ADSM-L

Re: [ADSM-L] How to restore (already expired) objects from tape?

2015-04-14 12:16:24
Subject: Re: [ADSM-L] How to restore (already expired) objects from tape?
From: Gregor van den Boogaart <gregor.boogaart AT RZ.UNI-AUGSBURG DOT DE>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 14 Apr 2015 18:14:35 +0200
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Krzysztof, hi Steven, hi Steven,

thanks for all the hints on restoring the DB and commercial
alternatives! I had the feeling that this would be a supported
approach, but was not aware of the details. I am now... Due to unlucky
timing our DB backups are probably expired, but there would be a
chance the older DB backup tapes could be identified and have not been
reused. So it would theoretically be possible.

As only a single - although important - object has been lost, the
decision over here was to focus on preventing this from happening
again. Restoring the DB and all that would possibly have a to big
impact on normal TSM operations for us this time.

However I still would like to restore the object. Therefore the last
question. I came across this

http://www.backupcentral.com/mr-backup-blog-mainmenu-47/13-mr-backup-blog/179-reading-tsm-tapes.html

which mentions

http://sourceforge.net/projects/tsmtape/

as an open source possibility to read TSM written tapes:

"TSMtape allows recovery of files found on TSM v5.x (tested against
5.2) tapes."

I wonder whether anybody has used this successfully (we are still
running TSM 5.5)?

Gregor



On 04/14/2015 03:14 PM, Steven Langdale wrote:
> We had a data loss "incident" a couple of years back.  There is no
> way to re-populate the TSM DB from a tape. You have 2 options: 1.
> Do a DB restore to before the data was deleted from the TSM DB 2.
> Use a data recovery process to re-read the tapes.
> 
> We used Option 2, and got most of it back.  IBM supplied the
> program to read the data from tape and put it into a directory
> structure.
> 
> Steven
> 
> On Tue, 14 Apr 2015 at 14:05 Krzysztof Przygoda <przygod AT gmail DOT com>
> wrote:
> 
>> Hi I guess its better to disable schedules before you start (by
>> adding disablescheds yes in dsmserv.opt file) the resorted server
>> than to do it after start (as some jobs can start just after
>> startup).
>> 
>> http://www-01.ibm.com/support/knowledgecenter/SSGSG7_7.1.0/com.ibm.itsm.srv.ref.doc/r_opt_server_disablescheds.html
>>
>>
>> 
Kind Regards
>> Krzysztof
>> 
>> 2015-04-14 14:48 GMT+02:00 Steven Harris
>> <steve AT stevenharris DOT info>:
>>> Hi Gregor.
>>> 
>>> IBM used to offer a special service to get data off tape. Not
>>> sure if it is still available.
>>> 
>>> Other than that the only way I know is to restore your DB to a
>>> point before expiration.  Start up the restored instance with
>>> NOMIGRECL and immediately stop your daily schedules, in
>>> particular stop running expiration.  At this point you can
>>> restore your file.  I would tend to mark the primary volumes as
>>> unavailable and restore from the copypool. That way you if you
>>> have restored to another instance you can keep your main
>>> instance running close to normal (except for copypool stuff)
>>> for however long the restore takes you.  But it is a lot of
>>> work and a lot of fiddling about.
>>> 
>>> Regards and good luck.
>>> 
>>> Steve
>>> 
>>> Steven Harris TSM Admin Canberra Australia.
>>> 
>>> 
>>> 
>>> 
>>> On 14/04/2015 10:10 PM, Gregor van den Boogaart wrote:
>>>> Dear List,
>>>> 
>>>> the retention settings in the copygroup were to strict... The
>>>> version of the file we would like to restore has already been
>>>> expired. :-( However it is very likely the desired version of
>>>> the file is still on tape, as no reclamation has taken place.
>>>> So my understanding would be: if the file is still on tape,
>>>> the "only" thing missing is the database entry. Correct?
>>>> 
>>>> So the question would be: How to restore already expired
>>>> objects from a tape cartridge? Or: How to generate database
>>>> entries for all versions of all objects on a tape cartridge?
>>>> 
>>>> I guess the first step would be to move the respective node
>>>> to a copygroup, with "better" retention settings. I looked on
>>>> the documentation for "audit volume", but I did not get the
>>>> feeling this would understand checking "for inconsistencies
>>>> between database information and a storage pool volume" in
>>>> un-expiring previously expired objects...
>>>> 
>>>> Any tips, pointers, ideas? Perhaps I am just searching for
>>>> the wrong key words?
>>>> 
>>>> Gregor
>>>> 
>>>> 
>>>> 
>> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=yp3D
-----END PGP SIGNATURE-----