RE: Database recovery problem

From: Harel Safra <>
Date: Tue, 25 May 2010 09:19:22 +0300
Message-ID: <06cc01cafbd2$3928ce80$ab7a6b80$>


It's not surprising, I've seen it happen many times.  

When the data files are in backup mode the SCNs in their headers are frozen. If the database crashes (server reboot/shutdown abort/instance crash) it tries to do a crash recovery when you start it back up. At that point it looks at the headers of the data files and sees the old (frozen) SCN and since it's an old SCN it decides you should do a media recovery.

It doesn't take the files out of backup mode by itself since from the database's point of view it can't distinguish (by just looking at the file headers) if the database was in backup mode and it crashed OR you restored from a hot backup and really need to do a recovery.  

That's the reason for the alter database datafile end backup command.  

Harel Safra  

From: [] On Behalf Of Niall Litchfield
Sent: Tuesday, May 25, 2010 8:53 AM
Subject: Re: Database recovery problem  

Hi Yechiel, first do a cold backup of what you have & then talk to Oracle I rather suspect. I'm a little surprised that you are getting that message on the list of actions you give, but you may well end up working through a non-standard recovery. The reason for the cold backup is so that any fix won't make things worse.  

Received on Tue May 25 2010 - 01:19:22 CDT

Original text of this message