手工 XTTS CPT 迁移-跨平台迁移数据库-阿里云开发者社区

开发者社区> 数据库> 正文
登录阅读全文

手工 XTTS CPT 迁移-跨平台迁移数据库

简介: -- xtts 跨平台迁移数据库 -- 步骤: 1、初始化参数设置; 2、将源端数据文件传输到目标系统; 3、转换数据文件为目标系统的字节序; 4、在源端创建增量备份,并传输到目标端; 5、在目标端恢复增量备份; 6、重复多次操作4和5步骤; 7、将源端数据库表空间设置为 READ ONLY 模式; 8、最后一次执行4和5步骤; 9、在源端导出元数据,并在目标端导入; 10、将目标端的数据库表空间设置为 READ WRITE; 11、数据验证。
-- xtts 跨平台迁移数据库

-- 步骤:
1、初始化参数设置;
2、将源端数据文件传输到目标系统;
3、转换数据文件为目标系统的字节序;
4、在源端创建增量备份,并传输到目标端;
5、在目标端恢复增量备份;
6、重复多次操作4和5步骤;
7、将源端数据库表空间设置为 READ ONLY 模式;
8、最后一次执行4和5步骤;
9、在源端导出元数据,并在目标端导入;
10、将目标端的数据库表空间设置为 READ WRITE;
11、数据验证。

-- 建议xtts目录和/rman 备份的路径目录都使用nfs共享

alter system set deferred_segment_creation=false sid='*' scope=spfile;

-- ###########################增量临时文件在asm######### xtt.properties ##########################
tablespaces=SPROD_TBS,IDX_TBS
platformid=13
dfcopydir=/xttrman/full -- 本地全量备份数据的路径
backupformat=/xttrman/incr -- 本地增量备份数据的路径
stageondest=/xttrman/full -- 目标端所需要恢复数据的路径,后续生成的增量也要拷贝到这里
storageondest=+DATA/prod/datafile -- 目标端数据文件路径 -c 的时候放的位置
backupondest=+DATA/prod/datafile -- 目标端增量文件的临时路径,做完增量后会自动删除,如果放asm需要设置asm_home和asm_sid
asm_home=/oracle/app/grid/product/11.2.0/grid
asm_sid=+ASM
parallel=2
rollparallel=2
getfileparallel=2
-- #############################建议使用这种方式,增量临时文件在文件系统###### xtt.properties #######################
tablespaces=SPROD_TBS,IDX_TBS
platformid=13
dfcopydir=/xttrman/full
backupformat=/xttrman/incr
stageondest=/xttrman/full
storageondest=+DATA/prod/datafile
backupondest=/xttrman/rec
#asm_home=/oracle/app/grid/product/11.2.0/grid
#asm_sid=+ASM
parallel=2
rollparallel=2
getfileparallel=2
-- ###########################################################################################

-- 源端全量备份脚本
cat full_backup.sh 
export TMPDIR=/oracle/soft/xttconvert
perl xttdriver.pl -p
-- 源端执行全量备份
sh full_backup.sh

-- 生成文件
$ cat xttplan.txt 
SPROD_TBS::::967892
6
IDX_TBS::::967892

$ cat rmanconvert.cmd
host 'echo ts::SPROD_TBS'; 
convert from platform 'Linux x86 64-bit' 
datafile 
'/xttrman/full/SPROD_TBS_5.tf' 
,'/xttrman/full/SPROD_TBS_6.tf' 
format '+DATA/prod/datafile/%N_%f.dbf' 
parallelism 2; 
host 'echo ts::IDX_TBS'; 
convert from platform 'Linux x86 64-bit' 
datafile 
'/xttrman/full/IDX_TBS_7.tf' 
format '+DATA/prod/datafile/%N_%f.dbf' 
parallelism 2;

-- 目标端,执行恢复转换
vi convert.sh
export XTTDEBUG=1
export TMPDIR=/oracle/soft/xttconvert
perl xttdriver.pl -c

sh convert.sh

vi do_incr.sh
export TMPDIR=/oracle/soft/xttconvert
perl xttdriver.pl -i
-- 源端增量备份
sh do_incr.sh

$ cat xttplan.txt.new 
SPROD_TBS::::968168
6
IDX_TBS::::968196

[oracle@localhost xttconvert]$ cat tsbkupmap.txt
IDX_TBS::7:::1=07u31cir_1_1
SPROD_TBS::5,6:::1=06u31ciq_1_1

-- 注意每次做完增量 需要将增量的数据文件拷贝到全量备份的目录下,恢复完删除/xttrman/incr/*
cp /xttrman/incr/-* /xttrman/full/

-- 目标端增量恢复
vi restore_incr.sh
export TMPDIR=/oracle/soft/xttconvert
perl xttdriver.pl -r

-- 注意如果增量恢复文件放在asm可能会出现以下错误,可以忽略,这个是xtts无法自动删除asm的临时文件
Can't locate strict.pm in @INC (@INC contains: /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0 /oracle/app/grid/product/11.2.0/grid/lib /oracle/app/grid/product/11.2.0/grid/lib/asmcmd /oracle/app/grid/product/11.2.0/grid/rdbms/lib/asmcmd /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl .) at /oracle/app/grid/product/11.2.0/grid/bin/asmcmdcore line 143.
BEGIN failed--compilation aborted at /oracle/app/grid/product/11.2.0/grid/bin/asmcmdcore line 143.
Can't locate strict.pm in @INC (@INC contains: /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0 /oracle/app/grid/product/11.2.0/grid/lib /oracle/app/grid/product/11.2.0/grid/lib/asmcmd /oracle/app/grid/product/11.2.0/grid/rdbms/lib/asmcmd /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0/x86_64-linux-thread-multi /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl/5.10.0 /oracle/app/grid/product/11.2.0/grid/perl/lib/site_perl .) at /oracle/app/grid/product/11.2.0/grid/bin/asmcmdcore line 143.
BEGIN failed--compilation aborted at /oracle/app/grid/product/11.2.0/grid/bin/asmcmdcore line 143.

-- 下次增量要记录当前恢复的scn位置,产生最新 xttplan.txt 文件,要不然下次增量还会在全量进行备份
perl xttdriver.pl -s

-- 后续增量循环以下三步骤
sh do_incr.sh
sh restore_incr.sh
perl xttdriver.pl -s

-- 最后一次增量,表空间设置为只读模式
SQL> alter tablespace SPROD_TBS read only;

Tablespace altered.

SQL> alter tablespace IDX_TBS read only;

Tablespace altered.

-- 最后一次恢复
sh do_incr.sh 
sh restore_incr.sh


----------------------------------------------------------------------------------------
-- 传输表空间
传输表空间的步骤可以和最后一次增量备份同时执行,可开并行
方法1:
-- 目标端执行,得到impdp脚本
$ORACLE_HOME/perl/bin/perl xttdriver.pl –e
impdp system/oracle directory=ddmp logfile=impdp_xtts_20180527.lognetwork_link=TEST \
transport_full_check=no \
transport_tablespaces=TEST,TEST_IND \
transport_datafiles='/data3/oradata/targetdb/TEST_5.dbf','/data3/oradata/targetdb/TEST_IND_6.dbf'

方法2:
-- 手工导出导入表空间
源端
expdp system/oracle directory=sdmp dumpfile=xtts.dmplogfile=expdp_xtts_20180527.log transport_tablespaces=TEST,TEST_INDtransport_full_check=no;

-- 复制导出的文件到目标端
目标端
impdp system/oracle directory=ddmp dumpfile=xtts.dmplogfile=impdp_xtts_20180527.logtransport_datafiles='/app/oradata/targetdb/TEST_5.dbf,/app/oradata/targetdb/TEST_IND_6.dbf'

-- 验证,传输完成后应该能够看到所有表空间,并且能够查询其中的数据
set pagesize 2000
set linesize 200
col file_name for a40
col OBJECT_NAME for a30
select TABLESPACE_NAME from dba_tablespaces;
----------------------------------------------------------------------------------------


-- 导入元数据

-- 源端检查表空间的自包含性
SQL> EXECUTE DBMS_TTS.TRANSPORT_SET_CHECK('TEST02', TRUE);
SQL> SELECT * FROM TRANSPORT_SET_VIOLATIONS; 
no rows selected

/*
$ expdp system/oracle directory=source_dbdata dumpfile=test02.dmp transport_tablespaces=test02 transport_full_check=y
$ impdp test02/test02 DUMPFILE=test02.dmp DIRECTORY=source_dbdata TRANSPORT_DATAFILES=/home/oracle/xtts/target_data/TEST02_7.dbf
*/

-- 方式一
-- 建议不要导入统计信息,目标库直接收集
create or replace directory xtts_dir as '/oracle/soft/'
create database link ttslink connect to system identified by "oracle" USING 'sprod';

ASMCMD> ls -al
WARNING:option 'a' is deprecated for 'ls'
please use 'absolutepath'

Type Redund Striped Time Sys Name
DATAFILE UNPROT COARSE JUN 01 21:00:00 Y +DATA/PROD/DATAFILE/idx_tbs_7.dbf => IDX_TBS.274.1009834557
DATAFILE UNPROT COARSE JUN 01 21:00:00 Y +DATA/PROD/DATAFILE/sprod_tbs_5.dbf => SPROD_TBS.273.1009834553
DATAFILE UNPROT COARSE JUN 01 21:00:00 Y +DATA/PROD/DATAFILE/sprod_tbs_6.dbf => SPROD_TBS.275.1009834553
DATAFILE UNPROT COARSE JUN 01 21:00:00 Y none => SYSAUX.257.1009808403
DATAFILE UNPROT COARSE JUN 01 21:00:00 Y none => SYSTEM.256.1009808403
DATAFILE UNPROT COARSE JUN 01 21:00:00 Y none => UNDOTBS1.258.1009808403
DATAFILE UNPROT COARSE JUN 01 21:00:00 Y none => USERS.259.1009808403
N idx_tbs_7.dbf => +DATA/PROD/DATAFILE/IDX_TBS.274.1009834557
N sprod_tbs_5.dbf => +DATA/PROD/DATAFILE/SPROD_TBS.273.1009834553
N sprod_tbs_6.dbf => +DATA/PROD/DATAFILE/SPROD_TBS.275.1009834553
ASMCMD>

impdp \’/ as sysdba\‘ directory=xtts_dir logfile=tts_imp.log network_link=ttslink transport_full_check=no transport_tablespaces=SPROD_TBS,IDX_TBS exclude=STATISTICS transport_datafiles='+DATA/PROD/DATAFILE/IDX_TBS.275.1009832163','+DATA/PROD/DATAFILE/SPROD_TBS.274.1009832159','+DATA/PROD/DATAFILE/SPROD_TBS.273.1009832159'

ls -al|grep -v *.zip|grep -v .sh|awk -F' ' '{print $9}'|xargs rm -rf

-- 方式二
-- 源端导出,如果不对数据字典做任何修改操作,可提前导出
expdp system/oracle DIRECTORY=xtts_dir LOGFILE=dp_fullexp_meta.log DUMPFILE=dp_full.dmp FULL=y CONTENT=METADATA_ONLY

-- 复制导出的文件到目标端
-- 目标端导入
impdp system/oracle DIRECTORY=xtts_dir LOGFILE=dp_fullexp_meta.log DUMPFILE=dp_full.dmp FULL=y CONTENT=METADATA_ONLY


-- 自定义 Role 处理
select GRANTED_ROLE from dba_role_privs whereGRANTEE='XTTS';
create role R_INIT_USER;
select 'grant '||PRIVILEGE||' to '||GRANTEE||';' fromdba_sys_privs where grantee='R_INIT_USER';

-- 自定义 Profile 处理
set long 1000000
SET PAGESIZE 3000
set lines 200
SET HEADING OFF
SET VERIFY OFF
SET FEEDBACK OFF
set echo on
set timing off
set wrap On
SET LONGCHUNKSIZE 400
spool create_profile.sql
with profile as(
select distinct profile from dba_profiles)
select dbms_metadata.get_ddl('PROFILE',PROFILE) fromprofile;

-- 临时表处理
set long 1000000
SET PAGESIZE 3000
set lines 200
SET HEADING OFF
SET VERIFY OFF
SET FEEDBACK OFF
set echo on
set timing off
set wrap On
SET LONGCHUNKSIZE 400
select dbms_metadata.get_ddl('TABLE',TABLE_NAME,OWNER)
from DBA_TABLES
where TEMPORARY='Y' and owner in ();


-- 解释:
tablespaces=    -- 需要迁移的表空间
platformid=   -- 源 OS 平台 ID
dfcopydir=   -- 源数据库备份文件存放目录
backupformat=    -- 源数据库增备文件存放目录
stageondest=   -- 目标据库备份文件存放目录
storageondest=   -- 目标据库正式文件存放目录
backupondest=   -- 目标据库增备文件存放目录
parallel=   -- 备份,转化的并行度
rollparallel=   -- 增备的并行度
getfileparallel=  -- 使用 dbms_file_transfer 方式的并行度











-- xtts 跨平台迁移数据库 2
1、源和目标库创建数据库对应的目录,目标库创建公共DBLink使用 system 用户
# source db
create or replace directory ora10g_dbf_dir as '+DATA/ora10g/datafile';
create or replace directory xttsfiles as '/home/oracle/xttsfiles';
grant read,write on directory ... to public;
# target db
create or replace directory ora12c_dbf_dir as '+DATA/ora12c/datafile';
create or replace directory xttsfiles as '/home/oracle/xttsfiles';
grant read,write on directory ... to public;
create database link ORA10G_LINK connect to system identified by system using 'ORA10G';

2、解压 XTTS
unzip -d /home/oracle/xtts rman_xttconvert_1.4.zip

3、编辑 xtts.properties 配置文件
vi xtts.properties
tablespace=AP_DATA,AP_IDX
# source db param
platform=2
srcdir=ora10g_dbf_dir
srclink=ora10g_link
dfcopydir=/home/oracle/xttsfiles
backupformat=/home/oracle/xttsfiles
# target db param
dstdir=ora12c_dbf_dir
stageondest=/home/oracle/xttsfiles
storageondest=+DATA
backupondest=+FRA
asm_home=/u01/app/12.1.0/grid
asm_sid=+ASM
parallel=4
rollparallel=2

4、执行脚本
-- 初始提取
Generate initial datafile image copies:
$ ./xttdriver.pl -S
Copy scripts & files to destination platform:
$ scp ... ...
Pull initial datafile image copies to destination:
$ ./xttdriver.pl -G
Conversion handled implicitly!
-- 增量同步
Generate INCREMENTAL LEVEL 1 backup sets:
$ ./xttdriver.pl -i
Copy scripts & files to destination platform:
$ scp ... ...
Capture SCN of last synchronization:
$ ./xttdriver.pl -s
Use INCREMENTAL LEVEL 1 backup sets to recover datafile image copies to current SCN:
$ ./xttdriver.pl -r
-- 当源库与目标库期望接近时,DBA决定完成迁移,则将所有需要XTTS迁移的对象设置为只读模式,
alter tablespace ... read only;
-- 直到最后一个1级增量备份被捕获兵传输到目标库,然后应用于0级备份,在最后一次恢复。
-- 最终同步
Generate INCREMENTAL LEVEL 1 backup sets:
$ ./xttdriver.pl -i
Copy scripts & files to destination platform:
$ scp ... ...
Capture SCN of last synchronization:
$ ./xttdriver.pl -s
Use INCREMENTAL LEVEL 1 backup sets to recover datafile image copies to final SCN:
$ ./xttdriver.pl -r
Import metadata into destination database:
$ ./xttdriver.pl -e




-- CPT(Cross-Platform Transport) 跨平台迁移数据库
# source 12.1.0.1 
Backup READ ONLY tablespace as backup sets:
rman target /
backup for transport format '+DATA' datapump format '/home/oracle/cpt/dp_file.dmp' tablespace tpch_data,tpch_idx;
Copy datafile backup sets to target database 12.1.0.2
Use DBMS_FILE_TRANSFER or other OS tools transport to destination OS.
# target 12.1.0.2 non-CDB
rman target /
restore foreign tablespace tpch_data,tpch_idx format '+DATA' from backupset '+RECO' dump file from backupset '/home/oracle/cpt/dp_file.dmp';

版权声明:本文内容由阿里云实名注册用户自发贡献,版权归原作者所有,阿里云开发者社区不拥有其著作权,亦不承担相应法律责任。具体规则请查看《阿里云开发者社区用户服务协议》和《阿里云开发者社区知识产权保护指引》。如果您发现本社区中有涉嫌抄袭的内容,填写侵权投诉表单进行举报,一经查实,本社区将立刻删除涉嫌侵权内容。

分享: