我的记录 2016-05-04
某数据库配置了Physical DataGuard(以下简称dg),由于没有检查,所以在主库上把归档日 删除了,而备库又没有完全应用,所以导致备库一直停留在之前的归档日志中。
主库检查,发现dest_id为的25号日志就丢失了.
PRIMARY> select dest_id,sequence#,applied,status,first_change#,next_change# from v$archived_log
where sequence#>20 order by dest_id,sequence#;
DEST_ID SEQUENCE# APPLIED S FIRST_CHANGE# NEXT_CHANGE#
---------- ---------- --------- - ------------- ------------
1 21 NO D 1765541 1765764
1 22 NO D 1765764 1768228
1 23 NO D 1768228 1768269
1 24 NO D 1768269 1768400
1 25 NO X 1768400 1768516
1 26 NO X 1768516 1768531
1 27 NO X 1768531 1768611
2 21 YES A 1765541 1765764
2 22 YES A 1765764 1768228
2 23 YES A 1768228 1768269
2 24 NO A 1768269 1768400
DEST_ID SEQUENCE# APPLIED S FIRST_CHANGE# NEXT_CHANGE#
---------- ---------- --------- - ------------- ------------
2 26 NO A 1768516 1768531
2 27 NO A 1768531 1768611
所以备库一直无法使用.这种情况下要么就重做整个备库,或是从备库断点的日志那个时候起恢复数据库.
由于数据库比较大,决定采用第二种方案;
在备库上,我们查看当前的数据文件最大的SCN
STANDBY> select name,file#,checkpoint_change# from v$datafile order by checkpoint_change#;
NAME FILE# CHECKPOINT_CHANGE#
------------------------------------------------------- ---------- ------------------
/u01/app/Oracle/oradata/dc1stby/undotbs01.dbf 4 1768269
/u01/app/oracle/oradata/dc1stby/sysaux01.dbf 3 1768269
/u01/app/oracle/oradata/dc1stby/system01.dbf 1 1768269
/u01/app/oracle/oradata/dc1stby/users01.dbf 6 1768269
在主库上做一个增量SCN备份,起始的位置就是备库数据文件的最小SCN号.
[oracle@newplat ~]$ rman target / nocatalog
[oracle@newplat ~]$ $ORACLE_HOME/bin/rman target / nocatalog
RMAN> run
2> {
3> allocate channel c3 device type disk;
4> backup as compressed backupset incremental from scn 1768269 database format '/home/oracle/bak/%U';
5> release channel c3;
6> }
released channel: ORA_DISK_1
allocated channel: c3
channel c3: SID=42 device type=DISK
.....................................................
including current control file in backup set
channel c3: starting piece 1 at 03-APR-16
channel c3: finished piece 1 at 03-APR-16
piece handle=/home/oracle/bak/0rr22q7u_1_1 tag=TAG20160403T022709 comment=NONE
channel c3: backup set complete, elapsed time: 00:00:01
Finished backup at 03-APR-16
备份完成后,我们通过SCP很快把备份文件传到备库的机器上来
[oracle@newplat bak]$ scp *1_1 192.168.56.22:/home/oracle/bak
这时候我们需要关闭备库,然把把实例启动到nomount关态
STANDBY> shutdown immediate;
ORA-01109: database not open
Database dismounted.
ORACLE instance shut down.
STANDBY> startup nomount;
ORACLE instance started.
Total System Global Area 1224736768 bytes
Fixed Size 2923824 bytes
Variable Size 939524816 bytes
Database Buffers 268435456 bytes
Redo Buffers 13852672 bytes
实例启动到mount状态后,我们进入备库的rman,这时候需要恢复控制文件,原备库的控制文件因为
包含了redo apply老的应用信息所以不能用了.
[oracle@dg2 bak]$ rman target / nocatalog
RMAN> restore standby controlfile from '/home/oracle/bak/0rr22q7u_1_1';
Starting restore at 03-APR-16
allocated channel: ORA_DISK_1
channel ORA_DISK_1: SID=23 device type=DISK
............................................
很快控制文件就了恢复完成了,我们打开另一个窗口,通过sqlplus把备库以standby的方式mount起来
STANDBY> alter database mount standby database;
Database altered.
数据库mount起来后,我们返回到rman提示符,这个时候我们可以对传过来的备份文件并进行编目了.这样我们在下一步操作中,就可以使用这些文件来行恢复
RMAN> catalog start with '/home/oracle/bak';
searching for all files that match the pattern /home/oracle/bak
List of Files Unknown to the Database
=====================================
File Name: /home/oracle/bak/08r1ekse_1_1
File Name: /home/oracle/bak/0or22q7t_1_1
File Name: /home/oracle/bak/0kr22phj_1_1
File Name: /home/oracle/bak/0nr22pic_1_1
File Name: /home/oracle/bak/09r1eku6_1_1
File Name: /home/oracle/bak/0rr22q7u_1_1
Do you really want to catalog the above files (enter YES or NO)? yes
cataloging files...
cataloging done
文件编目后,我们就可以开始正式恢复数据库了.
RMAN> recover database;
Starting recover at 03-APR-16
using channel ORA_DISK_1
starting media recovery
.....................................
media recovery complete, elapsed time: 00:03:01
Finished recover at 03-APR-16
备库完成恢复完成后,我们就可以启动redo apply 了.
STANDBY> alter database recover managed standby database disconnect from session using current logfile;
Database altered.
STANDBY> select process,client_process,sequence#,status from v$managed_standby;
PROCESS CLIENT_P SEQUENCE# STATUS
--------- -------- ---------- ------------
ARCH ARCH 0 CONNECTED
ARCH ARCH 0 CONNECTED
ARCH ARCH 0 CONNECTED
ARCH ARCH 0 CONNECTED
ARCH ARCH 0 CONNECTED
RFS ARCH 0 IDLE
RFS LGWR 33 IDLE
MRP0 N/A 33 WAIT_FOR_LOG
现在我们查看备库状态,MRP进程已经开始最新的日志应用了.到此我们通过增量SCN备份和恢复来修复备库归档日志丢失的过程已经全部完成了。