11G RAC TO 11G RAC ADG配置

联系:手机/微信(+86 17813235971) QQ(107644445)QQ咨询惜分飞

标题:11G RAC TO 11G RAC ADG配置

作者:惜分飞©版权所有[未经本人同意,不得以任何形式转载,否则有进一步追究法律责任的权利.]

RDBMS tnsnames.ora配置
在主备库的数据库的tns中均增加如下配置

q9dbdg =
  (DESCRIPTION =
    (ADDRESS = (PROTOCOL = TCP)(HOST = q9db01-dg-vip)(PORT = 1522))
    (ADDRESS = (PROTOCOL = TCP)(HOST = q9db02-dg-vip)(PORT = 1522))
    (CONNECT_DATA =
      (SERVER = DEDICATED)
      (SERVICE_NAME = q9db)
    )
)

q9adgdg =
  (DESCRIPTION =
    (ADDRESS = (PROTOCOL = TCP)(HOST = q9adg01-dg-vip)(PORT = 1522))
    (ADDRESS = (PROTOCOL = TCP)(HOST = q9adg02-dg-vip)(PORT = 1522))
    (CONNECT_DATA =
      (SERVER = DEDICATED)
      (SERVICE_NAME = q9db)
    )
)

主库修改参数配置

alter system set DB_UNIQUE_NAME=q9db scope=spfile;
alter system set LOG_ARCHIVE_CONFIG='DG_CONFIG=(q9db,q9db_adg)' scope=both;
alter system set LOG_ARCHIVE_DEST_1='LOCATION=+arch/q9db/ARCHIVELOG VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=q9db' scope=both;
alter system set LOG_ARCHIVE_DEST_2='SERVICE=q9adg1dg LGWR SYNC AFFIRM VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=q9db_adg';
alter system set standby_file_management='AUTO' sid='*' scope=both;
alter system set fal_client='q9dbdg';
alter system set fal_server='q9adgdg';
alter system set db_file_name_convert='+DATA','+DATA' SCOPE=SPFILE;
ALTER SYSTEM SET log_file_name_convert='+DATA','+DATA' SCOPE=SPFILE;

备库修改参数配置

alter system set DB_UNIQUE_NAME=q9db_adg scope=spfile;
alter system set LOG_ARCHIVE_CONFIG='DG_CONFIG=(q9db,q9db_adg)' scope=both;
alter system set LOG_ARCHIVE_DEST_1='LOCATION=+arch/q9db/ARCHIVELOG VALID_FOR=(ALL_LOGFILES,ALL_ROLES) DB_UNIQUE_NAME=q9db_adg';
alter system set LOG_ARCHIVE_DEST_2='SERVICE=q9db1dg LGWR SYNC AFFIRM VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE) DB_UNIQUE_NAME=q9db';
alter system set standby_file_management='AUTO' sid='*' scope=both;
alter system set fal_client='q9adgdg';
alter system set fal_server='q9dbdg';
alter system set service_names='q9db';
alter system set db_file_name_convert='+DATA','+DATA' SCOPE=SPFILE;
ALTER SYSTEM SET log_file_name_convert='+DATA','+DATA' SCOPE=SPFILE;

主库增加standby redo log

alter database add standby logfile THREAD 1 group 51 '+DATA' size 512m;
…………
alter database add standby logfile THREAD 1 group 61 '+DATA' size 512m;

alter database add standby logfile THREAD 2 group 71 '+DATA' size 512m;
…………
alter database add standby logfile THREAD 2 group 81 '+DATA' size 512m;

主库修改强制日志

SQL> ALTER DATABASE FROCE LOGGING;

RMAN备份主库

rman target /
run{
configure device type disk parallelism 16;
backup database filesperset = 10  format '/q9bak/df_%U';
backup current controlfile format '/q9bak/ctl_%U';
}

创建standby controlfile

ALTER DATABASE CREATE STANDBY CONTROLFILE AS '/q9bak/stb_ctl';

还原standby controlfile

ASMCMD> cp /q9bak/stb_ctl control01.ctl
copying /q9bak/stb_ctl -> +data/Q9DB/control01.ctl
ASMCMD> cp /q9bak/stb_ctl control02.ctl
copying /q9bak/stb_ctl -> +arch/q9db/control02.ctl

还原备库

SQL>startup mount;

rman target / nocatalog <<EOF
run{
configure default device type to disk;
configure device type disk parallelism 16;
restore database;
}
<<EOF

修改备库文件名

RMAN>catalog start with '+DATA/Q9DB/DATAFILE/';
RMAN> switch database to copy;

清理redo

--清理redo(所有的group#)
SQL> select group# from v$log; 
SQL> alter database clear logfile group 1; 
……

--清理standby redo(所有的group#)
SQL> select group# from v$standby_log; 
SQL> alter database clear logfile group 51; 
……

开启备库实时应用模式

SQL> alter database recover managed standby database using current logfile disconnect from session;

备库注册DB服务(oracle用户)

srvctl add database -d q9db_adg -o $ORACLE_HOME 
srvctl add instance -d q9db_adg -i q9db1 -n q9adg01
srvctl add instance -d q9db_adg -i q9db2 -n q9adg02
srvctl modify database -d q9db_adg -a DATA,ARCH
此条目发表在 Data Guard 分类目录。将固定链接加入收藏夹。

11G RAC TO 11G RAC ADG配置》有 2 条评论

  1. 惜分飞 说:

    只在一个节点开启,另外你归档放到共享目录,当一个节点的mrp出现问题的时候,你可以在另外一个节点启动mrp

    飞哥,请教一个问题;
    在RAC-RAC的ADG环境下,DG端的RAC有两个节点,如果设置的是主库把日志发送到备库的scan-ip这种情况下,备库这两个节点都需要开始mrp进程吗?如果只需要在一个节点开启mrp,那么没有开启的那个节点就无法接收并应用日志了吧?

  2. 冷场哥 说:

    飞哥,请教一个问题;
    在RAC-RAC的ADG环境下,DG端的RAC有两个节点,如果设置的是主库把日志发送到备库的scan-ip这种情况下,备库这两个节点都需要开始mrp进程吗?如果只需要在一个节点开启mrp,那么没有开启的那个节点就无法接收并应用日志了吧?