1.long 字段的无法使用OGG 同步

2.clob字段的导入导出Bug , 生产使用network-link 导入导出太慢了，本地导入导出速度会快3到4倍 。但是测试环境的情况却相反

测试环境和生产环境测试结果如下： 最后还是以生产环境的数据为主。

原因有2点，第一 ： 测试环境的网络没有拥塞，所以感觉测试环境network-link I/O 速度可以接受。第二测试环境的I/O 太慢了，跟生产环境不是一个等量级，所以感觉生产环境 本地I/O 会快很多。

数据局迁移遇到的Bug

1.DRAFT_LOG 这个表有6G ，有clob 字段

2．_LOG 这张表因为有clob 字段，这张表是核心表，导出时间和导入时间会有比较长。大概有19G ，

因为有clob 字段，导出时间会比较长。根据我们测试，

expdp /impdp 索引分离迁移  http://blog.51cto.com/itbull/1078724

数据泵 expdp导出

nohup expdp  system/******** dumpfile=lysb_20121113_%U.dmp directory=dmp_dir schemas=scott logfile=expdp_20121113.log job_name=lysbexpdp &

index导入时会占用大量时间，导入时先导入处index外的其他对象包括数据

nohup impdp system/******** dumpfile=lysb_20121113_%U.dmp directory=dmp_dir schemas=scott logfile=impdp_20121113.log job_name=lysbexpdp TABLE_EXISTS_ACTION=REPLACE EXCLUDE=INDEX &

index可以再重建或者单独导入

nohup impdp system/******** dumpfile=lysb_20121113_%U.dmp directory=dmp_dir schemas=scott logfile=impdp_201211132.log job_name=lysbexpdp include=INDEX &

测试1：（测试环境），不使用network_link 方式，

EXPDP 这2张表时间会达到3个小时。

IMPDP 这2张表时间也会超过3个小时。

测试２:  (测试环境) ，使用network_link 方式，

时间会缩短一倍，达到3个半小时。

测试３：（生产环境），使用network_link 方式，

从上午11点25发起，持续到16点，已经跑了4个半小时，还在导入。

测试4:  (生产环境) ，使用本地导入导出方式，不使用network_link 方式，除去BAK_RGCT_DRAFT_LOG这个6G表。

EXPDP 所有生产用户时间会达到1个小时10分钟。

IMPDP 所有生产用户时间会达到1个小时10分钟。

可以在2个半小时左右完成，

1.

检查项 "1.创建空库不需要初始化
2.检查生产的FRA空间检查
生产的在维护期间会创建一个回滚点并保留7天，所以要检查灾备FRA的卷空间是否能保留7天的日志量。不足需申请扩空间。"
新建目标空库 "1，只包含system,sysaux,undotbs1,temp,users 表空间，不需要初始化。
set line 200 pagesize 200
select tablespace_name from dba_tablespaces a
where (a.tablespace_name not in ('TEMP','SYSAUX','SYSTEM','USERS')
and a.contents not in ('TEMPORARY','UNDO')) ;
2，字符集、block size与源库保持一致
select * from v$nls_parameters where parameter='NLS_CHARACTERSET';
show parameter block_size;
3，安装组件和源生产一致。
select comp_name,version,status from dba_registry order by 1;
4.源库和目标库具有相同的timeone；
SELECT VERSION FROM V$TIMEZONE_FILE;

8.db_cache_size
设置为和源库相同
show parameter db_cache_size
9.注：9.1.目标库cpu_count 不要在spfile里进行设置，不需要和源库保持一致
9.2.目标库的parallel_max_servers设置要和源库一致。
10.job_queue_processes，aq_tm_processes 设置为0。
11.修改新建的目标库为非归档模式
12，11g数据库参数按标准新建库手册。
13. 关闭Flashback。
select flashback_on from v$database; --结果为NO
15.awr配置为间隔15分钟，保留90天
exec dbms_workload_repository.modify_snapshot_settings(interval=>15, retention=>90*24*60);
col SNAP_INTERVAL format a20
col RETENTION format a20
select * from dba_hist_wr_control;"
检查迁移对象是否是闭包的 "1：检查迁移用户引用非迁移用户的对象，如果引用的是系统表酌情处理；如果引用的是非系统表则要整改。
select * from dba_dependencies dd where dd.owner in
('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') --替换为实际用户
and dd.referenced_owner not in
('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE')
2：检查非迁移用户引用迁移用户的对象，如果存在则与开发讨论处理方法。
select * from dba_dependencies dd where dd.type<>'SYNONYM' and dd.owner not in
('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') ----替换为实际用户
and dd.referenced_owner in
('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') ----替换为实际用户

/"
检查迁移schema的trigger是否是闭包的 "1:检查触发器的owner<>table_owner且owner不在迁移用户列表而table_owner在迁移用户列表的情况，这种触发器需要手工迁移
select * from dba_triggers where owner<>table_owner and owner not in ('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') and table_owner in ('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') ;

2:检查触发器的owner<>table_owner且owner在迁移用户列表而table_owner不在迁移用户列表的情况，这种触发器视情况决定是否迁移
select * from dba_triggers where owner<>table_owner and owner not in ('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') and table_owner in ('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') ;

3:检查触发器的triggering_eventowner'LOGON '且owner在迁移用户列表或table_owner在迁移用户列表的情况，这种触发器expdp不能导出需要手工迁移
select * from dba_triggers where triggering_event='LOGON 'and (owner in ('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE') or table_owner in ('BBSP','BBSPODS','BCDPETL','DBMONOPR','DBMGR','OVSEE'));

"
根据10g生产库中待迁移job清单信息，生成job迁移脚本并在测试环境验证通过。特别注意：脚本所有待迁移job其next_date都必须设为无限大。 t_update_status();
检查10g生产库中是否有业务scheduler存在，这些scheduler也要进行迁移，如果存在有，本步要把迁移scheduler的脚本准备好 0

"
创建11g目标库到10g生产库的dblink" "create public database link dblink_migup connect to dbmgr identified by &pw
using ' 生产库连接串 ';
3:验证dblink是否通
select * from dual@dblink_migup; --有结果返回表示通的否则不通需处理
"
"
1. 创建要迁移的用户清单配置表(sys..t_user_cfg)

" "create table sys.t_user_cfg as SELECT username, default_tablespace
FROM dba_users@dblink_migup
WHERE username not in
('SYS', 'SYSTEM', 'PUBLIC', 'OUTLN', 'WMSYS', 'ORDSYS', 'MDSYS',
'CTXSYS', 'XDB', 'WKSYS', 'ANONYMOUS', 'ORDPLUGINS', 'EXFSYS',
'LBACSYS', 'DIP', 'LBACSYS', 'ORACLE_OCM', 'TSMSYS', 'ADAMS',
'TRACESVR', 'DBSNMP', 'PERFSTAT', 'DBSTATS', 'EXPFULL',
'MDDATA', 'MGMT_VIEW', 'OLAPSYS', 'SI_INFORMTN_SCHEMA', 'OWBSYS',
'SYSMAN', 'ORDDATA', 'OWBSYS_AUDIT', 'APEX_030200', 'APPQOSSYS',
'MDSYS', 'FLOWS_FILES', 'SPATIAL_WFS_ADMIN_USR',
'SPATIAL_CSW_ADMIN_USR', 'APEX_PUBLIC_USER', 'XS$NULL','DMSYS')
minus
select username, default_tablespace from dba_users;"
目标端新增表空间 "1.查看源库表空间名称及大小，在目标库新建相同大小表空间
set line 200 pagesize 200
select tablespace_name from dba_tablespaces a
where a.tablespace_name not in ('TEMP','SYSAUX','SYSTEM','USERS') ;

--注意临时表空间是否有特殊的，CREATE TEMPORARY TABLESPACE RLMSTMP TEMPFILE '+DATA_HDG' SIZE 100M AUTOEXTEND ON;

2.打开所有表空间的自动扩展
select 'alter database datafile '''||file_name||''' autoextend on next 100M;' from dba_data_files ;
select 'alter database tempfile '''||file_name||''' autoextend on next 100M;' from dba_temp_files ;
执行上面生成的脚本
复核：
select df.tablespace_name,df.autoextensible from dba_temp_files df where df.autoextensible='NO';
select df.tablespace_name,df.autoextensible from dba_data_files df where df.autoextensible='NO';"
目标端创建role "SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/01_create_role.sql
select 'create role ' || role || ';'
from (select role
from dba_roles@dblink_migup
minus
select role from dba_roles);
spool off;
在目标库执行01_create_role.sql"
创建密码函数 "执行附件脚本创建密码函数sys.VERIFY_FUNCTION_11G
PS:这个密码函数如果老库的名称不一样，需要把老库的函数迁移过来。"
创建profile "SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/02_deal_profile.sql
select (case
when rn = 1 then
('create profile ' || profile ||' limit ' || resource_name ||' ' || limit || (case
when rn = max_rn then
';'
end)) else(resource_name || ' ' || limit || (case
when rn = max_rn then
';'
end)) end)
from (select profile,
resource_name,
limit,
row_number() over(partition by profile order by profile, resource_name) rn,
count(*) over(partition by profile) max_rn
from dba_profiles@dblink_migup
where profile in
(select distinct profile
from dba_profiles@dblink_migup
minus
select distinct profile from dba_profiles)
order by profile, resource_name)
order by profile, resource_name;
spool off;

执行02_deal_profile.sql

alter profile default limit PASSWORD_VERIFY_FUNCTION VERIFY_FUNCTION_11G;
alter profile default limit FAILED_LOGIN_ATTEMPTS 10;
复核：
select * from dba_profiles minus select * from dba_profiles@dblink_migup;
select * from dba_profiles@dblink_migup minus select * from dba_profiles;"
创建公共同义词 "SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/03_create_public_synonym.sql
SELECT
'create or replace public synonym '||synonym_name||' for '||table_owner||'.'||table_name||';'
FROM dba_synonyms@dblink_migup
WHERE owner = 'PUBLIC'
AND table_owner IN ( select username from sys.t_user_cfg t) ;
spool off;
在目标库执行03_create_public_synonym.sql"
移植cron "
将原hp平台上的crontab移植到新linux主机上。
"

2.

21:00 导出BBSP.RGCT_DRAFT_LOG "export NLS_LANG=AMERICAN_AMERICA.ZHS16GBK
cd /db/db1/oradata/dba/exp_table
vi expdp_rgct_log.par
userid='/ as sysdba'
directory=schema_exp
dumpfile=expdp_rgct_log_%U.dmp
logfile=expdp_rgct_draft_log.log
parallel=4
filesize=10240M
tables=(BBSP.RGCT_DRAFT_LOG)
content=all "
22:00 scp传输到linux scp *.dmp cradmin@10.198.217.193:/db/db1/app/opdb1/db1_impdp
22:00 新库创建用户 "create user bbsp identified by values 'DF01EB216CE6E6BE' default tablespace bbsp profile default;
grant unlimited tablespace to bbsp;
alter user bbsp quota unlimited on users;
alter user bbsp quota unlimited on bbsp;
"
22:10 导入BBSP.RGCT_DRAFT_LOG "chown -R opdb1:dba /db/db1/app/opdb1/db1_impdp/*.dmp

export NLS_LANG=AMERICAN_AMERICA.ZHS16GBK
cd /db/db1/app/opdb1/db1_impdp
vi impdp_rgct_log.par
userid='/ as sysdba'
directory=schema_imp
dumpfile=expdp_rgct_log_%U.dmp
logfile=impdp_rgct_draft_log.log
parallel=4
tables=(BBSP.RGCT_DRAFT_LOG)
content=all

nohup impdp parfile=impdp_rgct_log.par &

检查导出导入日志是否有报错"
T-0 "DBA检查DB中的long oper、
rbs占用，
并请运营确认和做相应处理" "sqlplus '/ as sysdba'
alter system switch logfile ;
alter system switch logfile ;
alter system switch logfile ;
alter system archive log current;

1）检查long operation
select sid, target,opname, sofar,totalwork,time_remaining
from gv$session_longops where time_remaining>0;
2）检查占用回滚段较多的事务：
select xidusn,s.last_call_et, s.username, used_ublk,USED_UREC,space,
RECURSIVE, NOUNDO, START_UBAFIL,start_ubablk, start_uext
from gv$transaction t, gv$session s
where t.ses_addr=s.saddr;
"
1:00 维护时间开始处理，通知应用停止应用 应用将库的所有相关应用系统和job、ETL任务等停止下来。
T-0 检查卷，冻结源库双机软件 "冻结源库的双机，
冻结目标库的双机"
T-0 全备，磁带永久保留
T-0 DBA停止报警，屏蔽cron "cd $HOME
touch cron_null
crontab cron_null
crontab -l 返回值应该为空"
T-0 修改端口 "lsnrctl stop db1
修改监听端口为1537
cd $ORACLE_HOME/network/admin
vi listener.ora
lsnrctl start db1"
T-0 重建到旧库的db links。 "1:在目标库重建迁移用的临时dblink ---直接使用连接串
drop public database link dblink_migup;
create public database link dblink_migup connect to dbmgr identified by db1234DBA
using ' (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST =10.198.69.10)(PORT = 1537)) (CONNECT_DATA =(sid = db1)))';
2:验证dblink是否通
select * from dual@dblink_migup; --有结果返回表示通的否则不通需处理"
T-0 修改job_queue_processes，aq_tm_processes为0，关闭audit "保留记录job_queue_processes及aq_tm_processes修改前的值。
show parameter job_queue_processes --- 1000
show parameter aq_tm_processes ---1
alter system set job_queue_processes=0 scope=both;
alter system set aq_tm_processes=0 scope=both;
alter system set audit_trail=none scope=spfile;"
T-0 "DBA检查DB中的long oper、
rbs占用，
并请运营确认和做相应处理" "sqlplus '/ as sysdba'
alter system switch logfile ;
alter system switch logfile ;
alter system switch logfile ;
alter system archive log current;

1）检查long operation
select sid, target,opname, sofar,totalwork,time_remaining
from gv$session_longops where time_remaining>0;
2）检查占用回滚段较多的事务：
select xidusn,s.last_call_et, s.username, used_ublk,USED_UREC,space,
RECURSIVE, NOUNDO, START_UBAFIL,start_ubablk, start_uext
from gv$transaction t, gv$session s
where t.ses_addr=s.saddr;
"
T-0 修改stream_pool_size alter system set streams_pool_size=128M scope=both;
T-0 1.设置JOB的next_date为无限大 "备份原库的Job:create table dbmgr.dbajobs_old_bak as select * from dba_jobs;
复核：SELECT COUNT(*) FROM dbmgr.dbajobs_old_bak;
设置JOB的next_date为无限大："
T-0 原库创建用户清单表sys.t_user_list "create table sys.t_user_cfg as SELECT username, default_tablespace
FROM dba_users
WHERE username not in
('SYS', 'SYSTEM', 'PUBLIC', 'OUTLN', 'WMSYS', 'ORDSYS', 'MDSYS',
'CTXSYS', 'XDB', 'WKSYS', 'ANONYMOUS', 'ORDPLUGINS', 'EXFSYS',
'LBACSYS', 'DIP', 'LBACSYS', 'ORACLE_OCM', 'TSMSYS', 'ADAMS',
'TRACESVR', 'DBSNMP', 'PERFSTAT', 'DBSTATS', 'EXPFULL',
'MDDATA', 'MGMT_VIEW', 'OLAPSYS', 'SI_INFORMTN_SCHEMA', 'OWBSYS',
'SYSMAN', 'ORDDATA', 'OWBSYS_AUDIT', 'APEX_030200', 'APPQOSSYS',
'MDSYS', 'FLOWS_FILES', 'SPATIAL_WFS_ADMIN_USR',
'SPATIAL_CSW_ADMIN_USR', 'APEX_PUBLIC_USER', 'XS$NULL','DMSYS');
grant select on sys.t_user_cfg to public;"
T-0 部署数据比对的procedure "grant select on dba_tab_columns to dbmgr;
create table dbmgr.TAB_HASH
(
OWNER VARCHAR2(30),
TABLE_NAME VARCHAR2(50),
SQL_VALUE VARCHAR2(4000),
HASH_VALUE NUMBER(20)
) ;

CREATE OR REPLACE PROCEDURE DBMGR.GET_TAB_HASH8(V_OWNER VARCHAR2,
V_TABLE_NAME VARCHAR2) AS
v_sql_temp varchar2(4000);
v_hash number(20);
BEGIN
FOR r in (SELECT COLUMN_NAME
FROM DBA_TAB_COLUMNS
WHERE OWNER = V_OWNER
AND TABLE_NAME = V_TABLE_NAME
AND DATA_TYPE NOT IN ('CLOB', 'BLOB')
and OWNER IN (SELECT USERNAME FROM sys.t_user_cfg)
order by 1 asc) loop
v_sql_temp := v_sql_temp || '||' || R.COLUMN_NAME;
end loop;
v_sql_temp := 'select/*+ parallel(t 8) full(t)*/ sum(dbms_utility.get_hash_value(' ||
substr(v_sql_temp, 3) || ', 0, 999999999)) from ' ||
V_OWNER || '.' || V_TABLE_NAME || ' t';

execute immediate v_sql_temp
into v_hash;

insert into tab_hash
(owner, table_name, sql_value, hash_value)
values
(V_OWNER, V_TABLE_NAME, v_sql_temp || ';', v_hash);
commit;
EXCEPTION
WHEN OTHERS THEN
NULL;
END;
/"
T-0 为GG回滚库做的disable/enable trigger以及constraint脚本。 "set linesize 300

SPOOL $HOME/disable_constraint_r.sql
SELECT 'spool ./disable_constraint_r.log' FROM dual;
SELECT 'alter table '||owner||'.'||table_name||' disable constraint '||constraint_name||';'
FROM dba_constraints
WHERE owner IN (select username from sys.t_user_cfg)
AND constraint_type = 'R'
ORDER BY constraint_type desc, owner, table_name, constraint_name;
SELECT 'spool off' FROM dual;
SPOOL OFF

SPOOL ./disable_trigger.sql
SELECT 'spool ./disable_trigger.log' FROM dual;
SELECT
'alter trigger '||owner||'.'||trigger_name||' disable;'
FROM dba_triggers
WHERE owner IN (select username from sys.t_user_cfg)
AND status <> 'DISABLED'
ORDER BY owner, table_name, trigger_name;
SELECT 'spool off' FROM dual;
SPOOL OFF
SPOOL ./enable_trigger.sql
SELECT 'spool ./enable_trigger.log' FROM dual;
SELECT
'alter trigger '||owner||'.'||trigger_name||' enable;'
FROM dba_triggers
WHERE owner IN (select username from sys.t_user_cfg)
AND status = 'ENABLED'
ORDER BY owner, table_name, trigger_name;
SELECT 'spool off' FROM dual;
SPOOL OFF

SPOOL ./enable_constraint_r_novalidate.sql
SELECT 'spool ./enable_constraint_r_novalidate.log' FROM dual;
SELECT
'alter table '||owner||'.'||table_name||' modify constraint '||constraint_name||' enable novalidate;'
FROM dba_constraints
WHERE owner IN (select distinct username from sys.t_user_cfg)
AND constraint_type = 'R'
AND STATUS = 'ENABLED'
ORDER BY constraint_type desc, owner, table_name, constraint_name;
SELECT 'spool off' FROM dual;
SPOOL OFF

SPOOL ./enable_constraint_src_r.sql
SELECT 'spool ./enable_constraint_src_r.log' FROM dual;
SELECT
'alter table '||owner||'.'||table_name||' modify constraint '||constraint_name||' enable novalidate;'
FROM dba_constraints
WHERE owner IN (select distinct username from sys.t_user_cfg)
AND constraint_type = 'R'
AND status = 'ENABLED'
ORDER BY constraint_type desc, owner, table_name, constraint_name;
SELECT 'spool off' FROM dual;
SPOOL OFF"
T-0 重启生产库 "
shutdown immediate;
startup
"
T-0 停止同城灾备db 没有灾备库
T-0 清空回收站 Purge dba_recyclebin;
T-0 expdp导出 "export NLS_LANG=AMERICAN_AMERICA.ZHS16GBK
connect / as sysdba
create or replace directory schema_exp as '/db/db1/oradata/expdp_db1';
cd /db/db1/oradata/expdp_db1

userid='/ as sysdba'
directory=schema_exp
dumpfile=expdp_schemas_%U.dmp
logfile=expdp_schemas.log
parallel=6
filesize=10240M
schemas=BBSP,BBSPODS,BCDPETL,DBMONOPR
content=all
EXCLUDE=TABLE:""in ('BAK_RGCT_DRAFT_LOG','RGCT_DRAFT_LOG')""

nohup expdp parfile=expdp.par &"
检查error more expdp_schemas.log |grep -i error
T-0 impdp导入 "scp *.dmp cradmin@10.198.217.193:/db/db1/app/opdb1/db1_impdp
chown -R opdb1:dba /db/db1/app/opdb1/db1_impdp/*.dmp
export NLS_LANG=AMERICAN_AMERICA.ZHS16GBK

userid='/ as sysdba'
directory=schema_imp
dumpfile=expdp_schemas_%U.dmp
logfile=impdp_schemas.log
parallel=6
schemas=BBSP,BBSPODS,BCDPETL,DBMONOPR
content=all

nohup impdp parfile=impdp.par &"
T-0 检查error more impdp_schemas.log |grep -i error
T-0 grant GRANT SELECT ON "BBSP"."RGCT_DRAFT_LOG" TO "BCDPETL"
T-0 查出时点ID记录数最大的做保存 select max(l.id) from bbsp.rgct_draft_log l
T-0 用记录的最大id查出新增数据 select count(*) from bbsp.rgct_draft_log l where l.id>xxxxx; --替换为上一步查出的值
T-0 补录新增数据到bbsp.rgct_draft_log "insert into bbsp.rgct_draft_log select * from bbsp.rgct_draft_log@dblink_migup where id>xxxxxxxx ;
commit;"
T-0 "数据核对
并且点开lob字段查询是否有值" "select count(*) from bbsp.rgct_draft_log ;
select * from bbsp.rgct_draft_log ;"
T-0 清除impdp导到目标库sys用户下的job并复核 "select * from dba_jobs;
清除impdp导到sys用户下的job
DECLARE X NUMBER;
BEGIN
for cur in ( select * from dba_jobs j where j.log_user = 'SYS') loop
SYS.DBMS_iJOB.remove(job => cur.job );
COMMIT;
end loop ;
commit;
END;
/"
T-1 创建公共同义词 "SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/03_create_public_synonym.sql
SELECT
'create or replace public synonym '||synonym_name||' for '||table_owner||'.'||table_name||';'
FROM dba_synonyms@dblink_migup
WHERE owner = 'PUBLIC'
AND table_owner IN ( select username from sys.t_user_cfg t) ;
spool off;
在目标库执行03_create_public_synonym.sql"
T-0 生成用户审计信息 "SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/04_audit_users.sql
SELECT
'audit connect by '||user_name||';'
FROM dba_stmt_audit_opts@dblink_migup
WHERE audit_option = 'CREATE SESSION'
and user_name in ( select username from sys.t_user_cfg )
and user_name is not null;
SELECT
'audit '||audit_option||' by '||user_name||' by session whenever successful;'
FROM dba_stmt_audit_opts@dblink_migup
WHERE audit_option <> 'CREATE SESSION'
and user_name in ( select username from sys.t_user_cfg )
and user_name is not null;
SPOOL OFF;"
T-0 添加用户审计信息 执行04_audit_user.sql
T-0 给用户和角色授权 "1.执行附件脚本04_grant.sql
2.可以看到生成的脚本$HOME/grant_role_priv.sql,$HOME/grant_sys_priv.sql,$HOME/grant_sys_tab_priv.sql
3.检查没问题后执行上面生成的3个脚本"
T-0 目标库编译失效对象 sqlplus> @?/rdbms/admin/utlrp.sql;
T-0 对比源库和目标库确保无新增失效对象 "select owner ,object_name, status from dba_objects where status = 'INVALID'
minus
select owner ,object_name, status from dba_objects@dblink_migup where status = 'INVALID';"
T-0 检查表和索引的统计信息是否被锁住 "select owner,table_name,partition_name,t.stattype_locked from dba_tab_statistics t where t.stattype_locked is not null and t.owner in (select username from sys.t_user_cfg);
select owner,index_name,table_owner,table_name,partition_name,stattype_locked from dba_ind_statistics t where t.stattype_locked is not null and t.owner in (select username from sys.t_user_cfg);
如果有结果，说明有表或索引统计信息被锁定

执行下列拼出来的语句
select case when partition_name is null then 'exec dbms_stats.unlock_table_stats(ownname =>''' ||owner || ''',tabname =>''' || table_name || ''');'
else 'exec dbms_stats.unlock_table_stats(ownname =>''' ||owner || ''',tabname =>''' || table_name ||''',partname =>'''|| partition_name||''');' end
from dba_tab_statistics t
where t.stattype_locked is not null
and t.owner in (select username from sys.t_user_cfg);

select case when partition_name is null then 'exec dbms_stats.unlock_table_stats(ownname =>''' ||table_owner || ''',tabname =>''' || table_name || ''');'
else 'exec dbms_stats.unlock_table_stats(ownname =>''' ||table_owner || ''',tabname =>''' || table_name ||''',partname =>'''|| partition_name||''');' end
from dba_ind_statistics t where t.stattype_locked is not null and t.owner in (select username from sys.t_user_cfg);"
T-0 "复核以下项在源和目标环境中是否相等
" " 检查迁移用户系统权限
检查迁移用户角色权限
检查迁移用户对象权限
检查迁移用户审计
检查迁移用户对象同义词
检查迁移用户db_link
检查迁移用户context
检查迁移用户的对象总数
迁移用户的constraint
检查迁移用户的index(含索引字段)
检查迁移用户的表结构
迁移用户的trigger
检查迁移用户的表记录数
检查迁移用户的seq数及last_number"
T-0 核心表数据比对 "select count(*)登记中心数 from BBSP.rgct_bill_info
union
select count(*)登记中心历史数 from BBSP.rgct_bill_hist

T-0 打开归档 "shutdown immediate;
startup nomount;
alter database mount;
alter database archivelog;
alter database open;"
T-0 关闭undo和temp表空间的自动扩展 "select 'alter database datafile '''||file_name||''' autoextend off;' from dba_data_files where tablespace_name like 'UNDO%';
select 'alter database tempfile '''||file_name||''' autoextend off;' from dba_temp_files ;
执行产出脚本"
T-0 复核job参数 "检查Job参数设置,设置为0
SQL> show parameter job_queue_processes --0
SQL> show parameter aq_tm_processes --0"
force logging开启 alter database force logging;
T-0 gg混滚链路处理 "alter table BBSP.CODE logging;
alter table BBSP.PRODUCT logging;
"
T-0 gg链路处理 按gg回滚手册进行部署
T-0 配置监听 "【新库恢复监听】
lsnrctl stop db1
并修改监听端口为1528

cd $ORACLE_HOME/network/admin
vi listener.ora

lsnrctl start db1 --监听端口是1528"
T-0 DNS切换，清理缓存 找应用值班同事执行
T-0 验证连通性 sqlplus客户端验证
T-0 启动中间件和相关应用
T-0 通知开发、应用验证应用
T-0 打开Job参数 "SQL> alter system set job_queue_processes = 1000 scope = both;
SQL> alter system set aq_tm_processes=1 scope = both;;
复核：
SQL> show parameter job_queue_processes
SQL> show parameter aq_tm_processes
"
T-0 恢复cron文件 无
T-0 注册catalog "Rman注册：
步骤如下（主库）：
shell>rman target / catalog rman11g/rman11g@cat11g
RMAN> register database;
RMAN> report schema;
RMAN> CONFIGURE RETENTION POLICY TO RECOVERY WINDOW OF 90 DAYS;
RMAN> CONFIGURE ARCHIVELOG DELETION POLICY TO BACKED UP 1 TIMES TO 'SBT_TAPE';
RMAN>show all;"

2.1 set_job_nextdate.sql

SET termout on feedback on verify OFF echo on heading OFF trims ON escape ON pages 0 linesize 999 long 3000
set SERVEROUTPUT on size 200000
alter session set nls_date_format='yyyy/mm/dd hh24:mi:ss' ;

begin
for cur in ( select * from dba_jobs j where j.BROKEN = 'N' ) loop
sys.dbms_ijob.next_date(job => cur.job,next_date=>to_date('2999/9/9','yyyy/mm/dd')) ;
end loop ;
commit ;
end;
/

select count(1) from dba_jobs j where j.next_date < to_date('2999-09-09','yyyy-mm-dd') and j.broken = 'N';

2.2 04_grant.sql

SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/grant_role_priv.sql
SELECT 'spool $HOME/grant_role_priv.log' FROM dual;
SELECT 'grant ' || granted_role || ' to ' || grantee || CASE admin_option
WHEN 'YES' THEN
' with admin option;'
ELSE
';'
END
FROM dba_role_privs@dblink_migup
WHERE grantee IN
( select username from sys.t_user_cfg
union
select distinct granted_role from dba_role_privs@dblink_migup WHERE grantee IN (select username from sys.t_user_cfg)
and granted_role like 'R\_%' escape '\'
) ;
SELECT 'spool off;' FROM dual;
SPOOL OFF;
commit;
SET termout on echo on

--SET termout OFF feedback on verify OFF echo on heading OFF trims ON escape ON pages 0 linesize 999 long 3000

--@$HOME/grant_role_priv.sql
--commit;

SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/grant_sys_priv.sql
SELECT 'spool $HOME/grant_sys_priv.log' FROM dual;
SELECT
'grant '||privilege||' to '||grantee
||CASE admin_option WHEN 'YES' THEN ' with admin option;' ELSE ';' END
FROM dba_sys_privs@dblink_migup
WHERE grantee IN
( select username from sys.t_user_cfg
union
select distinct granted_role from dba_role_privs@dblink_migup WHERE grantee IN (select username from sys.t_user_cfg)
and granted_role like 'R\_%' escape '\'
) ;
SELECT
'grant '||privilege||' to '||grantee
||CASE admin_option WHEN 'YES' THEN ' with admin option;' ELSE ';' END
FROM dba_sys_privs@dblink_migup
WHERE grantee IN (SELECT role FROM dba_roles WHERE password_required <> 'NO' AND ROLE LIKE 'R\_%' ESCAPE '\');
SELECT 'spool off;' FROM dual;
SPOOL OFF;
commit;
SET termout on echo on

--SET termout OFF feedback on verify OFF echo on heading OFF trims ON escape ON pages 0 linesize 999 long 3000

--@$HOME/grant_sys_priv.sql
--commit;

SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
spool $HOME/grant_sys_tab_priv.sql
SELECT 'spool $HOME/grant_sys_tab_priv.log' FROM dual;

SELECT
'grant '||privilege||' on '
||CASE WHEN substr(table_name,1,2)='V$' THEN 'V_$'||substr(table_name,3,length(table_name-2)) ELSE table_name END
||' to '||grantee||CASE grantable WHEN 'YES' THEN ' with grant option;' ELSE ';' END
FROM dba_tab_privs@dblink_migup
WHERE grantee IN
( select username from sys.t_user_cfg
union
select distinct granted_role from dba_role_privs@dblink_migup WHERE grantee IN (select username from sys.t_user_cfg)
and granted_role like 'R\_%' escape '\'
)
AND owner = 'SYS'
and table_name NOT in('EXPDIR','V_$TEMP_HISTOGRAM','V_$STANDBY_APPLY_SNAPSHOT','GV_$TEMP_HISTOGRAM','GV_$STANDBY_APPLY_SNAPSHOT' ,'DBA_ADVISOR_DIRECTIVES','DBA_CACHEABLE_TABLES' , 'DBA_CACHEABLE_OBJECTS_BASE' , 'DBA_CACHEABLE_OBJECTS' ,'DBA_USERS2') -----opms Special
AND privilege NOT IN ('READ','WRITE');
SELECT
'grant '||privilege||' on directory '||table_name||' to '||grantee
||CASE grantable WHEN 'YES' THEN ' with grant option;' ELSE ';' END
FROM dba_tab_privs@dblink_migup
WHERE grantee IN
( select username from sys.t_user_cfg
union
select distinct granted_role from dba_role_privs@dblink_migup WHERE grantee IN (select username from sys.t_user_cfg)
and granted_role like 'R\_%' escape '\'
)
AND owner = 'SYS'
AND privilege IN ('READ', 'WRITE')
AND table_name not in ( 'EXPDIR' );
SELECT 'spool off;' FROM dual;
SPOOL OFF;
commit;
SET termout on echo on

--SET termout OFF feedback on verify OFF echo on heading OFF trims ON escape ON pages 0 linesize 999 long 3000

--@$HOME/grant_sys_tab_priv.sql
--commit;

2.3 check.sql

----check object -----------------
--CHECK OBJECT

1.准备工作 (11G RAC)
step 0:
alter user dbmgr identified by dbmgr1234;

alter system set local_listener = '(DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=10.197.1.203)(PORT=1528)))' sid='db1' scope = memory;

step 1:

drop public database link dblink_migup;

create public database link dblink_migup connect to dbmgr identified by dbmgr1234 using ' (DESCRIPTION = (ADDRESS = (PROTOCOL = TCP)(HOST =10.197.1.103)(PORT = 1528)) (CONNECT_DATA =(sid = db1)))';

step 2:

drop table sys.t_user_cfg;

create table sys.t_user_cfg as SELECT username, default_tablespace
FROM dba_users
WHERE username not in
('SYS', 'SYSTEM', 'PUBLIC', 'OUTLN', 'WMSYS', 'ORDSYS', 'MDSYS',
'CTXSYS', 'XDB', 'WKSYS', 'ANONYMOUS', 'ORDPLUGINS', 'EXFSYS',
'LBACSYS', 'DIP', 'LBACSYS', 'ORACLE_OCM', 'TSMSYS', 'ADAMS',
'TRACESVR', 'DBSNMP', 'PERFSTAT', 'DBSTATS', 'EXPFULL',
'MDDATA', 'MGMT_VIEW', 'OLAPSYS', 'SI_INFORMTN_SCHEMA', 'OWBSYS',
'SYSMAN', 'ORDDATA', 'OWBSYS_AUDIT', 'APEX_030200', 'APPQOSSYS',
'MDSYS', 'FLOWS_FILES', 'SPATIAL_WFS_ADMIN_USR','SQLTXPLAIN','DBMONOPR','SCOTT',
'SPATIAL_CSW_ADMIN_USR', 'APEX_PUBLIC_USER', 'XS$NULL','DMSYS','SQLTXADMIN','ZABBIX','DBMGR','OGG','OVSEE','FOGLIGHT','PAO');

set pagesize 0
select * from sys.t_user_cfg ;

2:验证dblink是否通

select * from dual@dblink_migup;

--有结果返回表示通的否则不通需处理" 未完成

3.开始使用连接PL/SQL developer 端连接数据库 RAC 11G
用户名:dbmgr/dbmgr1234
字符串:
10.197.1.205:1528/db

4.PL/SQL devoper 开一个SQL 窗口，执行以下语句，（重点关注 jobs dbmgr.save_ash('db') 在11g是否存在;）

--CHECK jobs
select log_user,schema_user,what,broken
from dba_jobs@dblink_migup
where schema_user in (select username from sys.t_user_cfg)
and last_date is not null
minus
select log_user,schema_user,what,broken
from dba_jobs
where schema_user in (select username from sys.t_user_cfg)
and last_date is not null;

--CHECK scheduler_jobs
select job_name from dba_scheduler_jobs@dblink_migup where owner !='SYS'
minus
select job_name from dba_scheduler_jobs where owner !='SYS';

--CHECK OBJECT
select owner,object_name,object_type,status
from dba_objects@dblink_migup
where owner in (select username from sys.t_user_cfg)
and status <> 'INVALID'
minus
select owner,object_name,object_type,status
from dba_objects
where owner in (select username from sys.t_user_cfg)
and status <> 'INVALID';

----check system priv-------------
select grantee,privilege from dba_sys_privs@dblink_migup
where grantee in ( select username from sys.t_user_cfg)
minus
select grantee,privilege from dba_sys_privs
where grantee in ( select username from sys.t_user_cfg) ;

----check role priv-------------
select grantee,granted_role,admin_option from dba_role_privs@dblink_migup
where grantee in ( select username from sys.t_user_cfg)
minus
select grantee,granted_role,admin_option from dba_role_privs
where grantee in ( select username from sys.t_user_cfg) ;

---check role----
select role from dba_roles@dblink_migup
minus
select role from dba_roles;

--CHECK SYS PRIVS
select grantee,privilege,admin_option
from dba_sys_privs@dblink_migup
where grantee in (select username from sys.t_user_cfg)
minus
select grantee,privilege,admin_option
from dba_sys_privs
where grantee in (select username from sys.t_user_cfg) ;

--CHECK ROLE PRIVS
select grantee,granted_role,admin_option
from dba_role_privs@dblink_migup
where grantee in (select username from sys.t_user_cfg)
minus
select grantee,granted_role,admin_option
from dba_role_privs
where grantee in (select username from sys.t_user_cfg) ;

----check user object priv-------------
select grantee,owner,table_name,grantor,privilege,grantable from dba_tab_privs@dblink_migup
where grantee in ( select username from sys.t_user_cfg)
minus
select grantee,owner,table_name,grantor,privilege,grantable from dba_tab_privs
where grantee in ( select username from sys.t_user_cfg) ;

-----check quota---------------
select username,max_bytes
from dba_ts_quotas@dblink_migup
where username in (select username from sys.t_user_cfg)
minus
select username,max_bytes
from dba_ts_quotas
where username in (select username from sys.t_user_cfg) ;
--check audit-------------
select user_name,audit_option,success,failure
from dba_stmt_audit_opts@dblink_migup
where user_name in ( select username from sys.t_user_cfg)
minus
select user_name,audit_option,success,failure
from dba_stmt_audit_opts
where user_name in ( select username from sys.t_user_cfg) ;

--check synonym no rows selected
select owner,synonym_name,table_owner,table_name
from dba_synonyms@dblink_migup
where table_owner in ( select username from sys.t_user_cfg)
minus
select owner,synonym_name,table_owner,table_name
from dba_synonyms
where table_owner in ( select username from sys.t_user_cfg) ;

--check db links no rows selected
select owner,db_link,username,host from dba_db_links@dblink_migup
where owner in ( select username from sys.t_user_cfg)
minus
select owner,db_link,username,host from dba_db_links
where owner in ( select username from sys.t_user_cfg) ;

--check db context no rows selected
select namespace,schema,package,type
from dba_context@dblink_migup
where schema in ( select username from sys.t_user_cfg)
minus
select namespace,schema,package,type
from dba_context
where schema in ( select username from sys.t_user_cfg) ;

--CHECK OBJECT COUNT
--select owner,object_type,status,count(*)
-- from dba_objects@dblink_migup
-- where owner in (select username from sys.t_user_cfg)

-- And (owner,object_name) Not In (Select owner,segment_name From dba_lobs@dblink_migup
-- Union All Select owner,index_name From dba_lobs@dblink_migup)
-- group by owner,object_type,status
--minus
--select owner,object_type,status,count(*)
-- from dba_objects
-- where owner in (select username from sys.t_user_cfg)
--- group by owner,object_type,status;

--check objects
--老库比新库，结果应为空
select owner,object_name,object_type
from dba_objects@dblink_migup
where owner in (select username from sys.t_user_cfg)
minus
select owner,object_name,object_type
from dba_objects
where owner in (select username from sys.t_user_cfg) ;

--新库比老库，结果显示出来的对象需要邓玉坤确认是否是新建对象
select * from (
select owner,object_name,object_type
from dba_objects
where owner in (select username from sys.t_user_cfg)
minus
select owner,object_name,object_type
from dba_objects@dblink_migup
where owner in (select username from sys.t_user_cfg) ) t where t.object_name not like 'SYS_IL000%' and t.object_name not like 'AQ$_%';

--check table no rows selected
select owner,table_name,partitioned,iot_type,temporary
from dba_tables@dblink_migup
where owner in ( select username from sys.t_user_cfg)
minus
select owner,table_name,partitioned,iot_type,temporary
from dba_tables
where owner in ( select username from sys.t_user_cfg) ;

--check constraint no rows selected
select owner, constraint_name, constraint_type, table_name, r_owner, status
from dba_constraints@dblink_migup
where owner in (select username from sys.t_user_cfg)

and constraint_name not like 'SYS%' and constraint_name not like 'BIN$%'
minus
select owner, constraint_name, constraint_type, table_name, r_owner, status
from dba_constraints
where owner in (select username from sys.t_user_cfg)

and constraint_name not like 'SYS%' and constraint_name not like 'BIN$%';

------check index---------------------
select index_owner,index_name,table_owner,table_name,column_name,column_position
from dba_ind_columns@dblink_migup
where table_owner in (select username from sys.t_user_cfg)

minus
select index_owner,index_name,table_owner,table_name,column_name,column_position
from dba_ind_columns
where table_owner in (select username from sys.t_user_cfg)
;

--check tab columns no rows selected
select owner,table_name,column_name,data_type,data_length,nullable,column_id from dba_tab_columns@dblink_migup
where owner in (select username from sys.t_user_cfg) and table_name not like 'AQ$%'
and table_name not like 'BIN$%'
minus
select owner,table_name,column_name,data_type,data_length,nullable,column_id from dba_tab_columns
where owner in (select username from sys.t_user_cfg) and table_name not like 'AQ$%'
;

--check comments no rows selected
select owner,table_name,column_name,comments from dba_col_comments@dblink_migup
where owner in (select username from sys.t_user_cfg)
and table_name not like 'BIN$%'
minus
select owner,table_name,column_name,comments from dba_col_comments
where owner in (select username from sys.t_user_cfg)
;

------check profile
select * from dba_profiles@dblink_migup
minus
select * from dba_profiles ;

------check mview
select owner,mview_name,container_name,query_len,rewrite_enabled,refresh_mode,refresh_method from dba_mviews@dblink_migup
where owner in (select username from sys.t_user_cfg)

minus
select owner,mview_name,container_name,query_len,rewrite_enabled,refresh_mode,refresh_method from dba_mviews
where owner in (select username from sys.t_user_cfg)
;

-----check user
select a.username,a.password apwd,b.password bpwd,a.profile apf,b.profile bpf,
a.account_status acs,b.account_status bcs
From
(
select du.username,su.password,du.profile,du.account_status
from dba_users@dblink_migup du,sys.user$ su
where username in (select username from sys.t_user_cfg) and du.username = su.name
) a
,
(
select du.username,su.password,du.profile,du.account_status
from dba_users du,sys.user$ su
where du.username in (select username from sys.t_user_cfg)
and du.username = su.name
) b
Where a.username = b.username And (a.password <> b.password Or a.profile <> b.profile Or a.account_status <> b.account_status);

--------check trigger--
select owner,trigger_name,trigger_type,table_owner,table_name,column_name,status
from dba_triggers@dblink_migup
where owner in (select username from sys.t_user_cfg)

minus
select owner,trigger_name,trigger_type,table_owner,table_name,column_name,status from dba_triggers
where owner in (select username from sys.t_user_cfg)
;

--check table owner trigger no rows selected

select owner,trigger_name,trigger_type,table_owner,table_name,column_name
from dba_triggers@dblink_migup where owner in (select username from sys.t_user_cfg)
minus
select owner,trigger_name,trigger_type,table_owner,table_name,column_name
from dba_triggers where owner in (select username from sys.t_user_cfg);

----------check sequence 结果为空
select ds.sequence_owner,
ds.sequence_name,
ds.min_value,
ds.max_value,
ds.increment_by,
ds.cycle_flag,
ds.order_flag,
ds.cache_size
from dba_sequences@dblink_migup ds
where sequence_owner in (select username from sys.t_user_cfg)
minus
select ds.sequence_owner,
ds.sequence_name,
ds.min_value,
ds.max_value,
ds.increment_by,
ds.cycle_flag,
ds.order_flag,
ds.cache_size
from dba_sequences ds
where sequence_owner in (select username from sys.t_user_cfg);

--比对sequence值，结果应为空，新库SEQUNCE值比老库大
select *
from (select sequence_owner,
sequence_name,
lastnum_10,
lastnum_11,
lastnum_11 - lastnum_10 delta_num
from (select ds.sequence_owner sequence_owner,
ds.sequence_name sequence_name,
ds.last_number lastnum_10,
dt.last_number lastnum_11
from dba_sequences dt, dba_sequences@dblink_migup ds
where dt.sequence_owner in
(select username from sys.t_user_cfg)
and dt.sequence_owner = ds.sequence_owner
and dt.sequence_name = ds.sequence_name))
where delta_num < 0;

5.删除db_link
drdrop public database link dblink_migup;op public database link dblink_migup;

###################################

3.OGG 安装

上传不同平台的goldengate软件 gg软件两个平台都要安装
"goldengate临时装到data卷上
1.cd /db/{ORACLE_SID}/data
mkdir goldendate
2.unzip goldengate软件
" 请注意 goldengate 软件区分os平台和oracle版本 ,另外对于RAC，goldengate软件安装目录需要设置为共享存储
"cd
vi .profile
##追加以下内容
GGS_HOME=/db/{ORACLE_SID}/data/goldengate
export GGS_HOME
LD_LIBRARY_PATH=$ORACLE_HOME/lib:$GGS_HOME:$LD_LIBRARY_PATH
export LD_LIBRARY_PATH
PATH=$GGS_HOME:$PATH
export PATH

hp-unix:step:
. ./profile_bcds

" "1、请注意平台不同，profile文件配置也不同
2、对于RAC，每个节点上的profile都需要配置；"
"1.检查:
select SUPPLEMENTAL_LOG_DATA_MIN from v$database;
2.如果是未有, 打开supplementary log:
set timing on
alter system checkpoint;
alter system checkpoint;
ALTER DATABASE ADD SUPPLEMENTAL LOG DATA;" "1.linux 结果如下：
SUPPLEME
--------
YES

2.hp unix 结果如下：
SQL> select SUPPLEMENTAL_LOG_DATA_MIN from v$database;

SUPPLEME
--------
YES

"
"create user ggmgr identified by ggmgr123 default tablespace users temporary tablespace temp;
grant create session,alter session to ggmgr;
grant alter system to ggmgr;
grant resource to ggmgr;
grant connect to ggmgr;
grant select any dictionary to ggmgr;
grant flashback any table to ggmgr;
grant select any table to ggmgr;
grant select any table to ggmgr;
grant insert any table to ggmgr;
grant update any table to ggmgr;
grant delete any table to ggmgr;
grant select on dba_clusters to ggmgr;
grant execute on dbms_flashback to ggmgr;
grant create table to ggmgr;
grant create sequence to ggmgr;
grant alter any table to ggmgr;
grant dba to ggmgr;
grant lock any table to ggmgr;" 1.linux/hp-unix done
"cd $GGS_HOME
mkdir direnv"
./ggsci
"create subdirs
exit"
"cd $GGS_HOME
./ggsci
GGSCI> EDIT PARAMS ./GLOBALS
--输入如下内容
CHECKPOINTTABLE ggmgr.gg_checkpoint_tab
syslog none" "1.linux/hp-unix done
当使用GG版本11.x时，需要在GG manager参数中配置syslog none，使gg日志不写入操作系统的syslog。"
"GGSCI> DBLOGIN USERID ggmgr, PASSWORD ggmgr123
GGSCI> ADD CHECKPOINTTABLE ggmgr.gg_checkpoint_tab
GGSCI>exit" 1.linux/hp-unix done
"cd $GGS_HOME/direnv
1.vi db.oby
SETENV (ORACLE_SID = ""bcds"")

2.vi user.oby
userid ggmgr, password ggmgr123
" "1.linux/hp-unix done
ORACLE_SID = ""bcds"" 注意大小写要和环境变量保持一致"
"cd $GGS_HOME/dirprm
vi mgr.prm
port 7809
dynamicportlist 7809-7819
autostart er *
autorestart er *, retries 24, waitminutes 3, resetminutes 1440
lagreportminutes 10
laginfoseconds 30
LAGCRITICALMINUTES 45
purgeoldextracts ./dirdat/*, usecheckpoints, minkeephours 168
" "
1.linux/hp-unix done
删除PURGEDDLHISTORY MINKEEPDAYS 7, MAXKEEPDAYS 10 "
"cd $GGS_HOME
./ggsci
start mgr"
info mgr
GGSCI>dblogin userid ggmgr, password ggmgr12

2.迁移当天设置回滚环境

创建2条链路，链路一(step 1)是非CLOB 表，   链路2 （step 2）是clob 表

准备参数文件 准备e进程、d进程、r进程的参数文件
为ggmgr用户临时授权 "grant insert any table to ggmgr;
grant delete any table to ggmgr;
grant update any table to ggmgr;
"
设置参数 alter system set ENABLE_GOLDENGATE_REPLICATION=true scope=both;
创建gg_send表 "CREATE TABLE ggmgr.GG_SEND(
TESTER VARCHAR2(30) NOT NULL,
SOURCE_DATABASE VARCHAR2(10) NOT NULL,
TARGET_DATABASE VARCHAR2(10) NOT NULL,
TEST_TIME DATE DEFAULT SYSDATE,
Constraint PK_GG_SEND primary KEY (TESTER, SOURCE_DATABASE, TARGET_DATABASE, TEST_TIME) );

COMMENT ON COLUMN ggmgr.GG_SEND.TESTER IS '测试人员UM账号';
COMMENT ON COLUMN ggmgr.GG_SEND.SOURCE_DATABASE IS '源数据库SID';
COMMENT ON COLUMN ggmgr.GG_SEND.TARGET_DATABASE IS '目标数据库SID';
COMMENT ON COLUMN ggmgr.GG_SEND.TEST_TIME IS '测试时间';
COMMENT ON TABLE ggmgr.GG_SEND is 'GoldenGate接收端连通性测试表';

2. step 2 table 2:
CREATE TABLE ggmgr.GG_SEND_2(
TESTER VARCHAR2(30) NOT NULL,
SOURCE_DATABASE VARCHAR2(10) NOT NULL,
TARGET_DATABASE VARCHAR2(10) NOT NULL,
TEST_TIME DATE DEFAULT SYSDATE,
Constraint PK_GG_SEND_2 primary KEY (TESTER, SOURCE_DATABASE, TARGET_DATABASE, TEST_TIME) );

COMMENT ON COLUMN ggmgr.GG_SEND_2.TESTER IS '测试人员UM账号';
COMMENT ON COLUMN ggmgr.GG_SEND_2.SOURCE_DATABASE IS '源数据库SID';
COMMENT ON COLUMN ggmgr.GG_SEND_2.TARGET_DATABASE IS '目标数据库SID';
COMMENT ON COLUMN ggmgr.GG_SEND_2.TEST_TIME IS '测试时间';
COMMENT ON TABLE ggmgr.GG_SEND_2 is 'GoldenGate接收端连通性测试表';

"
添加trandata "cd $GGS_HOME
./ggsci
dblogin userid ggmgr password ggmgr123
delete trandata BBSP.*
delete trandata BBSPODS.*
delete trandata BCDPETL.*

add trandata BBSP.*
add trandata BBSPODS.*
add trandata BCDPETL.*
"

--WARNING OGG-00869  All viable columns will be used to represent the key, but may not guarantee uniqueness.  KEYCOLS may be used to define the key.  这个报错可以忽略

对超过32列无主键的表添加trandata "
执行/tmp/dba/add_trandata_long32.sql
--报错WARNING OGG-01387需要执行"

1.部署pkg（排除了clob表）
2.执行查询
SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000

spool /tmp/dba/add_trandata_long32.sql
select trim(sql_add_supp) from TABLE(
select cast(gg_supplemental_32.sql_supplemental_32('AFA')
as gg_supplemental_32_table_type )
from dual
);
select trim(sql_add_supp) from TABLE(
select cast(gg_supplemental_32.sql_supplemental_32('AUTEK')
as gg_supplemental_32_table_type )
from dual
);

检查trandata是否添加成功 "执行下面sql，返回空为成功。( --如果不为 空 ，add trandata BBSPODS.*  ，会报错error no data found retrieving导致漏掉一些表，所以需要手工add trandata BBSPODS.table_name) 
select 'add trandata '||owner||'.'|| table_name from dba_tables where owner in ('BBSP',' BBSPODS','BCDPETL') and table_name not in (select distinct table_name from dba_log_group_columns where owner in ('BBSP',' BBSPODS','BCDPETL'));

--排除掉一些临时备份表或者CLOB表或者名称为小写表，剩余的表就需要手工add   trandata BBSPODS.table_name
--clob 表会自动创建唯一索引的

--add trandata BBSPODS.*  会漏掉一些表，所以需要手工add trandata BBSPODS.table_name
--同时在GGSCI 重新运行返回的结果SQL，

"
上传参数文件 "cd $GGS_HOME/dirprm

cd $GGS_HOME/dirprm
step 1:
将e_rollbk.prm,d_rollbk.prm上传到此目录下
step 2:
将e_rollbk_2.prm,d_rollbk_2.prm上传到此目录下
chown -R opDB:dba /crdb/DB/data/goldengate/dirprm

chown -R opbcds:dba /ogg/bcds/dirprm

"
创建GG进程 "cd $GGS_HOME  (注意  OGG 的进程名总数不能超过8个 )
./ggsci

cd $GGS_HOME
./ggsci
dblogin userid ggmgr password ggmgr123

step 1:
add extract e_rollbk, tranlog, begin now
add exttrail ./dirdat/e1, extract e_rollbk, megabytes 50

add extract d_rollbk, exttrailsource ./dirdat/e1, params dirprm/d_rollbk.prm
add rmttrail ./dirdat/dr, extract d_rollbk, megabytes 50

step 2:
add extract e_roll_2, tranlog, begin now
add exttrail ./dirdat/e2, extract e_roll_2, megabytes 50

add extract d_roll_2, exttrailsource ./dirdat/e2, params dirprm/d_roll_2.prm
add rmttrail ./dirdat/d2, extract d_roll_2, megabytes 50

"
修改回滚链路r进程参数文件 "cd $GGS_HOME/dirprm

cd $GGS_HOME/dirprm
step 1:
上传r_rollbk.prm
step 2:
上传r_rollbk_2.prm

chown -R opUSER:dba /crdb/DB/oradata/goldengate/dirprm

chown -R opbcds:dba /oggfs/ogg/dirprm "
创建GG进程 "cd $GGS_HOME
./ggsci
dblogin userid ggmgr password ggmgr123

step 1:
add replicat r_rollbk, exttrail ./dirdat/dr, params dirprm/r_rollbk.prm , CHECKPOINTTABLE ggmgr.gg_checkpoint_tab

step 2:
add replicat r_roll_2, exttrail ./dirdat/d2, params dirprm/r_roll_2.prm , CHECKPOINTTABLE ggmgr.gg_checkpoint_tab

"
创建目标端连通性测试表

step 1:
CREATE TABLE ggmgr.GG_RECEIVE_R_ROLLBK(
TESTER VARCHAR2(30) NOT NULL,
SOURCE_DATABASE VARCHAR2(10) NOT NULL,
TARGET_DATABASE VARCHAR2(10) NOT NULL,
TEST_TIME DATE DEFAULT SYSDATE,
Constraint PK_GG_RECEIVE_R_ROLLBK primary KEY (TESTER, SOURCE_DATABASE, TARGET_DATABASE, TEST_TIME) );

COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK.TESTER IS '测试人员UM账号';
COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK.SOURCE_DATABASE IS '源数据库SID';
COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK.TARGET_DATABASE IS '目标数据库SID';
COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK.TEST_TIME IS '测试时间';
COMMENT ON TABLE ggmgr.GG_RECEIVE_R_ROLLBK is 'GoldenGate接收端连通性测试表';

step 2:
CREATE TABLE ggmgr.GG_RECEIVE_R_ROLLBK_2(
TESTER VARCHAR2(30) NOT NULL,
SOURCE_DATABASE VARCHAR2(10) NOT NULL,
TARGET_DATABASE VARCHAR2(10) NOT NULL,
TEST_TIME DATE DEFAULT SYSDATE,
Constraint PK_GG_RECEIVE_R_ROLLBK_2 primary KEY (TESTER, SOURCE_DATABASE, TARGET_DATABASE, TEST_TIME) );

COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK_2.TESTER IS '测试人员UM账号';
COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK_2.SOURCE_DATABASE IS '源数据库SID';
COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK_2.TARGET_DATABASE IS '目标数据库SID';
COMMENT ON COLUMN ggmgr.GG_RECEIVE_R_ROLLBK_2.TEST_TIME IS '测试时间';
COMMENT ON TABLE ggmgr.GG_RECEIVE_R_ROLLBK_2 is 'GoldenGate接收端连通性测试表';

';

"
禁用级联删除和触发器 "SELECT 'alter table '||owner||'.'||table_name||' disable constraint '||constraint_name||';'
FROM dba_constraints
WHERE owner in ('BBSP',' BBSPODS','BCDPETL')
AND constraint_type = 'R'
ORDER BY constraint_type desc, owner, table_name, constraint_name;

select 'alter trigger '||owner||'.'||trigger_name||' disable ;'
from all_triggers
where owner in ('BBSP',' BBSPODS','BCDPETL')
order by owner,trigger_name;

复核：
select owner,table_name,constraint_name,status from dba_constraints where owner in ('BBSP',' BBSPODS','BCDPETL') and constraint_type = 'R' and status <> 'DISABLED'; --无结果产出

select owner,trigger_name,table_owner,table_name,status from dba_triggers where owner in ('BBSP',' BBSPODS','BCDPETL') and status <> 'DISABLED'; --无结果产出
"
启动extract和dump进程

step 1:
start e_rollbk
start d_rollbk
step 2:
start e_roll_2
start d_roll_2
info all
<进程都是RUNNING状态>

启动同步到unix的回滚R进程

cd $GGS_HOME

./ggsci
step 1:
start r_rollbk

step 2:
start r_roll_2

插入测试信息 "sqlplus '/as sysdba'

step 1
sqlplus '/as sysdba'
insert into ggmgr.GG_SEND values('测试','source_sid','target_sid',SYSDATE);
commit;

step 2:
insert into ggmgr.GG_SEND_2 values('测试','source_sid','target_sid',SYSDATE);
commit;

检测测试信息是否同步过来 "sqlplus '/as sysdba'

step 1:
sqlplus '/as sysdba'
select * from ggmgr.GG_RECEIVE_R_ROLLBK ;

step 2:
sqlplus '/as sysdba'
select * from ggmgr.GG_RECEIVE_R_ROLLBK_2 ;

for clob table:
select count(*) from f_cus_reprintinfo;

for no-clob table:

select count(*) from t_ibps_paymentbook

"

3.3 OGG PKG

查看32列以上 没有主键 的非clob表   (排除了clob表，如果有 clob 表，那么可能需要另外单独为clob表 单独创建一个链路,解释2  CLOB 表会自动 创建一个唯一性的索引。可以正常使用ogg 同步，所以下面pkg 不检测clob 表) ：

1.部署pkg
2.执行查询
SET termout OFF feedback OFF verify OFF echo OFF heading OFF trims ON escape ON pages 0 linesize 999 long 3000
select trim(sql_add_supp) from TABLE(
select cast(gg_supplemental_32.sql_supplemental_32('BBSP')
as gg_supplemental_32_table_type )
from dual
);

Pkg info：

create or replace type gg_supplemental_32_row_type as object
(
owner varchar2(32),
table_name varchar2(32),
seq# number,
sql_add_supp varchar2(2048)
);
/
create or replace type gg_supplemental_32_table_type as table of gg_supplemental_32_row_type;
/
create or replace package gg_supplemental_32 AUTHID CURRENT_USER is
function sql_supplemental_32(i_owner varchar2)
return gg_supplemental_32_table_type;
end gg_supplemental_32;
/

create or replace package body gg_supplemental_32 as
function sql_supplemental_32(i_owner varchar2)
return gg_supplemental_32_table_type as
z_data gg_supplemental_32_table_type := gg_supplemental_32_table_type();
n number;
cursor c is
select dt.owner || '.' || dt.table_name obj,
dt.column_name,
dt.table_name,
dt.owner,
Row_Number() over(partition by dt.owner || '.' || dt.table_name order by dt.column_name) rn,
do.object_id
from dba_tab_columns dt, dba_objects do
where dt.owner = i_owner
and dt.data_type not in
('CLOB', 'BLOB', 'NCLOB', 'LONG', 'LONG RAW', 'ANYDATA',
'ALERT_TYPE', 'XMLTYPE', 'WM_PERIOD', 'SQL_PLAN_TABLE_TYPE',
'BFILE')
and dt.data_type not like '%$%'
and dt.owner = do.owner
and dt.table_name = do.object_name

and do.OBJECT_TYPE='TABLE'
and dt.table_name in
(SELECT distinct (table_name) obj
FROM dba_tab_columns
WHERE owner = i_owner
AND column_id > 32
AND table_name in
(SELECT table_name
FROM dba_tables
WHERE owner = i_owner
MINUS (SELECT obj1.name
FROM SYS.user$ user1,
SYS.user$ user2,
SYS.cdef$ cdef,
SYS.con$ con1,
SYS.con$ con2,
SYS.obj$ obj1,
SYS.obj$ obj2
WHERE user1.name = i_owner
AND cdef.type# = 2                                            
AND con2.owner# = user2.user#(+)
AND cdef.robj# = obj2.obj#(+)
AND cdef.rcon# = con2.con#(+)
AND obj1.owner# = user1.user#
AND cdef.con# = con1.con#
AND cdef.obj# = obj1.obj#
UNION
SELECT idx.table_name
FROM dba_indexes idx
WHERE idx.owner = i_owner
AND idx.uniqueness = 'UNIQUE')));
obj_prev varchar2(200);
obj_flag number;
sql_add_supp varchar2(4000);
sql_post varchar2(2000);
n_g number;
pre_n_g number;
z_owner varchar2(32);
z_table_name varchar2(32);
begin
sql_post := ') always;';
obj_flag := 0;
for x in c loop
n_g := trunc((x.rn - 1) / 32) + 1;
n := mod(x.rn, 32);
z_owner := x.owner;
z_table_name := x.table_name;

if obj_prev is null then
sql_add_supp := 'ALTER TABLE ' || x.obj ||
' ADD SUPPLEMENTAL LOG GROUP GGS_' ||
substr(x.table_name, 1, 16) || '_' || x.object_id || '_' || n_g || '(' ||
x.column_name;
obj_prev := x.obj;
pre_n_g := n_g;
elsif (obj_prev = x.obj) then
if (n_g <> pre_n_g) then
-- print out sql statement
sql_add_supp := sql_add_supp || sql_post;
z_data.extend;
z_data(z_data.count) := gg_supplemental_32_row_type(z_owner,
z_table_name,
n_g,
sql_add_supp);
-- reset
pre_n_g := n_g;
sql_add_supp := 'ALTER TABLE ' || x.obj ||
' ADD SUPPLEMENTAL LOG GROUP GGS_' ||
substr(x.table_name, 1, 16) || '_' || x.object_id || '_' || n_g || '(' ||
x.column_name;
else
sql_add_supp := sql_add_supp || ' ,' || x.column_name;
end if;
else
-- print out sql statement
sql_add_supp := sql_add_supp || sql_post;
z_data.extend;
z_data(z_data.count) := gg_supplemental_32_row_type(z_owner,
z_table_name,
n_g,
sql_add_supp);
-- reset
pre_n_g := n_g;
obj_prev := x.obj;
sql_add_supp := 'ALTER TABLE ' || x.obj ||
' ADD SUPPLEMENTAL LOG GROUP GGS_' ||
substr(x.table_name, 1, 16) || '_' || x.object_id || '_' || n_g || '(' ||
x.column_name;
end if;
end loop;
-- print last sql if exist --
sql_add_supp := sql_add_supp || sql_post;
z_data.extend;
z_data(z_data.count) := gg_supplemental_32_row_type(z_owner,
z_table_name,
n_g,
sql_add_supp);

return z_data;
end sql_supplemental_32;
end;
/

###解释说明：

解释1 非clob 表带有唯一主键的表。

SELECT obj1.name
FROM SYS.user$ user1,
SYS.user$ user2,
SYS.cdef$ cdef,
SYS.con$ con1,
SYS.con$ con2,
SYS.obj$ obj1,
SYS.obj$ obj2
WHERE user1.name = i_owner
AND cdef.type# = 2                                            
AND con2.owner# = user2.user#(+)
AND cdef.robj# = obj2.obj#(+)
AND cdef.rcon# = con2.con#(+)
AND obj1.owner# = user1.user#
AND cdef.con# = con1.con#
AND cdef.obj# = obj1.obj#

###以下是查  clob 表和非clob 表都带有唯一主键的表

SELECT idx.table_name
FROM dba_indexes idx
WHERE idx.owner = i_owner
AND idx.uniqueness = 'UNIQUE'

解释2  CLOB 表会自动 创建一个唯一性的索引。所以OGG

You are not allowed to operate directly on system-defined LOB index.

Whenever you create a LOB column the database automatically creates two segments (LOB Data Segment, LOB index Segment).  You cannot drop/alter any of these segments separately, the only way to do so is to drop the LOB column itself and consequently the two associated segments will be dropped.

TEST CASE
=========
SQL> create user user1 identified by user1;

User created.

SQL> grant dba to user1;

Grant succeeded.

SQL> connect user1/user1
Connected.
SQL> create table LOB_TAB(id number, name CLOB);

Table created.

SQL> insert into LOB_TAB values (1,'Ahmed');

1 row created.

SQL> commit;

Commit complete.

SQL> select index_name, index_type from dba_indexes where table_name like 
'LOB_TAB' and owner like 'USER1';

INDEX_NAME INDEX_TYPE
------------------------------ ---------------------------
SYS_IL0000051347C00002$$ LOB

SQL> drop index SYS_IL0000051347C00002$$;
drop index SYS_IL0000051347C00002$$
*
ERROR at line 1:
ORA-22864: cannot ALTER or DROP LOB indexes

SQL> alter table LOB_TAB drop column name;

Table altered.

SQL>

You can use the DBA_LOBS view to query for all the LOB objects related to a 
certain user and then drop their columns one by one:

SQL> col table_name format a20
SQL> col column_name format a20
SQL> select table_name, column_name from dba_lobs where owner like 'USER1';

TABLE_NAME COLUMN_NAME
-------------------- --------------------
LOB_TAB NAME