一、Goldengate双向复制配置
1.1.在进行如下配置之前，先在源数据库source system（原来的目标数据库）端
添加辅助的redolog配置：
SQL>alter database add supplemental log data;
SQL>alter system switch logfile;
SQL>alter database force logging;
1.2.在原Source和原Target分别用scott用户创建一张emp_ogg表
SQL> create table emp_ogg as select * from emp where 1=0; //原source库建立表但不插入数据
Table created.
单项复制时Source端是EINI_1进程，Target端是RINI_1进程,双向复制时反之，以便完成两表的初始化同步，配置如下：
原Source
GGSCI (gc2) 36> ADD REPLICAT RINI_1, SPECIALRUN
REPLICAT added.
GGSCI (gc2) 43> EDIT PARAMS RINI_1
-- GoldenGate Initial Load Delivery
--
REPLICAT RINI_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
ASSUMETARGETDEFS
USERID ogg, PASSWORD ogg
DISCARDFILE ./dirrpt/RINIaa.dsc, PURGE
MAP scott.*, TARGET scott.*;
原Target
GGSCI (oraclelinux54.cuug.net) 10> ADD EXTRACT EINI_1, SOURCEISTABLE
EXTRACT added.
GGSCI (oraclelinux54.cuug.net) 11> INFO EXTRACT *, TASKS
EXTRACT EINI_1 Initialized 2014-08-12 23:05 Status STOPPED
Checkpoint Lag Not Available
Log Read Checkpoint Not Available
First Record Record 0
Task SOURCEISTABLE
GGSCI (oraclelinux54.cuug.net) 12> EDIT PARAMS EINI_1
-- GoldenGate Initial Data Capture
-- for EMP_OGG and DEPT_OGG
--
EXTRACT EINI_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
USERID ogg, PASSWORD ogg
RMTHOST gc2, MGRPORT 7809
RMTTASK REPLICAT, GROUP RINI_1
TABLE scott.EMP_OGG; //这里我们只为初始化scott用户下的emp_ogg表
GGSCI (oraclelinux54.cuug.net) 19> START EXTRACT EINI_1
Sending START request to MANAGER ...
EXTRACT EINI_1 starting
GGSCI (oraclelinux54.cuug.net) 20> VIEW REPORT EINI_1
……...
***********************************************************************
* ** Run Time Statistics ** *
***********************************************************************
Report at 2014-08-12 23:11:04 (activity since 2014-08-12 23:10:59)
Output to RINI_1:
From Table SCOTT.EMP_OGG:
# inserts: 14 //可以看出两边已经同步成功
# updates: 0
# deletes: 0
# discards: 0
1.3.在ogg环境下，添加日志跟踪：
GGSCI (oraclelinux54.cuug.net) DBLOGIN USERID ogg, PASSWORD ogg
GGSCI (oraclelinux54.cuug.net) ADD TRANDATA scott.* //这里为了试验方便，我们设置为scott下全部表均设为可同步状态
2013-08-13 03:21:18 GGS WARNING 109 No unique key is defined for table EMP_OGG. All viable columns will be used to represent the key, but may not guarantee uniqueness. KEYCOLS may be used to define the key.
2013-08-13 03:21:18 GGS WARNING 301 Failed to add supplemental log group on table SCOTT.EMP_OGG due to ORA-01031: insufficient privileges, SQL ALTER TABLE "SCOTT"."EMP_OGG" ADD SUPPLEMENTAL LOG GROUP "GGS_EMP_OGG_74686" ("EMPNO","ENAME","JOB","MGR","HIREDATE","SAL","COMM","DEPTNO") ALWAYS .
解决办法：
SQL>alter table emp_ogg add constraint emp_ogg_pk primary key(empno);
SQL> grant alter any table to ogg;
Grant succeeded.
1.4.配置extract
GGSCI (gc2) 71> EDIT PARAMS EORA_1 //编辑前先停止进程
EXTRACT EORA_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
USERID ogg, PASSWORD ogg
EXTTRAIL ./dirdat/aa
TABLE scott.*;
DDL INCLUDE OBJNAME "scott.*"
TRANLOGOPTIONS EXCLUDEUSER ogg //双向复制关键
GGSCI (oraclelinux54.cuug.net) 74> EDIT PARAMS EORA_1 //编辑前先停止进程
EXTRACT EORA_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
USERID ogg, PASSWORD ogg
TRANLOGOPTIONS EXCLUDEUSER ogg
EXTTRAIL ./dirdat/aa
TABLE scott.*;
DDL INCLUDE OBJNAME "scott.*"
GGSCI (双节点执行)> ADD EXTRACT EORA_1, TRANLOG, BEGIN NOW
注：
ADD EXTRACT EORA_1:添加EXTRACT进程，这就是一直运行，一段停止redo 里的日志就没人去抓取了
TRANLOG, BEGIN NOW：现在开始同步日志，也可以用异步，那就要另外配置
EXTRACT added.
GGSCI (双节点执行)> ADD EXTTRAIL ./dirdat/aa, EXTRACT EORA_1, MEGABYTES 5 //添加跟踪文件给EORA_1用，大小为5M
EXTTRAIL added.
GGSCI (双节点执行)> START EXTRACT EORA_1
1.5.配置pump进程
GGSCI (gc2) 72> EDIT PARAMS PORA_1 //编辑前先停止进程
EXTRACT PORA_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
PASSTHRU
RMTHOST oraclelinux54.cuug.net, MGRPORT 7809
RMTTRAIL /u01/app/ogg/dirdat/pa
TABLE scott.*;
GGSCI (oraclelinux54.cuug.net) 75> EDIT PARAMS PORA_1 //编辑前先停止进程
EXTRACT PORA_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
PASSTHRU
RMTHOST gc2, MGRPORT 7809
RMTTRAIL ./dirdat/pa
TABLE scott.*;
GGSCI (双节点执行)> ADD EXTRACT PORA_1, EXTTRAILSOURCE ./dirdat/aa // 告诉PORA_1要传送哪个路径下的跟踪信息
EXTRACT added.
GGSCI (双节点执行)> ADD RMTTRAIL ./dirdat/pa, EXTRACT PORA_1, MEGABYTES 5 // 表示把捕获到的信息传送到远程的哪个目录的文件中
RMTTRAIL added.
GGSCI (双节点执行)> START EXTRACT PORA_1
配置replicate
GGSCI (gc2) 74> EDIT PARAM RORA_1
REPLICAT RORA_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
USERID ogg, PASSWORD ogg
HANDLECOLLISIONS
ASSUMETARGETDEFS
DISCARDFILE ./dirrpt/RORA_aa.DSC, PURGE
DDL INCLUDE ALL
DDLERROR DEFAULT IGNORE RETRYOP MAXRETRIES 3 RETRYDELAY 5
DDLERROR DEFAULT DISCARD
DDLERROR DEFAULT IGNORE RETRYOP
MAP scott.*, TARGET scott.*;
GGSCI (oraclelinux54.cuug.net) 87> EDIT PARAM RORA_1
REPLICAT RORA_1
SETENV (NLS_LANG=AMERICAN_AMERICA.ZHS16GBK)
USERID ogg, PASSWORD ogg
HANDLECOLLISIONS
ASSUMETARGETDEFS
DISCARDFILE ./dirrpt/RORA_aa.DSC, PURGE
DDL INCLUDE ALL
DDLERROR DEFAULT IGNORE RETRYOP MAXRETRIES 3 RETRYDELAY 5
DDLERROR DEFAULT DISCARD
DDLERROR DEFAULT IGNORE RETRYOP
MAP scott.*, TARGET scott.*;
GGSCI (双节点执行) 87> ADD REPLICAT RORA_1, EXTTRAIL ./dirdat/pa //表示从哪里提取传送过来的信息
GGSCI (双节点执行) 87> START REPLICAT RORA_1
SQL> conn / as sysdba
Connected.
SQL> grant insert on scott.emp_ogg to ogg;
Grant succeeded.
SQL> conn / as sysdba
Connected.
SQL> grant delete on scott.emp_ogg to ogg;
Grant succeeded.
SQL> grant update on scott.emp_ogg to ogg;
Grant succeeded.
二、配置checkpoint
GGSCI (oraclelinux54.cuug.net) EDIT PARAMS ./GLOBALS
CHECKPOINTTABLE ogg.ggschkpt
GGSCI (oraclelinux54.cuug.net) exit
GGSCI (oraclelinux54.cuug.net) DBLOGIN USERID ogg, PASSWORD ogg
Successfully logged into database.
GGSCI (oraclelinux54.cuug.net) ADD CHECKPOINTTABLE //如果单项复制已经做了，这里不用再添加
应该看到，两节点的进程都为Running状态
GGSCI (gc2) 73> info all
Program Status Group Lag Time Since Chkpt
MANAGER RUNNING
EXTRACT RUNNING EORA_1 00:00:00 00:00:02
EXTRACT RUNNING PORA_1 00:00:00 00:00:04
REPLICAT RUNNING RORA_1 00:00:00 00:00:01
GGSCI (oraclelinux54.cuug.net) 88> info all
Program Status Group Lag Time Since Chkpt
MANAGER RUNNING
EXTRACT RUNNING EORA_1 00:00:00 00:00:09
EXTRACT RUNNING PORA_1 00:00:00 00:00:02
REPLICAT RUNNING RORA_1 00:00:00 00:00:05
三、测试双向传送结果
3.1.source 到target
Source system
SQL> INSERT INTO emp_ogg VALUES(8000,'HL','CLERK',7902,'12-DEC-80',800,100,20);
1 row created.
SQL> commit;
Commit complete.
SQL> select * from emp_ogg;
EMPNO ENAME JOB MGR HIREDATE SAL COMM DEPTNO
---------- ---------- --------- ---------- --------- ---------- ---------- ----------
8000 HL CLERK 7902 12-DEC-80 800 100 20
Target system
SQL> select * from emp_ogg;
EMPNO ENAME JOB MGR HIREDATE SAL COMM DEPTNO
---------- ---------- ---------- ---------- --------------------- ---------- ---------- ----------
8000 HL CLERK 7902 1980-12-12 : 00:00:00 800 100 20
3.2target 到 source
Target system
SQL> select * from tcustmer;
CUST NAME CITY ST
---- ------------------------------ -------------------- --
WILL BG SOFTWARE CO. SEATTLE WA
JANE ROCKY FLYER INC. DENVER CO
SQL> insert into tcustmer values ('HL','zai','cuug','en');
1 row created.
SQL> commit;
Commit complete.
Source system
SQL> select * from tcustmer;
CUST NAME CITY ST
---- ------------------------------ -------------------- --
WILL BG SOFTWARE CO. SEATTLE WA
JANE ROCKY FLYER INC. DENVER CO
helei zai cuug en
四、支持DDL复制配置
4.1在两个节点执行执行DDL同步脚本命令：
先进入goldengate软件安装目录，以SYSDBA身份登录oracle执行以下脚本，执行脚本过程中，需要输入的用户全部是ogg，安装模式为INITIALSETUP，如果数据字典或者某些内部的包有错误，则需要运行catalog.sql和catproc.sql脚本。
SQL>show parameter recyclebin;
NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
recyclebin string off 必须是off
这里执行第二个脚本@ddl_setup时会报一个recyclebin的错误，而且只能在Pfile中修改recyclebin=off后用pfile起库，执行两个清除脚本，再重新运行一下脚本
SQL>@marker_setup
SQL>@ddl_setup
SQL>@role_setup
SQL>grant GGS_GGSUSER_ROLE to ogg;
SQL>@ddl_enable
如果某项脚本执行错误，需要重新执行时，先要执行清除的脚本：ddl_remove.sql和marker_remove.sql
