Re: Válasz: Re: Insure DB consistency in case ofrecovery

Heba

your problem here is that you have two separately controlled "automatic"
recovery mechanisms.  Since journalling is at the OS level and the other
mechanism is at the application level it seems inevitable to me that you
will have differences. From what I can see you will always be in the
situation where someone will have to make a call on where to apply your
journals too.

Note that I'm ignoring committment control issues here for the purposes of
discussion

If you have some kind of crash, then the files that are journaled will be
potentially recoverable right up to what was written to the database
(obviously this is dependent on how you manage your journals and hardware).
In the case of the other files, it is not quite so clear which files will
be recovered. I presume that if you are not using journaling they won't be
recoverable except via backup.

If the application doesn't use journalling at any level and can't be
recovered, then its kind of pointless to recover your custom data as it has
nothing to refer to.

Maybe you colud explain a bit further - I feel like all I;ve done is
restate your problem :)

Regards
Evan Harris

>This is a multipart message in MIME format.
>--
>[ Picked text/plain from multipart/alternative ]
>Actually I was looking for a way to maintain the referential integrity
>between our files and application files.
>
>For example, if we have an order and order detail file in the application
>database, and for some reason we have an internal developed file containg
>extra order details. The recovery of the application failed to enter the
>order data in the original application database for any reason, but the
>APYJRNCHG to our external file was successful. Someone has to check the DB
>consistency after recovery using a query for example and my manager does
>not like this although I think such a case is very exceptional. but it be
>be some idea to overcome this...
>
>thanks in advance
>Heba