OGG在RAC上的初始化(下)-- OGG初始化

上篇地址http://blog.itpub.net/29047826/viewspace-1284906/

1，源端执行RMAN全库备份
备份路径为节点node1的本地磁盘。为了将该备份复制到灾备端，此次备份不能保存在ASM上。

run { allocate channel ch01 device type disk; allocate channel ch02 device type disk; backup format '/backup/full_%d_%s_%T_%p.bak'   database include current controlfile; sql 'alter system archive log current'; backup   archivelog all format '/backup/arch_%d_%s_%T_%p'; release channel ch02; release channel ch01; }

2，查看controlfile所在的备份片。灾备端恢复的时候也从这个备份片中读取controlfile。

RMAN> list backup of controlfile; List of Backup Sets =================== BS Key  Type LV Size       Device Type Elapsed Time Completion Time ------- ---- -- ---------- ----------- ------------ ------------------- 37      Full    17.80M     DISK        00:00:00     2014-09-28 23:53:06         BP Key: 37   Status: AVAILABLE  Compressed: NO  Tag: TAG20140928T235259         Piece Name: /backup/full_PROD_36_20140928_1.bak   Control File Included: Ckp SCN: 1260618      Ckp time: 2014-09-28 23:53:06

3，获取不活动的（INACTIVE）的已经归档的日志的最后一个SCN号，如果有多个已归档的INACTIVE的组，取最大的FIRST_CHANGE# 1260645
该SCN号作为灾备端recover database until 的SCN，并且在复制进程重新启动后也要参照这个SCN。

SYS@PROD1 > select inst_id,group#,thread#,sequence#,archived,status,first_change# from gv$log order by status;    INST_ID     GROUP#    THREAD#  SEQUENCE# ARC STATUS           FIRST_CHANGE# ---------- ---------- ---------- ---------- --- ---------------- -------------          1          3          2         25 NO  CURRENT                1260660          2          3          2         25 NO  CURRENT                1260660          1          1          1         29 NO  CURRENT                1260655          2          1          1         29 NO  CURRENT                1260655          1          2          1         28 YES INACTIVE               1260638          1          4          2         24 YES INACTIVE               1260645          2          2          1         28 YES INACTIVE               1260638          2          4          2         24 YES INACTIVE               1260645

4，将源端备份复制到灾备端

[oracle@node1 ~]$ scp /backup/* oracle@172.16.228.103:/backup/


5，停止灾备端的复制进程准备恢复数据库

GGSCI (node3) 4> stop rt1

Sending STOP request to REPLICAT RT1 ...
Request processed.

停止复制进程后，可以在源端继续执行一些DML操作。这些操作会正常的通过抽取进程获取，并有投递进程投递到灾备端的remote trail目录，只是暂时不用复制进程追加而已。

6，关闭灾备端数据库
[root@node3 ~]# srvctl stop database -d prod

7，灾备端从RMAN启动数据库到nomount状态

[root@node3 ~]# su - oracle [oracle@node3 ~]$ rman target / RMAN> startup nomount; Oracle instance started Total System Global Area     672256000 bytes Fixed Size                     2256272 bytes Variable Size                520094320 bytes Database Buffers             146800640 bytes Redo Buffers                   3104768 bytes

8，灾备端从RMAN恢复控制文件

RMAN> restore controlfile from '/backup/full_PROD_24_20140928_1.bak'; Starting restore at 2014-09-28 20:14:13 allocated channel: ORA_DISK_1 channel ORA_DISK_1: SID=34 instance=PROD1 device type=DISK channel ORA_DISK_1: restoring control file channel ORA_DISK_1: restore complete, elapsed time: 00:00:01 output file name=+DATA/prod/controlfile/current.260.859387343 output file name=+ARCH/prod/controlfile/current.256.859387345 Finished restore at 2014-09-28 20:14:14

9，灾备端控制文件加载复制过来的备份集信息,核对控制文件信息

RMAN> catalog start with '/backup/';

RMAN> list backup;

10，灾备端从RMAN恢复数据库

RMAN> restore database;

11，灾备端从RMAN恢复归档日志

RMAN> run{
allocate channel c1 type disk;
allocate channel c2 type disk;
restore archivelog all;
release channel c1;
release channel c2;
}

12，灾备端从RMAN指定SCN恢复。该SCN是源端数据库执行RMAN备份后状态为INACTIVE中数值最大的SCN。

RMAN> recover database until scn 1260645

13，灾备端在mount状态下通过控制文件查看文件头的scn是否一致

SYS@PROD1 > select checkpoint_change# from v$datafile_header; CHECKPOINT_CHANGE# ------------------            1260645            1260645            1260645            1260645            1260645            1260645            1260645 7 rows selected. SYS@PROD1 > select checkpoint_change# from v$datafile; CHECKPOINT_CHANGE# ------------------            1260645            1260645            1260645            1260645            1260645            1260645            1260645

14，灾备端resetlogs方式开启数据库

alter database open resetlogs;

[root@node3 ~]# srvctl start database -d prod

15，灾备端重新配置checkpoint

[oracle@node3 goldengate]$ ggsci GGSCI (node3) 1> DBLOGIN USERID oggadmin,PASSWORD oggadmin GGSCI (node3) 2> ADD CHECKPOINTTABLE oggadmin.checkpointtable GGSCI (node3) 14> DELETE REPLICAT rt1 GGSCI (node3) 15> ADD REPLICAT rt1, EXTTRAIL /goldengate/dirdat/rt, checkpointtable oggadmin.checkpointtable

16，灾备端重新启动复制进程，从SCN为1260645开始
注意这个单词不要写错aftercsn

GGSCI (node3) 16> start rt1 aftercsn 1260645 Sending START request to MANAGER ... REPLICAT RT1 starting

17，灾备端通过sqlplus查询恢复数据库期间生成的DML是否被成功追加
sqlplus / as sysdba
select * from snow.t1

本次初始化演练到此结束。