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 -> Re: Oracle Backup Log files and recovery

Re: Oracle Backup Log files and recovery

From: Howard J. Rogers <hjr_at_dizwell.com>
Date: Tue, 9 Dec 2003 20:18:06 +1100
Message-ID: <3fd5934f$0$1022$afc38c87@news.optusnet.com.au>

"Hari Om" <hari_om_at_hotmail.com> wrote in message news:d1d5ebe4.0312080802.43d05c52_at_posting.google.com...
> I am using Oracle DB 9.2.0.1 on IBM AIX 5.1L System
>
> I have a instance running on it named "utahprod" which is running in
> ARCHIVEMODE.
>
> My Archive destination is "/storage/archive/utahprod" and everday ONE
> LOG File is generated and Archived and stored in this directory.
>
> My Syntax for this file is "utahprod_1_1.dbf" all the way through
> "utahprod_1_108.dbf"

It is more usual to name archives ".rdo" or ".arc" or something similar. They aren't data files, and your naming convention shouldn't suggest that they are.

>
> My QUESTIONS:
> -------------
> 1) Is this INCREMENTAL or DIRRERENTIAL BACKUP Log files....?

I don't even understand the question. They are archive logs. They are therefore a copy of your last online redo log. And as such, they contain the instructions on how to re-perform all the transactions that took place whilst that last online log was your current log.

> 2) If my Instance crashes, how can I recover from these files?

Instance recoveries, by definition, never require the use of archivelogs to recover. Instance recovery only needs to replay all transactions since the time of the last checkpoint (because at the last checkpoint, dirty data was flushed to the data files, and is therefore safe). You can't produce an archive until you've switched logs. At a log switch, a checkpoint is issued. Therefore, the redo in the log just switched away from (and hence archived) is not needed for instance recovery.

A small subtlety: it is possible you might need to replay redo in the log-before-last, in that the instance crash might occur in the middle of the log-switch induced checkpoint. But even so, that's the online log we're talking about, and not the archived version. Archives are never needed for instance recoveries.

> 3) What happens if one of the above files (say utahprod_1_36.dbf")
> gets corrupted....? What would happen or what are the consequences of
> this corrupted Log file on the recovery of the DB?

It would prevent recovery from progressing to completion. In your case, the best you'd be able to do is to recover to log 35. You'd then have to cancel the recovery, and open your database with a resetlogs. Anything after that log 35 would be lost.

> Any related information on this is appreciated.
>
> THANKS!
>
> HARI
Check out www.geocities.com/lydian_third. Look under the Books link. And download the backup and recovery document you'll find there.

Regards
HJR

-- 
------------------------------------
Oracle insights at www.dizwell.com
------------------------------------
Received on Tue Dec 09 2003 - 03:18:06 CST

Original text of this message

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