Recently i did some test case on my desktop regarding Goldengate backout plan
change details:
1. Databases are replicated using Goldengate
2. Create the some tables/indexes both source/target
3. Add the column to existing tables both source/target
4. Configure these changes into Goldengate (UNI directional - new tables/ existing column changes)
Backout plan is
1. Drop the tables/indexes both source/target
2. Drop the column which was recently added both source/target
3. Remove these changes from Goldengate
Complete backout plan steps
1. Take the invalid objects list from both source/target
2. Make sure there is no active session connected the tables which is going to add the column
3. Make sure there is no replication delay between source and target databases.
4. Take the GG definition files/parameter files backup on source and target databases
5. Bring down the GoldenGate on both source and target
6. Execute the DB changes on both source and target
7. Make sure there is no invalid objects on databases
8. Add the new tables detail on *.prm files /extract files on source; add the new tables detail on target *.prm files also.
9. Generate the GG definition files using *.prm files on source database
10. Copy the GG definition (*.def) file from source to target
11. Start the GG on both side.
I have done the DB backout and GG configuration changes successfully. But while start the GG extract processes (step 11), it was failed…
I see the GGERR.log file and found the below information alert..
2012-07-27 18:40:35 ERROR OGG-01038 Oracle GoldenGate Capture for Oracle, extract_1.prm: Cannot fetch required data from table AM.XXXXXXXX due to missing key columns.
2012-07-27 18:40:35 ERROR OGG-01668 Oracle GoldenGate Capture for Oracle, extract_1.prm: PROCESS ABENDING.
I have dropped some columns from AM.XXXXXXXX as part of backout plan.
I have tried so many times to start the GG.. I have read some forum and suggested to check the table structure, primary key / Unique key on both source/target databases, drop/recreate the supplemental log and I have tried, doesn’t help.
After 2 hours attempt, I have noticed while start the GG and its processes on first time, its replicated the few records from source to target databases. After that only GG is not able to extract/replicate the AM.XXXXXX table from archive log files.
I have confirmed the DML statements on AM.XXXXXXXX only caused this issue.
So I have skipped the particular archive log file extract/replication and started the GG finally ..
ggsci> alter extract extract_1, extseqno 7855, extAMa 0
7854 -> having mismatch data archive log files. So we have started the extract on 7855 archive.
Root cause:
Example :
1. Stopped the Goldengate on both source and target
2. I am going to drop the 3 columns on a table and its having totally 10 columns.
3. Before I am dropped these 3 columns, meanwhile few DML I didon that table and it’s not replicated to target database due to GG shutdown.
4. I have dropped these column from that table and started the GG
5. GG extract and replicate the data from archive log files for other tables..
6. While extract/replicate the our table data, GG is not able to extract/replicate due to column mismatch
( In archive log files, that tables having 10 columns record data, while extract/replicate having 7 columns, so GG was failed)
7. I have skipped that archive file from GG extract and then started the GG successfully.
8. Using Diff script, we can synch up the data between source and target databases.
I Hope this article helped you. Suggestions are welcome.
Thanks !!!