Dataguard的强制切换

【前言】Dataguard的从库一般情况下都是出于数据的抽取和查询的作用的，但是万一在主库发生故障的情况下就需要切换到备库了。而这种故障的切换往往又是灾难性的情况：比如主库的服务器根本就起不来了，没有办法进行正常的切换，就需要强制的把从库切换成主库了。

【操作步骤】这个时候主库根本是操作不了的，所以只需在从库执行以下操作

【1】停止应用恢复模式

alter database recover managed standby database finish;

【2】从库切换成主库

alter database commit to switchover to primary;

【3】启动数据库

Alter database open;

【4】检查备库的状态

select open_mode,database_role from v$database;

OPEN_MODE       DATABASE_ROLE--------------      ---------------------- OPEN                 PRIMARY

【其他】切换完成后，就需要更改应用的连接IP地址和相关的一系列的连接（DBLINK等其他接口），如果再规划的时候就把这部分的内容考虑过去的话，那么就减少了很多切换后的工作。

如果服务器的IP地址不变的话，那么就可以节省很多的工作了。这就需要前期的规划：通过负载均衡设置一个虚拟ip，当服务器发生故障后，切换虚拟ip对应的实体ip就好了。如果没有这个硬件的话，通过软件也是可以实现的比如：keepalived工具等。

附：切换日志

1. alter database recover managed standby database finish 停止应用进行日志的恢复
2. Attempt to do a Terminal Recovery (DE2)
3. Media Recovery Start: Managed Standby Recovery (DE2)
4. started logmerger process
5. Wed Mar 29 10:21:35 2017
6. Managed Standby Recovery not using Real Time Apply
7. Parallel Media Recovery started with 8 slaves
8. Media Recovery Log /oracle/DE2/oraarch/DE2arch/1_565_912333510.dbf
9. Media Recovery Waiting for thread 1 sequence 566 (in transit)
10. krsv_proc_kill: Killing 4 processes (all RFS, wait for I/O)
11. Begin: Standby Redo Logfile archival
12. End: Standby Redo Logfile archival
13. Terminal Recovery timestamp is '03/29/2017 10:21:43'
14. Terminal Recovery: applying standby redo logs.
15. Terminal Recovery: thread 1 seq# 566 redo required
16. Terminal Recovery:
17. Recovery of Online Redo Log: Thread 1 Group 14 Seq 566 Reading mem 0
18. Mem# 0: /oracle/DE2/mirrlogA/sredo14.log
19. Identified End-Of-Redo (failover) for thread 1 sequence 566 at SCN 0xffff.ffffffff
20. Wed Mar 29 10:21:48 2017
21. Incomplete Recovery applied until change 21240383491 time 03/29/2017 10:27:41
22. Wed Mar 29 10:21:48 2017
23. Media Recovery Complete (DE2)
24. Terminal Recovery: successful completion
25. Forcing ARSCN to IRSCN for TR 4:4060514307
26. Attempt to set limbo arscn 4:4060514307 irscn 4:4060514307
27. Resetting standby activation ID 1489765315 (0x58cc03c3)
28. Wed Mar 29 10:21:51 2017
29. ARCH: Archival stopped, error occurred. Will continue retrying
30. ORACLE Instance DE2 - Archival Error
31. ORA-16014: log 14 sequence# 566 not archived, no available destinations
32. ORA-00312: online log 14 thread 1: '/oracle/DE2/mirrlogA/sredo14.log'
33. Completed: alter database recover managed standby database finish 完成日志的应用
34. Wed Mar 29 10:22:12 2017
35. alter database commit to switchover to primary 从库切换成主库
36. ALTER DATABASE SWITCHOVER TO PRIMARY (DE2)
37. Maximum wait for role transition is 15 minutes.
38. CLOSE: killing server sessions.
39. CLOSE: all sessions shutdown successfully.
40. Wed Mar 29 10:22:12 2017
41. SMON: disabling cache recovery
42. Backup controlfile written to trace file /oracle/DE2/saptrace/diag/rdbms/de2dg/DE2/trace/DE2_ora_9961494.trc
43. Standby terminal recovery start SCN: 21240295534
44. RESETLOGS after incomplete recovery UNTIL CHANGE 21240383491 重建redolog
45. Online log /oracle/DE2/mirrlogB/log_g15m1.dbf: Thread 1 Group 5 was previously cleared
46. Online log /oracle/DE2/mirrlogB/log_g15m2.dbf: Thread 1 Group 5 was previously cleared
47. Online log /oracle/DE2/mirrlogA/log_g16m1.dbf: Thread 1 Group 6 was previously cleared
48. Online log /oracle/DE2/mirrlogA/log_g16m2.dbf: Thread 1 Group 6 was previously cleared
49. Online log /oracle/DE2/mirrlogB/log_g17m2.dbf: Thread 1 Group 7 was previously cleared
50. Online log /oracle/DE2/mirrlogB/log_g17m1.dbf: Thread 1 Group 7 was previously cleared
51. Online log /oracle/DE2/mirrlogA/log_g18m2.dbf: Thread 1 Group 8 was previously cleared
52. Online log /oracle/DE2/mirrlogA/log_g18m1.dbf: Thread 1 Group 8 was previously cleared
53. Standby became primary SCN: 21240295533
54. Wed Mar 29 10:22:13 2017
55. Setting recovery target incarnation to 3
56. Switchover: Complete - Database mounted as primary
57. Completed: alter database commit to switchover to primary 完成切换
58. Wed Mar 29 10:22:37 2017
59. alter database open
60. Data Guard Broker initializing...
61. Data Guard Broker initialization complete
62. Data Guard: verifying database primary role...