ADSM-L

Re: query archive using description

1999-08-03 16:57:46
Subject: Re: query archive using description
From: "Thomas A. La Porte" <tlaporte AT ANIM.DREAMWORKS DOT COM>
Date: Tue, 3 Aug 1999 13:57:46 -0700
Robert,

You addressed two issues:

Descriptions

You're right, this is somewhat of a hassle. I can think of a
couple of ways in which you could get around this problem.

 1) Redirect the output of 'select name from v$datafile'
    to a file, then query the archive status for each entry
    in the resulting file.

 2) If you can afford the space, you could do your backups
    to a single location on disk; then you could archive *those*
    files, and the ADSM copies would all come from the same
    location. This would also allow you to restore the
    database more quickly, if you can afford the disk space.

Controlfiles

We do issue an "alter database backup controlfile to FILENAME reuse"
as one of the last commands of our backup procedure. Since the
physical structure of our databases rarely change, and because we
multiplex the control files across three different disks, we
believe that the need to restore using the backup controlfile is
minimized--remember, if you only lose one of your controlfiles,
or even all but one of them, you can simply copy the existing
(good) control file to the locations of the lost control files
and open the database: the contents of each controlfile are
identical.

In my experience, the need to do a complete restoration of a
database--which is what the Redbook deals with, as I recall--is
an extremely unlikely event. Much more likely you will lose a
single disk drive, and the datafiles or log files located on that
particular device. Depending upon what datafiles or log files are
lost, you will then have to make decisions regarding the best way
to proceed with recovery.

If you don't already have a copy, I would highly recommended the
Oracle Press book "Oracle Backup & Recovery Handbook" by Rama
Velpuri. It is very detailed in its explanations of Oracle backup
and recovery procedures, and it has an excellent section of
scenarios that walk the reader through various recovery
procedures.

 -- Tom

Thomas A. La Porte
DreamWorks Feature Animation
tlaporte AT anim.dreamworks DOT com

On Tue, 3 Aug 1999, Grigsby, Robert C wrote:

>Tom,
>
>Thanks for the help! This did work. Unfortunately this was not
>what I was hoping for. Let me explain what I am trying to do:
>
>In the redbook you mentioned documented for a FULL ONLINE backup.
>you do an end begin backup; archive; end backup for each
>tablespace/datafile.
>
>My hope was that by having the same description on each archive that I could
>quickly identify/verify the Oracle datafiles/controlfiles/redo that was
>backed up
>on any given day. It appears that you must specify at least the beginning
>part of the
>mount point for each mount that you have datafiles residing.
>
>Also I may have already found the answer to a question that I asked
>previously,
>Based on section 6.8.2 of the redbook entitled "Restoring All Tablespaces
>and Data Files",
>it appears you use the "recover database using backup controlfile" command
>essentially
>bypassing the controlfiles that never got archived successfully in the first
>place.
>
>Thanks again for the help,
>Robert C Grigsby III
>
>-----Original Message-----
>From: Thomas A. La Porte [mailto:tlaporte AT ANIM.DREAMWORKS DOT COM]
>Sent: Tuesday, August 03, 1999 12:31 PM
>To: ADSM-L AT VM.MARIST DOT EDU
>Subject: Re: query archive using description
>
>
>I suspect that the problem you are encountering here is in your
>source file description of "*", particularly since the file used
>in your second example, "/u01/this.file" is presumably located in
>a filespace called "/u01". Without a filespace modifier in your
>source file description, and probably more importantly, without a
>"-subdir=yes" parameter in your "query archive" command, you are
>effectively asking for files archived with that particular
>description, in the current working directory. So, if you don't
>issue the command from "/u01", you wont find "/u01/this.file".
>
>The command you probably want to issue is this:
>
>"dsmc query archive -desc="FULL ONLINE 990728" -subdir=yes
>"/u01/*"
>
> -- Tom
>
>Thomas A. La Porte
>DreamWorks Feature Animation
>tlaporte AT anim.dreamworks DOT com
>
>On Thu, 29 Jul 1999, Grigsby, Robert C wrote:
>
>>Is there something quirky about using the description option when doing a
>>dsmc query archive?
>>
>>I had created for the first time an archive the previous day.
>>When I went to do a dsmc query archive -description="FULL ONLINE 990728"
>"*"
>>I get ANS1092E No files matching search criteria were found
>>
>>
>>When I do dsmc query archive "/u01/this.file" It works.
>>       Size  Archive Date - Time    File - Expires on - Description
>>104,865,792  07/28/99   15:34:37    /u01/this.file 07/27/00 FULL ONLINE
>>990728
>>
>>I am using multiple archive commands with the same description.
>This ADSM
>>version3 UNIX.
>>
>>Robert C Grigsby III
>
<Prev in Thread] Current Thread [Next in Thread>