OracleGoldenGate Data Synchronization initialization Best Practices (DataPump)

Use the infoextractpumpreplicat and detail commands to verify whether the trail files and checkpoints of each process in the synchronization link are consistent. If they are inconsistent, use the following statements to connect them, as shown in

Run the info extract/pump/replicat and detail commands to check whether the trail files and checkpoints of each process in the synchronization link are consistent. If they are inconsistent, use the following statement to connect them, as shown in figure

Full-link initialization standardization process:

1) Verify synchronization link connectivity

Run the info extract/pump/replicat and detail commands to check whether the trail files and checkpoints of each process in the synchronization link are consistent. If they are inconsistent, use the following statement to connect them, if necessary, you can perform DML tests before initialization starts.

Alter extract/pump/repilcat extseqno, extrba

2) Check the gv $ transaction View to check whether there are long-running transactions. If there are long-running transactions, check the cause, wait for the transaction to be completed, or kill the transaction.

3) add additional logs for the synchronization table to the source.

If no transaction cannot be killed in step 1, execute the transaction at the source end (this step is optional)

Alter extract tranlog threads 2 begin now

4) After performing the preceding steps,

Start the extract and pump Processes on the source.

Stop the replicat process on the target end

5) after completing the preceding steps, start the export operation. First, retrieve the current SCN value and record it as SCN_INIT when the source database does not have a running transaction:

Select current_scnfrom v $ database;

Or

Select dbms_flashback.get_system_change_number current_scn from dual;

6) use the expdp tool on the source to specify the flashback_scn option to export the initialization data of the synchronization link.

Expdp exp_user/exp_passwd schemas = hx_zsj directory = dp_dir dumpfile = hx_zsj.dmp logfile = hx_zsj_2013 ***. log exclude = grant, statistics, trigger flashback_scn = SCN_INIT

7) use the impdp tool to import data to the target end and specify the table_exist_action = truncate option.

Impdp imp_user/imp_passwd schemas = hx_zsj directory = dp_dir dumpfile = hx_zsj.dmp logfile = hx_zsj_2013 ***. log table_exist_action = truncate

Primary data synchronization considerations:

To ensure successful initialization, you should disable triggers, foreign key constraints, and jobs with DML operations on the target database before initialization. These factors may cause data inconsistency between the two ends after initialization.

Because the golden three environment has cross-flow cascade synchronization, the data link is complex. When initializing data synchronization, try to choose to close the data O & M portal in the business prohibited range (usually at night, ensure that the data at the synchronization source is not allowed.

The master data synchronization part of the core collection and management system. During initialization, we should try to ensure full separation of work on each link. For example, after the initialization of the production data such as taxpayers and declarations to the master database is complete, then initialize the synchronization of the outgoing distribution of the master database.

During initialization, try to avoid importing data to a database and exporting data from the database to ensure sufficient undo resources are available.

Related reading:

Management OGG-HA in RAC Environment

Configure OGG synchronization in RAC Environment

GoldenGate unidirectional table DML Synchronization

Oracle GoldenGate series: restoration principles of Extract processes

Oracle GoldenGate installation Configuration

OGG-01004 OGG-1296 error for Oracle goldengate