一.背景说明
Oracle RMAN 备份的恢复分2个步骤:RESTRE 和 RECOVER。
在这个过程中,Recover 是依赖与归档文件的。
假设一种情况:周一对数据库做了全备,然后保留归档。周四发现数据库有异常,准备恢复,发现周二的时候少了一个归档。
按照正常的情况,我们只能将数据库恢复到周二缺失归档的之前的点。
那么我这里就是一个研究,如何跳过这个缺失的归档,让数据库继续进行Recover。
根据测试结果,Recover 是可以继续,但是测试的结果意义不是很大,因为还是有数据丢失。
所以这里更多的是对这种方法的抛砖引玉。
二.测试案例
2.1 使用RMAN 全备数据库
此步骤直接备份即可。
2.2 创建测试表dave1并切换归档
SQL> select sequence# from v$log wherethread#=1;
SEQUENCE#
----------
152
151
SQL> create table dave1 as select * fromdba_users;
Table created.
SQL> alter system switch logfile;
System altered.
SQL> select sequence# from v$log wherethread#=1;
SEQUENCE#
----------
152
153
2.3 创建测试表dave2并切换归档
SQL> create table dave2 as select * fromdba_users;
Table created.
SQL> alter system switch logfile;
System altered.
SQL> select sequence#,status from v$logwhere thread#=1;
SEQUENCE# STATUS
---------- ----------------
154 CURRENT
153 ACTIVE
SQL> select sequence# fromv$archived_log where thread#=1;
SEQUENCE#
----------
148
149
150
151
152
153
6 rows selected.
2.4 删除153的归档
[oracle@dave arch]$ ll
total 42200
-rw-r-----. 1 oracle oinstall 42715136Jul 5 22:56 1_125_816661296.dbf
-rw-r-----. 1 oracle oinstall 248320 Jul 6 23:14 1_152_816661296.dbf
-rw-r-----. 1 oracle oinstall 127488 Jul 6 23:15 1_153_816661296.dbf
-rw-r-----. 1 oracle oinstall 113664 Jul 6 23:19 1_154_816661296.dbf
[oracle@dave arch]$ rm-rf 1_153_816661296.dbf
[oracle@dave arch]$ ll
total 42072
-rw-r-----. 1 oracle oinstall 42715136Jul 5 22:56 1_125_816661296.dbf
-rw-r-----. 1 oracle oinstall 248320 Jul 6 23:14 1_152_816661296.dbf
-rw-r-----. 1 oracle oinstall 113664 Jul 6 23:19 1_154_816661296.dbf
[oracle@dave arch]$
2.5 然后进行restore 和recover
SQL> shutdown immediate
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL>
SQL> startup mount
RMAN> restore database;
Starting restore at 06-JUL-13
released channel: ORA_DISK_1
allocated channel: ORA_DISK_1
channel ORA_DISK_1: SID=20 device type=DISK
channel ORA_DISK_1: starting datafilebackup set restore
channel ORA_DISK_1: specifying datafile(s)to restore from backup set
channel ORA_DISK_1: restoring datafile00001 to /u01/app/oracle/oradata/dave/system.256.816661027
channel ORA_DISK_1: restoring datafile00003 to /u01/app/oracle/oradata/dave/undotbs1.258.816661037
channel ORA_DISK_1: restoring datafile00005 to /u01/app/oracle/oradata/dave/undotbs2.265.816661787
channel ORA_DISK_1: reading from backuppiece /u01/backup/dave_lev0_06oe3kdv_1_1_20130706
channel ORA_DISK_1: piecehandle=/u01/backup/dave_lev0_06oe3kdv_1_1_20130706 tag=DAVE_LEV0
channel ORA_DISK_1: restored backup piece 1
channel ORA_DISK_1: restore complete,elapsed time: 00:00:25
channel ORA_DISK_1: starting datafilebackup set restore
channel ORA_DISK_1: specifying datafile(s)to restore from backup set
channel ORA_DISK_1: restoring datafile00002 to /u01/app/oracle/oradata/dave/sysaux.257.816661033
channel ORA_DISK_1: restoring datafile00004 to /u01/app/oracle/oradata/dave/users.259.816661039
channel ORA_DISK_1: restoring datafile00006 to /u01/app/oracle/oradata/dave/dave01.dbf
channel ORA_DISK_1: restoring datafile00007 to /u01/app/oracle/oradata/dave/dave02.dbf
channel ORA_DISK_1: reading from backuppiece /u01/backup/dave_lev0_05oe3kdv_1_1_20130706
channel ORA_DISK_1: piece handle=/u01/backup/dave_lev0_05oe3kdv_1_1_20130706tag=DAVE_LEV0
channel ORA_DISK_1: restored backup piece 1
channel ORA_DISK_1: restore complete,elapsed time: 00:00:25
Finished restore at 06-JUL-13
RMAN>
RMAN> recoverdatabase;
Starting recover at 06-JUL-13
using channel ORA_DISK_1
starting media recovery
archived log for thread 1 with sequence 152is already on disk as file /u01/arch/1_152_816661296.dbf
archived log for thread 1 with sequence 154is already on disk as file /u01/arch/1_154_816661296.dbf
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGESTACK FOLLOWS ===============
RMAN-00571:===========================================================
RMAN-03002: failure of recover command at07/06/2013 23:23:48
RMAN-06053: unable to perform mediarecovery because of missing log
RMAN-06025: no backup ofarchived log for thread 1 with sequence 153 and starting SCN of 3836001 foundto restore
RMAN>
这个153 是我们刚才手工删掉的归档。如果这个不搞定,后面没办法恢复。
2.6 BBED 推荐SCN
2.6.1 修改原理说明
-- System Checkpoint SCN:
SQL> select checkpoint_change# fromv$database;
CHECKPOINT_CHANGE#
------------------
3836654
--- Datafile CheckpointSCN:
SQL> select name,checkpoint_change# fromv$datafile;
NAME CHECKPOINT_CHANGE#
-------------------------------------------------------------------------
/u01/app/oracle/oradata/dave/system.256.816661027 3836654
/u01/app/oracle/oradata/dave/sysaux.257.816661033 3836654
/u01/app/oracle/oradata/dave/undotbs1.258.816661037 3836654
/u01/app/oracle/oradata/dave/users.259.816661039 3836654
/u01/app/oracle/oradata/dave/undotbs2.265.816661787 3836654
/u01/app/oracle/oradata/dave/dave01.dbf 3836654
/u01/app/oracle/oradata/dave/dave02.dbf 3836654
7 rows selected.
---START SCN:
SQL> select name,checkpoint_change# fromv$datafile_header;
NAME CHECKPOINT_CHANGE#
-------------------------------------------------------------------------
/u01/app/oracle/oradata/dave/system.256.816661027 3835435
/u01/app/oracle/oradata/dave/sysaux.257.816661033 3835434
/u01/app/oracle/oradata/dave/undotbs1.258.816661037 3835435
/u01/app/oracle/oradata/dave/users.259.816661039 3835434
/u01/app/oracle/oradata/dave/undotbs2.265.816661787 3835435
/u01/app/oracle/oradata/dave/dave01.dbf 3835434
/u01/app/oracle/oradata/dave/dave02.dbf 3835434
7 rows selected.
+++++SCN号与数据库启动:
在数据库启动过程中,当SystemCheckpoint SCN、Datafile Checkpoint SCN和Start SCN号都相同时,数据库可以正常启动,不需要做mediarecovery.三者当中有一个不同时,则需要做media recovery。
如果在启动的过程中,EndSCN号为NULL,则需要做instance recovery。ORACLE在启动过程中首先检查是否需要media recovery,然后再检查是否需要instance recovery。
在进行recovery的时候,我们根据归档,推进START SCN,但是归档缺失,导致无法推荐,数据库也无法启动。
我们这里缺失的是153的归档,我们只需要手工的修改datafile header,让数据库认为这个归档已经恢复了,即可。 这是一种欺骗行为,虽然可以继续,但还是会出现问题。
可以使用如下方法确定具体缺失的归档SCN,然后使用BBED 跳过这些SCN 即可。
SQL> selectsequence#,first_change#,next_change# from v$archived_log;
SEQUENCE# FIRST_CHANGE# NEXT_CHANGE#
---------- ------------- ------------
148 3834837 3835155
149 3835155 3835184
150 3835184 3835498
151 3835498 3835507
152 3835507 3836001
153 3836001 3836079
154 3836079 3836303
7 rows selected.
这个正好与我们之前RMAN 错误一致:
RMAN-03002: failure of recover command at07/06/2013 23:23:48
RMAN-06053: unable to perform mediarecovery because of missing log
RMAN-06025: no backup ofarchived log for thread 1 with sequence 153 and starting SCN of 3836001 foundto restore
2.6.2 使用BBED 推进所有DATAFILE header SCN
这里,我们只修改kscnbas的值:
kscnbas (at offset 484) - SCN of lastchange to the datafile.
BBED> info
File# Name Size(blks)
----- ---- ----------
1 /u01/app/oracle/oradata/dave/system.256.816661027 129280
2 /u01/app/oracle/oradata/dave/sysaux.257.816661033 97280
3 /u01/app/oracle/oradata/dave/undotbs1.258.816661037 9600
4 /u01/app/oracle/oradata/dave/users.259.816661039 640
5 /u01/app/oracle/oradata/dave/undotbs2.265.816661787 12800
6 /u01/app/oracle/oradata/dave/dave01.dbf 393216
7 /u01/app/oracle/oradata/dave/dave02.dbf 6400
+++我们需要将所有datafile 的SCN从3836001 推到3836079:
SQL> selectto_char('3836079','xxxxxxxxx') from dual;
TO_CHAR('3
----------
3a88af
因此我们的kscnbas 的新值是:0x003a88af。
但是注意,对于little-endian的format,他存储是先存储低位的,因此实际block 存储的是:af883a00.
我们需要使用BBED 将所有datafileheader 的@484 的值修改成:af883a00。
BBED> d /v dba 1,1 offset 484
File:/u01/app/oracle/oradata/dave/system.256.816661027 (1)
Block: 1 Offsets: 484 to 499 Dba:0x00400001
-------------------------------------------------------
2b863a00 00000000 bfd1e130 01000000 l+.:........0....
<16 bytes per line>
BBED> modify /x af88 dba 1,1 offset 484
File: /u01/app/oracle/oradata/dave/system.256.816661027(1)
Block: 1 Offsets: 484 to 499 Dba:0x00400001
------------------------------------------------------------------------
af883a00 00000000 bfd1e130 01000000
<32 bytes per line>
BBED> sum dba 1,1 apply
Check value for File 1, Block 1:
current = 0xe9ba, required = 0xe9ba
+++按照同样的步骤,把剩下的6个datafile都修改。
--BBED 推荐成功:
SQL> selectfile#,checkpoint_change#,status from v$datafile_header;
FILE# CHECKPOINT_CHANGE# STATUS
---------- ------------------ -------
1 3836079 ONLINE
2 3836079 ONLINE
3 3836079 ONLINE
4 3836079 ONLINE
5 3836079 ONLINE
6 3836079 ONLINE
7 3836079 ONLINE
7 rows selected.
这里的datafile 的SCN 都跳过了我们缺失的归档,我们可以继续进行recover了。
2.7 重新进行Recover
RMAN> recover database;
Starting recover at 07-JUL-13
using channel ORA_DISK_1
starting media recovery
media recovery failed
RMAN-00571:===========================================================
RMAN-00569: =============== ERROR MESSAGESTACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of recover command at07/07/2013 01:04:43
ORA-00283: recovery session canceled due toerrors
RMAN-11003: failure during parse/executionof SQL statement: alter database recover if needed
start
ORA-00283: recovery session canceled due toerrors
ORA-00600: internal errorcode, arguments: [3020], [3], [8077], [12590989], [], [], [], [], [], [], [],[]
ORA-10567:Redo is inconsistent with data block (file# 3, block# 8077, file offset is 66166784 bytes)
ORA-10564: tablespaceUNDOTBS1
ORA-01110: data file 3:'/u01/app/oracle/oradata/dave/undotbs1.258.816661037'
ORA-10560: block type 'KTU UNDO BLOCK'
根据官网的说明,我们这是UNDO 表空间恢复无法继续了,详见:
Resolving ORA-600[3020] Raised During Recovery (文档 ID 361172.1)
尝试跳过坏块测试:
RMAN> recover database allow 50 corruption;
Starting recover at 07-JUL-13
using channel ORA_DISK_1
starting media recovery
media recovery complete, elapsed time:00:00:01
Finished recover at 07-JUL-13
RMAN>
恢复是没有问题,但是打开是有问题的:
SQL> alter database open;
alter database open
*
ERROR at line 1:
ORA-01092: ORACLE instance terminated.Disconnection forced
ORA-01578: ORACLE data block corrupted(file # 3, block # 128)
ORA-01110: data file 3:'/u01/app/oracle/oradata/dave/undotbs1.258.816661037'
Process ID: 32549
Session ID: 16 Serial number: 5
2.8 重建UNDO 表空间
这里里面的 3 就是我们的undo 表空间,我们把重新创建一个UNDO 在拉起数据库:
2.8.1 用spfile 创建pfile,然后修改参数
#*.undo_tablespace='UNDOTBS1'
*.undo_management='MANUAL'
*.rollback_segments='SYSTEM'
2.8.2 用修改之后的pfile,重启DB
SQL> startuppfile='/u01/app/oracle/product/11.2.0/dbhome_1/dbs/initdave.ora'
ORACLE instance started.
Total System Global Area 718188544 bytes
Fixed Size 2231832 bytes
Variable Size 436208104 bytes
Database Buffers 276824064 bytes
Redo Buffers 2924544 bytes
Database mounted.
Database opened.
SQL>
2.8.3 删除原来的表空间,创建新的UNDO 表空间
SQL> select tablespace_name fromdba_tablespaces;
TABLESPACE_NAME
------------------------------
SYSTEM
SYSAUX
UNDOTBS1
TEMP
USERS
UNDOTBS2
DAVE
7 rows selected.
SQL> droptablespace undotbs1;
SQL> create undo tablespace undotbs1datafile '/u01/app/oracle/oradata/dave/undotbs1.dbf' size 50M;
Tablespace created.
2.8.4 关闭数据库,修改pfile参数,然后用新的pfile创建spfile,在正常启动数据库。
*.undo_tablespace='UNDOTBS1'
#*.undo_management='MANUAL'
#*.rollback_segments='SYSTEM'
SQL> shutdown immediate
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup
ORACLE instance started.
Total System Global Area 718188544 bytes
Fixed Size 2231832 bytes
Variable Size 436208104 bytes
Database Buffers 276824064 bytes
Redo Buffers 2924544 bytes
Database mounted.
Database opened.
SQL>
库终于拉起来了。
2.9 验证
SQL> select name,checkpoint_change# fromv$datafile;
NAME CHECKPOINT_CHANGE#
-------------------------------------------------------------------------
/u01/app/oracle/oradata/dave/system.256.816661027 3857521
/u01/app/oracle/oradata/dave/sysaux.257.816661033 3857521
/u01/app/oracle/oradata/dave/undotbs1.dbf 3857521
/u01/app/oracle/oradata/dave/users.259.816661039 3857521
/u01/app/oracle/oradata/dave/undotbs2.265.816661787 3857521
/u01/app/oracle/oradata/dave/dave01.dbf 3857521
/u01/app/oracle/oradata/dave/dave02.dbf 3857521
7 rows selected.
SQL> select name,checkpoint_change# fromv$datafile_header;
NAME CHECKPOINT_CHANGE#
-------------------------------------------------------------------------
/u01/app/oracle/oradata/dave/system.256.816661027 3857521
/u01/app/oracle/oradata/dave/sysaux.257.816661033 3857521
/u01/app/oracle/oradata/dave/undotbs1.dbf 3857521
/u01/app/oracle/oradata/dave/users.259.816661039 3857521
/u01/app/oracle/oradata/dave/undotbs2.265.816661787 3857521
/u01/app/oracle/oradata/dave/dave01.dbf 3857521
/u01/app/oracle/oradata/dave/dave02.dbf 3857521
7 rows selected.
SQL> select checkpoint_change# fromv$database;
CHECKPOINT_CHANGE#
------------------
3857521
SQL> select count(1) from dave1;
select count(1) from dave1
*
ERROR at line 1:
ORA-00942: table or view does not exist
SQL> select count(1) from dave2;
select count(1) from dave2
*
ERROR at line 1:
ORA-00942: table or view does not exist
库是正常拉起来了,不过之前创建的表都没有成功恢复。
--------------------------------------------------------------------------------------------
版权所有,文章允许转载,但必须以链接方式注明源地址,否则追究法律责任!
QQ: 251097186
Skype: tianlesoftware
Email: tianlesoftware@gmail.com
Blog: http://blog.csdn.net/tianlesoftware
Weibo: http://weibo.com/tianlesoftware
Twitter: http://twitter.com/tianlesoftware
Facebook: http://www.facebook.com/tianlesoftware
Linkedin: http://cn.linkedin.com/in/tianlesoftware