Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Hot backups and rollback segments.

Hot backups and rollback segments.

From: Mike Burden <mburden_at_uk.att.com>
Date: Mon, 18 Jan 1999 11:58:23 +0000
Message-ID: <36A321DE.FD015CE5@capgemini.co.uk>


I've recently been playing about with Oracle in Archive Log Mode and Hot backups but one or two things are confusing me.

I understand the need for all the data files to be in line before Oracle allows access to the data but I'm a bit confused as to why the rollback segments need backing up.

Q1. Do you need to hot backup the rollback segments??? You can 'begin backup' on the tablespace but what on earth is this doing. Surly it doesn't write the information for the rollback segments to the redo logs.???? How can the system keep on running if the roll back segments are not writeable whilst a backup is taken.

Q2. Why does the RBS tablespace need to be synchronized with the other dbf files. Surly once you've rolled forward to a point in time (or whatever) then the contents of this table space are irrelevant as all truncations are now complete. Only the structure of the rollback segments is important for a recovery, not their contents?

Q3. Is it possible to tell Oracle that the contents of a dbf file or tablespace are not required? Say for instance you've backed up all required datafiles but forgotten to back up the the users table space. You knows there's nothing on it you require but somehow you need to indicate this to Oracle. This would be useful for the RBS tablespace.

Q4. How does one roll forward if you've lost the latest control file and your current (not archived) redo logs have been cleared. It seems to use the control file to figure out which redo logs need applying (v$loghist data comes from the control file). However, if you've lost the control file is there some manual way of apply all changes from the archived redo log files. The only way I could do it was to bring the database online from a fixed point backup. Switch the logs a few time to increase the SCN so it was above the SCN I new was on my redo logs. Restore the datafiles again and then roll forward. By luck it worked but there must be a correct way of doing it.

I could go on but lets just see what a fool I've made of myself first. Received on Mon Jan 18 1999 - 05:58:23 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US