DG物理standby，failover步骤

       环境为11gR2，oracle linux 6.4；物理standby，单实例；假设primary（db20）异常关闭数据库，需要切换物理standby（db66）为新primary 数据库；

步骤1：检查归档日志是否缺失

       如果异常的primary数据库还能够mount上，则启动到mount状态，然后执行SQL,补发日志；

       SQL> ALTER SYSTEM FLUSH REDO TO target_db_name;

target_db_name 为db_unique_name，log_archive_dest_n中指定的；如果执行成功，则可以保证数据无丢失；

       如果异常主库不能mount，则需要拷贝备库缺失的归档日志到备库，然后注册；

               SQL> ALTER DATABASE REGISTER PHYSICAL LOGFILE 'filespec1';

步骤2：备库停止redo应用

       db66：

       SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL;

步骤3：手动完成所有redo应用

       SQL> ALTER DATABASE RECOVER MANAGED STANDBY DATABASE FINISH;

如果执行没有报任何错误，则执行步骤4；如果报错并不能够解决，则执行activate，    SQL> ALTER DATABASE ACTIVATE PHYSICAL STANDBY DATABASE;

然后执行步骤6；

步骤4：查询备库状态

       SQL> SELECT SWITCHOVER_STATUS FROM V$DATABASE;

SWITCHOVER_STATUS

-----------------

TO PRIMARY

1 row selected

如果SWITCHOVER_STATUS 为TO PRIMARY 或 SESSIONS ACTIVE值，为正常状态，可以switchover为primary；

步骤5：切换到主库

       SQL> ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY WITH SESSION SHUTDOWN;

步骤6：打开新主库到open

       SQL> ALTER DATABASE OPEN;

       打开新主库；备份新主库，然后启动其他物理备库已经停止的日志应用；

打开新主库是报错：

SQL> startup open

ORACLE instance started.

Total System Global Area  839282688 bytes

Fixed Size                  2233000 bytes

Variable Size             637537624 bytes

Database Buffers          192937984 bytes

Redo Buffers                6574080 bytes

Database mounted.

ORA-03113: end-of-file on communication channel

Process ID: 12427

Session ID: 191 Serial number: 3

查看alert日志文件，如下：

Error 12514 received logging on to the standby

LGWR: Error 12514 verifying archivelog destination LOG_ARCHIVE_DEST_2

Destination LOG_ARCHIVE_DEST_2 is UNSYNCHRONIZED

LGWR: Continuing...

LGWR: Minimum of 1 applicable standby database required

Errors in file /u01/app/oracle/diag/rdbms/db66/oradb/trace/oradb_lgwr_12367.trc:

ORA-16072: a minimum of one standby database destination is required

LGWR (ospid: 12367): terminating the instance due to error 16072

System state dump requested by (instance=1, osid=12367 (LGWR)), summary=[abnormal instance termination].

System State dumped to trace file /u01/app/oracle/diag/rdbms/db66/oradb/trace/oradb_diag_12357.trc

Dumping diagnostic data in directory=[cdmp_20140228172732], requested by (instance=1, osid=12367 (LGWR)), summary=[abnormal instance termination].

Instance terminated by LGWR, pid = 12367

       通过上面的内容，发现报错在归档日志位置LOG_ARCHIVE_DEST_2，这个参数设置DG下primary数据库的日志传送；由于原来的DG是maximize protect模式，而在failover后，原primary无法达到，则报错；修改DG的保护模式为maximize availability即可

SQL> alter database set standby database to maximize availability;

Database altered.

SQL> select protection_mode,protection_level from v$database;

PROTECTION_MODE      PROTECTION_LEVEL

-------------------- --------------------

MAXIMUM AVAILABILITY RESYNCHRONIZATION