Control File Question

[jhmartin]

If you don't have a valid control file after the failure, then you won't really know if you've properly recovered the database up to the point of failure. The chances are that after applying all your archivelogs that the database will be correct, but you can't be 100% sure.

Also, depending on your backup methods, RMAN may be storing backup information into the control files.

[davey23uk]

you can be 100 sure, you keep applying the archive logs until you have no more and then the redo logs and then you are done.

You are very sure it is finished when you have nothing left to apply

[waitecj]

You may or may not find this useful. I asked a very similar question a few months back:

http://www.dbasupport.com/forums/sho...ighlight=inode

[dbbyleo]

we don't use RMAN. And if we did in this case, then we would simply option to store catalog in a database, not in the controlfile.

If I missed any datafiles when I re-created the control file, I simply need to check dba_data_files for any missing files to make.

I dont see anything else I could miss in the controlfile that I couldn't easily address.

Other than that, then it's a matter of checking the data itself. And we tested this in the following manner:

1.) Created a table with one date column.
2.) Run a procedure that repeatedly inserted sysdate into that table.
3.) "Pulled the plug" - removed all the datafiles and controlfiles, aborted the database, and noted system time
4.) Recovered the database having only the following files:
a. Data files from previous backup
b. Archived redo logs since previous backup
c. Online redo logs
d. Trace of controfile (this is really needed, but makes making the create-controlfile script easier.
5.) Checked the table and found an record with system date up to the time we aborted the database.