Re: Oracle 8i Restoring Hot backup on a different server - Solaris

Manoj Rajshekar wrote:

> Thanks Howard, for the reply. That was useful. Sorry I forgot to
> mention the version . It's 8.1.7.4 on Solaris 8.
>
> So, do I have to have a DB created on the new server, for me to
> restore, or do I create this from the hot backup files ?

From what you originally wrote, you are asking to clone your existing database onto a new server. That means restoring the hot backups for Database A onto server B. You will have to do a lot else to server B, of course, before it can make sense of and work with a restored backup from somewhere else, or even its own, "native", databases... but most of those things will be sorted by virtue of you having installed Oracle onto Server B in the first place (I'm referring to things such as O/S groups being created, environment variables being set and so on).

Assuming you have treated Server B as you would any machine onto which you are about to install Oracle, then you should be able to simply restore the hot backup from Server A, set an appropriate value for the ORACLE_SID on Server B, startup nomount and you're away.

> Also if you could specify the files which I need to have as part of
> the hot backup, that will get me started.

Everything except for the online redo logs, since you never, ever restore online redo logs... not for normal database recovery operations, anyway. If you happen to have a special CLOSED database backup which includes them because you had the intention of cloning the database, then that is a rather different matter, because it would avoid the need for a resetlogs on the clone and the cloning procedures would be a bit simpler.

But for routine backup and recovery, hot backups should include all data files, the init.ora, the spfile (if you were on 9i), the password file, and all previously not backed up archive logs. They should also include an alter database backup controlfile to trace and an alter database backup controlfile to '/somewhere/binary/controlfile/backup'. And I'd throw the alert log in there whilst I was at it.

But now I'm on the topic, what you *really* want to do is start learning and using RMAN. Because the command to perform a hot backup in RMAN would be something along the lines of:

run {allocate channel d1 type disk;
backup database format '/somewhere/%U';
sql 'alter system switch logfile';
backup archivelog all delete input format '/somewhere/%U';}

And in those four lines, you've transferred all the hassle of worrying about what gets backed up when to RMAN (which, incidentally, never backs up online redo logs).

Once you've got that under your belt, you're ready to experiment with RMAN's duplicate database command, and hot cloning of databases becomes about an 8-line job and (just to mix metaphors) a piece of cake.

Regards
HJR  
>
> Regards,
> -Manoj
Received on Tue Oct 26 2004 - 14:16:47 CDT