I am using Oracle 10g on Linux

We have a live 2 node RAC instance and we do have a DR instance located at a different place . Now the DR instance is a single instance database and the backups of RAC instance are shipped to the DR node through snapshots ( NetApp) . The backup would consist of redo logs , archived logs , datafiles , parameter files , control files and the password files.

Now when a media failure happens at live site I need to get the DR database running but I dont know how to convert the RAC database into a single instance database?

Since we would be having a list of redo logs , undo tablespace for the other instance and these might be required when we recover the database in the DR site but since it is a single instance database I can use redo logs , undo tablespaces of only one instance ? Am I correct ? Let me know how to tackle this issue ?

>> Now when a media failure happens at live site I need to get the DR database running

How would we know? How did you design it?

>> I dont know how to convert the RAC database into a single instance database?

Not needed.

Most RAC shops use Data Guard or Streams for replication to the DR site.

- With Data Guard, you can have the DR site open in a few minutes
- With streams, you can have the DR site oprn for read-onlt traffic anytime.

Without that, you would need to restore and roll-forward from whatever remote redo log you have collected.

For complete details, see the Replication book:

http://www.rampant-books.com/book_2003_2_replication.htm

Let me know if this makes sense ?

The Live Site is configured with ASM and the ASM volumes are shipped to DR site and when disaster occurs we use oracleasm to scan the disks .So directory structure remains the same with ASM configured in the DR site too.

Archived logs of both the instances are backed up.

So when I have all the backups ( including redo logs and undo tablespace of the second instance ) in the DR site , changing the init.ora such that the second instance parameters are removed and removing cluster_databases=true parameter too, will it not do the trick ?

I tried it and it worked but I was doubting the consistency , now my question is when the recovery happens will it apply the redo of both the threads and also the two instances in the Live database would have been pointing to two different undo tablespaces ?
When I set the undo_tablespace parameter in the DR site , what should I set , I did set it one of the two undo tablespaces and it did work but I dont know what I did is correct ?

What you did is correct. Oracle can understand that it needs two redo threads to recover, even though it is now a single instance database. All you need to do is change the init parameters as you did and recover.