How to replicate missing ddl / dml to target after intial loading

Question

Hi GG Experts,
I m new to GG, my test setup is like Oracle->Oracle (same version, 10.2.0.4) Replication. I did initial loading between source & target. There were some problem with my REPLICAT Process was getting abended, hence when i fired any dml/ddl are not transferred to target. Then, I put HANDLECOLLISIONS, APPLYNOOPUPDATES in REPLICAT Process, after which my replication got started but from that moment.
Now my question, how should I replicate the missing ddl/dml to target db which I fired on source db after initial loading and before my REPLICAT process successfully started.
is there any special consideration do I need to take for ddl replication.
Your help would be highly appreciated.
Regards,
Manish

Answer 1

Hi,
As both the source and target tables are out of sync currently, I would suggest you to re-synchronize the target using below procedure and restart the
online replication back again.
The export could be performed directly on the production system by using the FLASHBACK_SCN option. Then the FLASHBACK_SCN used for the export would then be the CSN value used for the Replicat. Note that you have to take the entire export using the same value for FLASHBACK_SCN for your entire export, even if you use multiple export files (e.g. you run multiple sessions in parallel, or in the case of data pump export, you use Oracle's parallelism).
Below are sample steps that you could use to instantiate an Oracle target system:
a.Switch the database to archivelog mode:
SQL> shutdown immediate
SQL> startup mount
SQL> alter database archivelog;
SQL> alter database open;
b. Enable minimal supplemental logging:
SQL> alter database add supplemental log data;
c. Prepare the database to support ddl replication
--Turn off recyclebin for the database . . .
SQL> alter system set recyclebin=off scope=both;
d.Create schema for ddl support replication & create a OGG user and provide necesaary user privileges
GGSCI> edit params ./GLOBAL
GGSCHEMA ogg
e.Run scripts for creating all necessary objects for support ddl replication:
SQL> @$ogg/marker_setup.sql
SQL> @$ogg/ddl_setup.sql
SQL> @$ogg/role_setup.sql
SQL> grant GGS_GGSUSER_ROLE to ogg;
SQL> @$ogg/ddl_enable.sql
GGSCI> dblogin userid ogg password ogg@321!
GGSCI> add trandata schema1.*
f. Create the extract group on the source side:
GGSCI>ADD EXTRACT EXT1, TRANLOG, BEGIN NOW, THREADS 2
GGSCI>ADD EXTTRAIL ./dirdat/P1, EXTRACT EXT1, MEGABYTES 100
GGSCI> edit params EXT1
EXTRACT EXT1
SETENV (ORACLE_HOME = "/oracle/orabin/product/10.2.4")
SETENV (ORACLE_SID = "prd1")
SETENV (NLS_LANG = "AMERICAN_AMERICA.WE8ISO8859P1")
USERID ogg@prd1, PASSWORD ogg@321!
TRANLOGOPTIONS ASMUSER sys@+ASM1, ASMPASSWORD asm321
RMTHOST dr, MGRPORT 7810
RMTTRAIL ./dirdat/P1
DISCARDFILE ./dirrpt/EXT1.DSC, PURGE, MEGABYTES 100
DDL INCLUDE MAPPED OBJNAME "schema1.*"
DDLOPTIONS ADDTRANDATA RETRYOP RETRYDELAY 60 MAXRETRIES 10, REPORT
TABLE schema1.*;
g.Create replicat group on target:
GGSCI> dblogin userid ogg password ogg@321!
GGSCI> add checkpointtable ogg.checkpoint
GGSCI>ADD REPLICAT REP1, RMTTRAIL ./dirdat/P1, checkpointtable ogg.checkpoint
GGSCI> edit params REP1
REPLICAT REP1
SETENV (ORACLE_HOME = “/oracle/orabin/product/10.2.4”)
SETENV (ORACLE_SID = “dr”)
SETENV (NLS_LANG = “AMERICAN_AMERICA.WE8ISO8859P1”)
ASSUMETARGETDEFS
USERID ogg@DR, PASSWORD ogg@321!
DISCARDFILE ./dirrpt/REP1.DSC, append, megabytes 100
DDL INCLUDE MAPPED OBJNAME "schema1.*"
MAP schema1.*, TARGET schema1.*;
h.Create a database directory:
SQLPLUS> create directory dumpdir as '<some directory>' ;
i.Get the current SCN on the source database:
SQLPLUS> select current_scn from v$database ;
28318029
j. Run the export using the flashback SCN you obtained in the previous step. The following example shows running the expdp utility at a Degree Of Parallelism (DOP) of 4. If you have sufficient system resources (CPU,memory and IO) then running at a higher DOP will decrease the amount of time it takes to take the export (up to 4x for a DOP of 4). Note that expdp uses Oracle Database parallel execution settings (e.g.parallel_max_servers) which have to be set appropriately in order to take advantage of parallelism. Other processes running in parallel may be competing for those resources. See the Oracle Documentation for more details.
expdp directory=dumpdir full=y parallel=4 dumpfile=ora102_%u.dmp flashback_scn=28318029
Username: sys as sysdba
Password:
Note: The export log needs to be checked for any errors.
h.Start an import using impdp to the target database when step 7 is complete.
i. GGSCI> start replicat <rename>, aftercsn <value returned from earlier step>
By following above procedure the resynchronisation of source and target can be done and the online change synchronuzation can be started without any data integrity issues. Hope this information helps.
Thanks & Regards
SK