要升级replicat端的原因为:目标端OGG软件版本与源端OGG软件版本不同,在实际生产应用中,经常发现replicat端事务丢失的情况,所以,需要将目标端的OGG软件升级为与源端OGG相同软件版本。
1、升级前环境情况
源端OGG版本11.2.1.0.1
目标端OGG版本11.1.1.1.2
升级前,为了解决源端、目标端OGG版本不一致不能正常同步的问题,在源端抽取Tail file格式时,加了format release 11.1的格式转换命令,在extract与data pump进程中均配置,其配置方式如下:
EXTTRAIL ./dirdat/tr, format release 11.1 |
2、升级目标
将目标端OGG版本从11.1.0.2升级到与源端一致的11.2.0.1版本
3、升级前准备
3.1 停止源端的extract和datapump进程
GGSCI>stop exttr
GGSCI>stop dpetr
3.2 停止目标端的replicat和mgr进程
GGSCI>stop reptr
GGSCI>stop mgr
3.3 去除extract和datapump进程中的format release参数
因为在extract与datadump进程中都配置有formatrelease 11.1关键字,在目标端升级后,该配置需要取消,但是,取消该配置,并不只是从配置文件中删除这么简单的事,还需要对进程做ETROLLOVER操作,否则进程启动时,会报如下错误:
ERROR OGG-01416 File ./dirdat/tr000008, with format RELEASE 10.4/11.1, does not match
current format specification of RELEASE 11.2.Modify the parameter file to specify format RELEASE 10.4/11.1
or issueETROLLOVER prior to restart.
操作方法:在OGG中以edit paramsextract_name和edit params datadump_name的方式进入进行修改(具体操作:略)。
3.4 对extract与datadump进程做ETROLLOVER操作
因为extract与datadump都修改了format release信息,所以,都需要做ETROLLOVER操作
GGSCI>alter extract exttrETROLLOVER
GGSCI>alter extract dpetrETROLLOVER
3.5 修改源端datapump进程的EXTSEQNO号和EXTRBA号
因为源端的extract-exttr进程执行ETROLLOVER操作后,进程的extseqno和extrba号被重置至下一个extseqno号的第0号extrba,而datapump并不知道extract进程发生了这个变化,仍然守望在extract ETROLLOVER前的extseqno和extrba号上,永远也等不到这个extseqno上有新的RBA变化,造成就无法将extract新抽取到的数据,传递到目标端去。
所以,在extract进程ETROLLOVER后,需要使用“Alterextract group_name EXTSEQNO X, EXTRBA 0”的命令,重置checkpoint位置,datadump进程才能正常的将数据继续传递到远端。
3.5.1 记录下exttr进程ETROLLOVER后的extseqno和extrba号
GGSCI> info exttr, detail
GGSCI (server1) 140> info exttr, detail EXTRACT EXTTR Initialized 2015-01-30 13:02 Status STOPPED Checkpoint Lag 00:00:00 (updated 00:00:27 ago) Log Read Checkpoint Oracle Redo Logs 2015-01-30 13:18:27 Seqno 365, RBA 17822208 SCN 0.17726669 (17726669) Target Extract Trails: Remote Trail Name Seqno RBA Max MB ./dirdat/tr 4 0 100 Extract Source Begin End …… |
3.5.2 修改datadump进程的extseqno和extrba号
GGSCI>Alter extract dpetr EXTSEQNO 4,EXTRBA 0
3.6 查出datadump进程ETROLLOVER后的Target ExtractTrails的extseqno和extrba号
GGSCI (server1) 160> info dpetr, detail EXTRACT DPETR Initialized 2015-01-30 13:03 Status STOPPED Checkpoint Lag 00:00:00 (updated 00:02:10 ago) Log Read Checkpoint File ./dirdat/tr000003 2015-01-30 13:18:18.000000 RBA 48953966 Target Extract Trails: Remote Trail Name Seqno RBA Max MB ./dirdat/tr 4 0 100
…… |
3.7 备份目标端的OGG目录
#cp –ra /u01/ogg /u01/ogg_backup
3.8 记录下目标端replicat的checkpoint信息
GGSCI (server2) 1> info reptr, showch REPLICAT REPTR Last Started 2015-01-30 13:03 Status STOPPED Checkpoint Lag 00:00:00 (updated 00:10:36 ago) Log Read Checkpoint File ./dirdat/tr000003 2015-01-30 13:18:16.296427 RBA 48953996 Current Checkpoint Detail: Read Checkpoint #1 GGS Log Trail Startup Checkpoint (starting position in the data source): Sequence #: 0 RBA: 0 Timestamp: Not Available Extract Trail: ./dirdat/tr Current Checkpoint (position of last record read in the data source): Sequence #: 3 RBA: 48953996 Timestamp: 2015-01-30 13:18:16.296427 Extract Trail: ./dirdat/tr …… |
Current Checkpoint
为 Sequence #: 3 RBA: 48953996
4、OGG软件升级
4.1 将新版OGG 11.2.1.0.1软件复制到原旧版OGG目录下
$ cp ogg112101_fbo_ggs_Linux_x64_ora11g_64bit.zip/u01/ogg
4.2 删除OGG目录下的fbo_ggs_Linux_x64_ora11g_64bit.tar文件
fbo_ggs_Linux_x64_ora11g_64bit.tar文件是旧版OGG安装时解压出来的tar文件,新版OGG软件unzip解压时,也会生成此文件,当然也可以使用unzip解压时的覆盖替代手工删除
$rm fbo_ggs_Linux_x64_ora11g_64bit.tar
4.3 在原版OGG目录中安装(解压)新版OGG软件
$unzip ogg112101_fbo_ggs_Linux_x64_ora11g_64bit.zip
$tar xvf fbo_ggs_Linux_x64_ora11g_64bit.tar
4.4 验证升级成功情况
$cd $OGG
[[email protected] u01]$ cd $OGG
[[email protected] ogg]$ ./ggsci
Oracle GoldenGate Command Interpreter for Oracle Version 11.2.1.0.1 OGGCORE_11.2.1.0.1_PLATFORMS_120423.0230_FBO Linux, x64, 64bit (optimized), Oracle 11g on Apr 23 2012 08:32:14 Copyright (C) 1995, 2012, Oracle and/or its affiliates. All rights reserved. GGSCI (server2) 1> |
从这里看到,OGG软件版本已经升级至11.2.1.0.1
5、验证升级后replicat的checkpoint是否为一致
5.1 查看升级后replicat的checkpoint是否与升级前一致
GGSCI (server2) 46> info reptr, detail REPLICAT REPTR Last Started 2015-01-30 13:03 Status STOPPED Checkpoint Lag 00:00:00 (updated 00:12:58 ago) Log Read Checkpoint File ./dirdat/tr000003 2015-01-30 13:18:16.296427 RBA 48953996 Extract Source Begin End ./dirdat/tr000003 * Initialized * 2015-01-30 13:18 ./dirdat/tr000000 * Initialized * First Record Current directory /u01/ogg Report file /u01/ogg/dirrpt/REPTR.rpt Parameter file /u01/ogg/dirprm/reptr.prm Checkpoint file /u01/ogg/dirchk/REPTR.cpr Checkpoint table GOLDENGATE.CHECKPOINT_REPTR_01 Process file /u01/ogg/dirpcs/REPTR.pcr Stdout file /u01/ogg/dirout/REPTR.out Error log /u01/ogg/ggserr.log |
Current Checkpoint 跟升级前一样,仍为Sequence #: 3 RBA: 48953996,进一步证明升级成功。
6、验证升级后可正常同步数据场景准备
6.1 先记录下源端、目标端数据量为一致
(生产库环境此步可以跳过,因为生产库的表受业务的操作数据不断发生变化,本案例为创建一张专用用于测试升级的表进行测试)
源端 |
目标端 |
SQL> select count(*) from goldengate.ogg_upg; COUNT(*) ---------- 2150000 |
SQL> select count(*) from goldengate.ogg_upg; COUNT(*) ---------- 2150000 |
源端和目标端数据量一致。
6.2 在目标端replicat进程启动前在源端删除部分数据
SQL> delete goldengate.ogg_upg where rownum <1000001; 1000000 rows deleted. SQL> commit; SQL> select count(*) from goldengate.ogg_upg; COUNT(*) ---------- 1150000 |
7、重建目标端replicat进程
重建replicat进程的原因:ogg_11.1.1.1.2每条replicat进程只有CHECKPOINT一张表,而ogg_11.2.x.x.1每条replicat进程有CHECKPOINT,CHECKPOINT_LOX两张表,所以,如果在升级后,直接启动replicat进程,是无法启动的,会报如下错误:
ERROR OGG-00665 OCI Error describe for query (status = 942-ORA-00942: table or view does not exist), SQL<SELECT a.current_dir, a.seqno, a.rba, a.audit_ts, a.log_csn, a.log_xid, a.log_cmplt_csn, a.log_cmplt_xids, b.log_cmplt_xids FROM GOLDENGATE.CHECKPOINT_REPTR_01 a LEFT JOIN GOLDENGATE.CHECKPOINT_REPTR_01_lox b ON a.group_name = b.group_name AND a.group_key = b.group_key AND a.log_cmplt_csn = b.log_cmplt_csn WHERE a.group_name = ‘REPTR‘ AND a.group_key = 2810015614>. 2015-01-28 05:12:59 ERROR OGG-01668 PROCESS ABENDING. |
提示表不存在,这张表是指:GOLDENGATE.CHECKPOINT_REPTR_01_lox表(CHECKPOINT_LOX表),重建replicat的最终目的是让在重建进程时,自动将两张checkpoint表都自动创建起来。
7.1 删除replicat进程与checkpoint表
GGSCI>dblogin userid goldengate,password goldengate
GGSCI>delete replicat reptr
GGSCI>delete checkpointtable GOLDENGATE.CHECKPOINT_REPTR_01
7.2 重新建立replicat进程
GGSCI>dbloginuserid goldengate, password goldengate
GGSCI>add
checkpointtable goldengate.checkpoint_reptr_01
GGSCI>add replicat
reptr, exttrail ./dirdat/tr,checkpointtablegoldengate.checkpoint_reptr_01
到这一步,replicat启动后,还是无法继续应用源端投递过来的trail的,因为源端的datapump进程做过ETROLLOVER了,所以需要手工修改replicat的sequence#和RBA号到与datadump进程的target extract trails中的seqno和RBA一致。
7.3 修改目标端replicat的extseqno和extrba号
根据上面一步,查出的源端的datadump进程的remotetrail file的seqno号和RBA,决定replicat要更改的extseqno和extrba号
GGSCI (server2) 37> Alter replicat reptrEXTSEQNO 4, EXTRBA 0
7.4 查看新建立的replicat进程的SEQ#和RBA号
GGSCI(server2) 27> info reptr, showch
GGSCI (server2) 3> info reptr, showch REPLICAT REPTR Initialized 2015-01-30 13:39 Status STOPPED Checkpoint Lag 00:00:00 (updated 00:00:07 ago) Log Read Checkpoint File ./dirdat/tr000004 First Record RBA 0 Current Checkpoint Detail: Read Checkpoint #1 GGS Log Trail Startup Checkpoint (starting position in the data source): Sequence #: 4 RBA: 0 Timestamp: Not Available Extract Trail: ./dirdat/tr Current Checkpoint (position of last record read in the data source): Sequence #: 4 RBA: 0 Timestamp: Not Available Extract Trail: ./dirdat/tr …… |
Start checkpoint 的sequence#和RBA号已经通过手工,都定位到了源端ETROLLOVER后的状态
Current checkpoint的sequence#和RBA号已经通过手工,也都定位到了源端ETROLLOVER后的状态
7.5 查看checkpoint表的数据内容
SQL>select * from goldengate.checkpoint_reptr_01; No rows selected SQL>select |
这时,两个checkpoint表中的数据还是空的,待replicat启动运行后,进行状态信息就会写入到checkpoint表中
8、启动源端和目标端进程
8.1 启动目标端的replicat和mgr进程
GGSCI>start mgr
GGSCI>start reptr
8.2 启动源端的datapump进程
GGSCI>start dpetr
9、验证是否能继续同步数据
此步最重要的,是要验证在升级时间时,源端所做的操作能否同步到目标库中来。本案例在升级过程中,对goldengate.ogg_upg表删除了1000000行数据。
9.1 查看目标端goldengate.ogg_upg表的数据变化
源端 |
目标端 |
SQL> select count(*) from goldengate.ogg_upg; COUNT(*) ---------- 2050000 |
SQL> select count(*) from goldengate.ogg_upg; COUNT(*) ---------- 2050000 |
9.2 目标端查看checkpoint表内check状态
select * from goldengate.checkpoint_reptr_01 ------------------------------------------------- REPTR 2149948420 4 19280017 2015-01-30 13:55:59.368501 2015/1/30 13:36:43 2015/1/30 14:01:01 /u01/ogg 17929533 7.16.20512 17929533 7.16.20512 1 |
从上面两步看来,升级后replicat正常的,接着升级前的状态在同步数据。
本文作者:黎俊杰(网名:踩点),从事”系统架构、操作系统、存储设备、数据库、中间件、应用程序“六个层面系统性的性能优化工作
欢迎加入 系统性能优化专业群,共同探讨性能优化技术。群号:258187244