Re: RBS01.dbf damaged

You can declare rollback segments corrupt in the init.ora and then get rid of them. This isn't considered kocher, and of course will possibly logically corrupt your data, ie the data is readable but not what you expect because it was stopped mid way and not rolled back. The kocher way to do things is have a good backup scheme in place.

Best
Kyle

On Mon, 19 Feb 2001 14:27:15 +0100, "Achille Carette" <achyl_at_infonie.be> wrote:

>No no, don't have to try - i'm doing it.
>
>I tried the following :
>While a transaction is active (doing inserts in a loop), shut the database
>down "abort". Then rename the file rbs01.dbf into rbs01.old and try to
>restart the DB.
>I mounted the DB, took the rbs01.dbf datafile offline drop, then was able
>to start the DB.
>- The transaction existing before shutdown cannot be rolled back at stratup,
>since rollback segment is unavailable
>- Rollback segment used by this transaction remains active, but is
>unavailable and cannot be dropped
>- Because it contains active rollback segment, RBS tablespace cannot be
>dropped
>- Oracle is unable to read data from the table modified by this transaction,
>since it cannot use the rollback segment to create a before-modification
>image.
>
>I don't konw how to resolve this situation (and to re-create rollback
>segments)...
>
>
>"Howard J. Rogers" <howardjr_at_www.com> wrote in message
>news:3a910636_at_news.iprimus.com.au...
>>
>> "Achille Carette" <achyl_at_infonie.be> wrote in message
>> news:3a90e78a$0$4493$73bec57d_at_news.be.uu.net...
>> > Great answer, thanks a lot.
>> >
>> > And what if there were active transaction when the database failed (it
 will
>> > become impossible to drop the RBS tablespace) ?
>> >
>>
>> I can't think why. Assuming you're starting the Instance from scratch.
>> It's only impossible to drop a rollback tablespace when there are active
>> transactions using a segment within that tablespace. Since the database
>> will be, at best, in mount stage, there can't be active transactions using
>> anything in that tablespace (assuming you do the drop *before* attempting
>> the recover)
>>
>> I should point out that I'm suggesting this all from theory... I guess I
>> could try it for real if you *really* wanted me to!!
>>
>> Regards
>> HJR
>
Received on Tue Feb 20 2001 - 02:47:18 CST