I am planning to migrate the characterset of my Oracle EBS 22.214.171.124 instance's database, which is in 10.2.0.5. I downloaded the DMU 1.0.2 and created the repository.
I have started the database scan, as per the v$session_longops I find the total_works column has value 15210 and the SOFAR column has the value around 2722 and it is getting incremented slowly. The total size of our database is 700GB.
Could anyone let me know, how long will it take to scan the database as I find the timeremaing column shows 31592 (almost 9 hrs)which is also increasing slowly.
Is this normal or do I need to apply any fix to speedup the scan.
I am doing this in my test server with 8 CPU's and no user is connected to that.
In Scan Wizard screen I find the Progress bar in "Calculating the Tablesplit Threshold" is still showing 1% progress.
This looks like an issue with the "Calculating the Tablesplit Threshold" step of the scan. This step should be short as it is only there to calculate the optimal scan method. Please, open an SR so that we can look at your configuration in detail.
I found that the initialization parameter in my DB was not set properly. I have changed the following initialization parameter which is a dynamic parameter and the table split step completed in just 1 minute.
optimizer_features_enable = 10.2.0.5
My database was 10.2.0.5 and the value for optimizer_features_enable was set to 9.2.0
Becasue of this parameter the explain plan went for nested loop query rather than going for Hash join.
This caused the performance issue and costed me 8 hrs to calculate the table split threashold.
Now after changing the value the DMU conversion is really fast.