I just did the following steps to recreate control file for migrating 10g DB from 32bit to 64 bit machine. I think this should work for your scenario too. 1) startup nomount; 2) create controlfile reuse ... noresetlogs ... (from backup controlfile to trace) 3) alter database open; 4) Add tempfile to temp tablespace manually; I think no recover command is needed if you have a current backup controlfile to trace. From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Roger Xu Sent: Monday, November 23, 2009 12:22 PM To: hrishy Cc: oracle-l@xxxxxxxxxxxxx Subject: Re: RMAN Restore Scenario: all copies of control files are lost First of all, thank you for your replies (Kumar,Martin,Jared and Hrishy). Would the following avoid resetlogs since everything else is assumed to be fine? 1) startup nomount; 2) create controlfile reuse ... noresetlogs ... (from backup controlfile to trace) 3) recover database; 4) alter system archvie log all; 5) alter database open; I am interest to see what kind of error produced on #3 since there is nothing to be recoverd? (I wish I have a sandbox to play with instead of bothering the list.) TIA, Roger On Mon, Nov 23, 2009 at 1:26 AM, hrishy <hrishys@xxxxxxxxxxx> wrote: Hi You can simply do 1) copy /oradata/controlfile.bak to the control_files location specified in the SPFILE (or PFILE); 2) startup mount; 3)recover database 5) alter database open resetlogs; if you configure controlfile auto backups then run{ restore controlfile from autobackup } --- On Fri, 20/11/09, Roger Xu <wellmetus@xxxxxxxxx> wrote: From: Roger Xu <wellmetus@xxxxxxxxx> Subject: RMAN Restore Scenario: all copies of control files are lost To: oracle-l@xxxxxxxxxxxxx Date: Friday, 20 November, 2009, 16:53 Hi RMAN gurus, Say one backs up a database using the following RMAN block daily: backup database archive log all; SQL 'alter database backup controlfile to '/oradata/controlfile.bak'; All copies of the current control files were in the same disk and the disk crashed. Assuming other parts of the database are fine, will the following steps work to restore the database to the minute when the disk crashed? 1) copy /oradata/controlfile.bak to the control_files location specified in the SPFILE (or PFILE); 2) startup mount; 3) find out the most recent archived redo log sequence number (12345) in log_archive_dest_1; 4) restore database until sequence 12346; 5) alter database open resetlogs; The reason this may work is that RMAN automatically searches for archived and online redo logs that are not recorded in the RMAN repository (in this case the backup controlfile). Do you think this is one of the many advantages using RMAN vs user-managed backup? Thanks, Roger Xu