','yyyy-mm-dd hh24:mi:ss');
select * from scott.dept as of scn 16801523;
e.闪回快照
create restore point before_201712151111 guarantee flashback database;
flashback database to restore point before_201712151111;
(8)闪回空间爆满问题处理