Re: recovery method

On Tue, 08 Jun 1999 20:08:06 -0600, Brian Yan <by2_at_gpu.srv.ualberta.ca> wrote:

>Customer has no
>plan to upgrade the hard disk. The database is run under archivelog
>mode. It is O.K. to lose one day data.

I manage a few databases where this is also true. Nothing wrong with it as long as the client has made a conscious decision that losing a day's data is acceptable.

>Currenly, we have all on-line
>datafile backup and archived log backed to tape every night. I am
>wondering in the case of disk failure, can I just copy all the datafile
>plus the archived logs to the same directory as before (assumed server
>directories have been restored to the image before crash)? then startup
>database, run the recover command?

You should consider turning off archive log mode. For the three databases I manage that fit the case you are describing, we do not use archivelog mode at all. We do cold backups nightly. The recovery plan, for any failure, is simply to restore the previous night's backup. That keeps things simple for the junior DBA who works w/me. He only needs to know one procedure. Assuming that you do a cold backup, then yes, you can just restore the files to the same disk, and start the database. If you are restoring to a different disk, then you will need to issue a bunch of ALTER DATABASE RENAME FILE commands. You will need to know the original filenames in order to do that. For this reason, I suggest maintaining a current list of these. You can generate one by issueing the following select:

	select name from v$datafile
	union
	select member from v$logfile;

We also do nightly exports of all our small databases. Yours must not be too big if it fits on one disk, so you might want to consider an export as well. It gives you another option in case everything else fails.

Jonathan Received on Thu Jun 10 1999 - 11:02:44 CDT