189 8069 5689
下面我们进行更复杂情况的处理，就是非一致性关闭的情况下日志损坏的情况。这种情况是很有实际意义的。因为在出现错误的时候，Oracle可能不会允许一致性关闭。没有经验的处理人员往往会贸然的强制关闭，我们获得到的场景往往也是非一致的情况。
员工经过长期磨合与沉淀，具备了协作精神，得以通过团队的力量开发出优质的产品。创新互联坚持“专注、创新、易用”的产品理念，因为“专注所以专业、创新互联网站所以易用所以简单”。公司专注于为企业提供成都做网站、网站设计、微信公众号开发、电商网站开发，小程序设计，软件按需求定制网站等一站式互联网企业服务。
4、非一致性关闭非当前日志组处理——Inactive日志组
非一致性关闭情况下，online日志组全员损坏是很复杂的，也是潜在会有数据损失的。我们这里说的非一致性关闭，就是shutdown abort或者强制断电之后。由于Oracle数据库采用写日志在先的策略，关闭点上我们不能保证脏块（Dirty Block）全都被写入到数据文件，各个数据文件和控制文件在SCN时间上保证一致性。
但是当数据库重新启动时，在open阶段，Oracle会进行instance recovery，从最后一个增量检查点check point追起，重演事务并且回滚事务。Instance Recovery过程也被称为前滚回滚过程，或者成为Cache Recovery和Transaction Recovery过程。
如果进行不一致关闭就会需要进行instance recovery，进行instance recovery最需要的内容就是online redo log。如果我们删除online redo log group恰好是进行instance recovery所需要的，那么我们就只能牺牲掉一部分数据，进行不完全恢复。
我们删除的非当前日志组也有很大风险，因为在日志切换成非当前日志之后，日志组有一段时间对应脏块是不能写入到数据文件中的。这也就是日志文件的active状态。本节讨论非当前日志成员被删除的情况，具体划分为Inactive日志组和非Inactive日志组。
我们下面实验选择Oracle 11g的Linux版本进行。
SQL> select * from v$version;
BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
PL/SQL Release 11.2.0.1.0 - Production
CORE 11.2.0.1.0 Production
当前日志情况如下：
SQL> select group#, status, archived, sequence# from v$log;
GROUP# STATUS ARCHIVED SEQUENCE#
---------- ---------------- -------- ----------
1 INACTIVE YES 108
2 CURRENT NO 110
3 INACTIVE YES 109
SQL> drop table t;
Table dropped
SQL> create table t as select * from dba_objects;
Table created
SQL> select group#, status, member from v$logfile;
GROUP# STATUS MEMBER
---------- ------- --------------------------------------------------------------------------------
3 /u01/oradata/WILSON/onlinelog/o1_mf_3_7xt456o0_.log
3 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_3_7xt45bvp_.log
2 /u01/oradata/WILSON/onlinelog/o1_mf_2_7xt44w3k_.log
2 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_2_7xt450rv_.log
1 /u01/oradata/WILSON/onlinelog/o1_mf_1_870jlj6v_.log
1 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_1_870jljvk_.log
6 rows selected
强制停机之后，删除日志组1对象。
[oracle@bspdev ~]$ sqlplus /nolog
SQL*Plus: Release 11.2.0.1.0 Production on Sat Oct 6 22:27:57 2012
Copyright (c) 1982, 2009, Oracle. All rights reserved.
SQL> conn / as sysdba
Connected.
SQL> shutdown abort
ORACLE instance shut down.
[oracle@bspdev ~]$ mv /u01/oradata/WILSON/onlinelog/o1_mf_1_870jlj6v_.log /u01/oradata/WILSON/onlinelog/redogroup1.log
[oracle@bspdev ~]$ rm /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_1_870jljvk_.log
[oracle@bspdev ~]$
重新启动之后，无法找到对应的日志文件，直接报错。
SQL> conn / as sysdba
Connected to an idle instance.
SQL> startup
ORACLE instance started.
Total System Global Area 849530880 bytes
Fixed Size 1339824 bytes
Variable Size 562040400 bytes
Database Buffers 281018368 bytes
Redo Buffers 5132288 bytes
Database mounted.
ORA-03113: end-of-file on communication channel
Process ID: 5059
Session ID: 1 Serial number: 5
ORA-27037: unable to obtain file status
Linux Error: 2: No such file or directory
Additional information: 3
Errors in file /u01/diag/rdbms/wilson/wilson/trace/wilson_ora_5059.trc:
ORA-00313: open failed for members of log group 1 of thread
ORA-00312: online log 1 thread 1: '/u01/oradata/WILSON/onlinelog/o1_mf_1_870jlj6v_.log'
ORA-00312: online log 1 thread 1: '/u01/flash_recovery_area/WILSON/onlinelog/o1_mf_1_870jljvk_.log'
USER (ospid: 5059): terminating the instance due to error 313
Instance terminated by USER, pid = 5059
重新启动到mount状态，进行处理。
SQL> conn / as sysdba
Connected to an idle instance.
SQL> startup mount
ORACLE instance started.
Total System Global Area 849530880 bytes
Fixed Size 1339824 bytes
Variable Size 562040400 bytes
Database Buffers 281018368 bytes
Redo Buffers 5132288 bytes
Database mounted.
SQL> select open_mode from v$database;
OPEN_MODE
--------------------
MOUNTED
SQL> select group#, status, archived, sequence# from v$log;
GROUP# STATUS ARC SEQUENCE#
---------- ---------------- --- ----------
1 INACTIVE YES 108
3 INACTIVE YES 109
2 CURRENT NO 110
SQL> select checkpoint_change#, last_change# from v$datafile;
CHECKPOINT_CHANGE# LAST_CHANGE#
------------------ ------------
1950243 1950243
1950243 1950243
1950243 1950243
1950243 1950243
1950243 1950243
1950243 1950243
1950243 1950243
7 rows selected.
SQL> select checkpoint_change# from v$datafile_header;
CHECKPOINT_CHANGE#
------------------
1950243
1950243
1950243
1950243
1950243
1950243
1950243
7 rows selected.
SQL>
虽然启动失效，但是我们没有看到很严重的一致性问题。我们尝试启动数据库。
SQL> alter database clear logfile group 1;
Database altered.
SQL> alter database open;
Database altered.
SQL> select group#, status, archived, sequence# from v$log;
GROUP# STATUS ARCHIVED SEQUENCE#
---------- ---------------- -------- ----------
1 CURRENT NO 111
2 INACTIVE YES 110
3 INACTIVE YES 109
这种场景是比较好处理的。下面我们讨论如果是Active状态的日志被删除如何？
5、非一致性关闭非当前日志组处理——Active日志组
我们构造出Active日志组对象。
SQL> select group#, status, member from v$logfile;
GROUP# STATUS MEMBER
---------- ------- --------------------------------------------------------------------------------
3 /u01/oradata/WILSON/onlinelog/o1_mf_3_7xt456o0_.log
3 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_3_7xt45bvp_.log
2 /u01/oradata/WILSON/onlinelog/o1_mf_2_7xt44w3k_.log
2 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_2_7xt450rv_.log
1 /u01/oradata/WILSON/onlinelog/o1_mf_1_870kg322_.log
1 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_1_870kg3mr_.log
6 rows selected
SQL> alter system switch logfile;
System altered
SQL> alter system switch logfile;
System altered
SQL> select group#, status, archived, sequence# from v$log;
GROUP# STATUS ARCHIVED SEQUENCE#
---------- ---------------- -------- ----------
1 CURRENT NO 114
2 ACTIVE YES 113
3 INACTIVE YES 112
之后强制关闭数据库，删除对应的日志组2。
SQL> shutdown abort;
ORACLE instance shut down.
SQL>
[oracle@bspdev trace]$ mv /u01/oradata/WILSON/onlinelog/o1_mf_2_7xt44w3k_.log /u01/oradata/WILSON/onlinelog/redogroup2.log
[oracle@bspdev trace]$ rm /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_2_7xt450rv_.log
重新启动数据库。
SQL> conn / as sysdba
Connected to an idle instance.
SQL> startup
ORACLE instance started.
Total System Global Area 849530880 bytes
Fixed Size 1339824 bytes
Variable Size 562040400 bytes
Database Buffers 281018368 bytes
Redo Buffers 5132288 bytes
Database mounted.
ORA-00313: open failed for members of log group 2 of thread 1
ORA-00312: online log 2 thread 1:
'/u01/flash_recovery_area/WILSON/onlinelog/o1_mf_2_7xt450rv_.log'
ORA-27037: unable to obtain file status
Linux Error: 2: No such file or directory
Additional information: 3
ORA-00312: online log 2 thread 1:
'/u01/oradata/WILSON/onlinelog/o1_mf_2_7xt44w3k_.log'
ORA-27037: unable to obtain file status
Linux Error: 2: No such file or directory
Additional information: 3
SQL> select open_mode from v$database;
OPEN_MODE
--------------------
MOUNTED
检查文件状态情况。
SQL> select checkpoint_change#, last_change# from v$datafile;
CHECKPOINT_CHANGE# LAST_CHANGE#
------------------ ------------
1970578
1970578
1970578
1970578
1970578
1970578
1970578
7 rows selected.
SQL> select checkpoint_change# from v$datafile_header;
CHECKPOINT_CHANGE#
------------------
1970578
1970578
1970578
1970578
1970578
1970578
1970578
7 rows selected.
从v$datafile的情况，我们就可以看出关闭是不完全关闭情况。我们只能尝试进行恢复。
SQL> recover database;
ORA-00279: change 1970578 generated at 10/06/2012 22:59:02 needed for thread 1
ORA-00289: suggestion :
/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_113_870kwzkl_.arc
ORA-00280: change 1970578 for thread 1 is in sequence #113
Specify log: {
cancel
Media recovery cancelled.
提出恢复使用sequence#=113的日志，但是这个恰恰是被删除的对象。所以，只能退而求其次选择使用不完全恢复。
SQL> recover database until cancel;
ORA-00279: change 1970578 generated at 10/06/2012 22:59:02 needed for thread 1
ORA-00289: suggestion :
/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_113_870kwzkl_.arc
ORA-00280: change 1970578 for thread 1 is in sequence #113
Specify log: {
auto
ORA-00279: change 1970779 generated at 10/06/2012 23:00:47 needed for thread 1
ORA-00289: suggestion :
/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_114_%u_.arc
ORA-00280: change 1970779 for thread 1 is in sequence #114
ORA-00278: log file
'/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_113_870kwzkl_.arc
' no longer needed for this recovery
ORA-00308: cannot open archived log
'/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_114_%u_.arc'
ORA-27037: unable to obtain file status
Linux Error: 2: No such file or directory
Additional information: 3
ORA-01547: warning: RECOVER succeeded but OPEN RESETLOGS would get error below
ORA-01194: file 1 needs more recovery to be consistent
ORA-01110: data file 1:
'/u01/oradata/WILSON/datafile/o1_mf_system_7xt3yzhj_.dbf'
SQL> alter database open resetlogs;
alter database open resetlogs
*
ERROR at line 1:
ORA-01194: file 1 needs more recovery to be consistent
ORA-01110: data file 1:
'/u01/oradata/WILSON/datafile/o1_mf_system_7xt3yzhj_.dbf'
尝试使用RMAN SCN方式打开。
SQL> select group#, sequence#, first_change# from v$log;
GROUP# SEQUENCE# FIRST_CHANGE#
---------- ---------- -------------
1 114 1970779
3 112 1950574
2 113 1970577
RMAN> recover database until sequence 112;
Starting recover at 06-OCT-12
allocated channel: ORA_DISK_1
channel ORA_DISK_1: SID=18 device type=DISK
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of recover command at 10/06/2012 23:19:42
RMAN-06556: datafile 1 must be restored from backup older than SCN 1950574
System文件的日期已经超过了recover范畴。只能尝试利用RMAN备份进行还原。
RMAN> restore database;
Starting restore at 06-OCT-12
using channel ORA_DISK_1
channel ORA_DISK_1: starting datafile backup set restore
channel ORA_DISK_1: specifying datafile(s) to restore from backup set
channel ORA_DISK_1: restoring datafile 00001 to /u01/oradata/WILSON/datafile/o1_mf_system_7xt3yzhj_.dbf
channel ORA_DISK_1: restoring datafile 00002 to /u01/oradata/WILSON/datafile/o1_mf_sysaux_7xt3yzkb_.dbf
channel ORA_DISK_1: restoring datafile 00003 to /u01/oradata/WILSON/datafile/o1_mf_undotbs1_7xt3yzl5_.dbf
channel ORA_DISK_1: restoring datafile 00004 to /u01/oradata/WILSON/datafile/o1_mf_users_805nxydh_.dbf
channel ORA_DISK_1: restoring datafile 00005 to /u01/oradata/WILSON/datafile/o1_mf_example_7xt46m9x_.dbf
channel ORA_DISK_1: restoring datafile 00006 to /u01/oradata/WILSON/datafile/o1_mf_nbscommo_820frtg1_.dbf
channel ORA_DISK_1: restoring datafile 00007 to /u01/oradata/WILSON/datafile/o1_mf_nbscommo_820ft5y5_.dbf
channel ORA_DISK_1: reading from backup piece /u01/flash_recovery_area/WILSON/backupset/2012_10_06/o1_mf_nnndf_TAG20121006T220912_870gwjoy_.bkp
channel ORA_DISK_1: piece handle=/u01/flash_recovery_area/WILSON/backupset/2012_10_06/o1_mf_nnndf_TAG20121006T220912_870gwjoy_.bkp tag=TAG20121006T220912
channel ORA_DISK_1: restored backup piece 1
channel ORA_DISK_1: restore complete, elapsed time: 00:02:46
Finished restore at 06-OCT-12
--进行部分恢复
RMAN> recover database until sequence 112;
Starting recover at 06-OCT-12
using channel ORA_DISK_1
starting media recovery
archived log for thread 1 with sequence 106 is already on disk as file /u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_106_870gzbfg_.arc
archived log for thread 1 with sequence 107 is already on disk as file /u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_107_870jlysq_.arc
archived log for thread 1 with sequence 108 is already on disk as file /u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_108_870jo98m_.arc
archived log for thread 1 with sequence 109 is already on disk as file /u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_109_870jx2gj_.arc
archived log for thread 1 with sequence 110 is already on disk as file /u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_110_870kgmdp_.arc
archived log for thread 1 with sequence 111 is already on disk as file /u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_111_870knhwb_.arc
archived log file name=/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_106_870gzbfg_.arc thread=1 sequence=106
archived log file name=/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_107_870jlysq_.arc thread=1 sequence=107
archived log file name=/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_108_870jo98m_.arc thread=1 sequence=108
archived log file name=/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_109_870jx2gj_.arc thread=1 sequence=109
archived log file name=/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_110_870kgmdp_.arc thread=1 sequence=110
archived log file name=/u01/flash_recovery_area/WILSON/archivelog/2012_10_06/o1_mf_1_111_870knhwb_.arc thread=1 sequence=111
media recovery complete, elapsed time: 00:00:20
Finished recover at 06-OCT-12
明显，得益于归档模式和备份集合。我们似乎成功的完成了recover过程。下面我们尝试开启数据库，注意这里已经是incomplete恢复了，我们必须承担这部分的数据损失。
SQL> conn / as sysdba
Connected.
SQL> alter database open resetlogs;
Database altered.
SQL> select group#, status, member from v$logfile;
GROUP# STATUS MEMBER
---------- ------- --------------------------------------------------------------------------------
3 /u01/oradata/WILSON/onlinelog/o1_mf_3_870n4lsg_.log
3 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_3_870n4o31_.log
2 /u01/oradata/WILSON/onlinelog/o1_mf_2_870n48hc_.log
2 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_2_870n4dtl_.log
1 /u01/oradata/WILSON/onlinelog/o1_mf_1_870n42n1_.log
1 /u01/flash_recovery_area/WILSON/onlinelog/o1_mf_1_870n44z3_.log
6 rows selected
SQL> select group#, status, archived, sequence# from v$log;
GROUP# STATUS ARCHIVED SEQUENCE#
---------- ---------------- -------- ----------
1 CURRENT NO 1
2 UNUSED YES 0
3 UNUSED YES 0
总结：当我们incomplete关闭数据库，又删除了Active状态日志组成员之后，我们必须要承担对应的数据损失，一部分数据可能因此丢失。笔者之后的处理，是借助之前rman的一个备份集合，进行非完全恢复到最后一个可以找到的日志文件组。这里，我们再次强调归档模式和备份的重要性。