Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Mailing Lists -> Oracle-L -> not time-critical: recovery on rawfiles

not time-critical: recovery on rawfiles

From: Oliver Artelt <>
Date: Tue, 19 Sep 2000 22:12:28 +0200
Message-Id: <>

Hello everyone,

I've encountered a situation that I've solved in the meantime, but I want to get an experienced opinion. Maybe someone could find a litte bit time for this. OK, the story:
penguin (SuSE 7.0), EE81610, test database. It is in archivelog mode, the last complete backup some days ago. I've expanded a ts with another datafile, this time a raw device. (Linux is other than sunos, you must bind /dev/raw??? to the block devices). And then the big mistake: I've forget to specify the size of the rawfile (the SIZE and the REUSE-option). dba_free_space now shows silly values for that tablespace (16 TByte free) and during creating objects on that file the db has crashed.
And now how to recover such thing? I cannot resize the file because I've assumed a media failure/physical error firstly and test that device by coping data on that file with dd (This action is completely bullsh.t but O.K. for a test-db, playing with recovery). Now the file header is missing. In my suggestions- in this situation I had to go on another file (RENAME AS) and apply the arcredologs there. How could I do this? If I select another raw device I got a message like 'not a oracle-device / corrupt header'. I had to select a file that is known in the ctrlfile but how can I do this- if I create another file in an ts and will use this oracle says 'already used'. I don't know how I can create a new or overwrite the corrupt file to replace the old one. And what's with the scn? Is the scn for the corrupt file stored in the ctrlfile to know where to start with the recovery?

(in german: Ich sehe den Wald vor Bäumen nicht!) Received on Tue Sep 19 2000 - 15:12:28 CDT

Original text of this message