Recover database calls for backup pieces far earlier than the level 0 backup

From: Gus Spier <gus.spier_at_gmail.com>
Date: Wed, 3 Sep 2014 21:48:20 -0400
Message-ID: <CAG8xnicZpbv3Qgo+RFkP_LEt80JPt6cOWi4e7DrRA1TGW08enA_at_mail.gmail.com>



Hi guys!

Production database became unusable when rows in certain key tables "disappeared". The cause of that is still under investigation. DBAs and application analysts determine the point in time just before this event occurred.

Backups are on hand for AUGUST 31 (Level 0), SEPTEMBER 1(Level 1), and SEPTEMBER 2 (Level 1). Archived redo logs are on hand

DBA team datapump the database in its current state to preserve the "crime scene" and elect to perform database point-in-time-recovery. The restore portion of this procedure appears to complete without complication.

When it comes time to recover the database, RMAN lists the backup pieces it needs to recover the database. The list includes backup piecdes, archived redo logs, gong back as far as January 31, 2013!! Does anyone have a clue why RMAN might act that way? We had expected that the latest INC 0, INC1, and archived redo logs would have sufficed to recover the database.

Any clues would be appreciated.

Thanks,

Gus

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Sep 04 2014 - 03:48:20 CEST

Original text of this message