Re: Academic question: How to recover through a gap of archived redol ogs with data loss?

  • From: Karl Arao <karlarao@xxxxxxxxx>
  • To: usn@xxxxxxxxx
  • Date: Fri, 7 Aug 2009 00:40:41 +0800

Below are the two options that have saved me from critical recovery scenarios..

1)  _allow_resetlogs_corruption=TRUE
I've already used it to open a pseudo standby database, we have some
missing archivelogs and we really have to open the database for some
DR testing. Well for me that's a failure, but on the client side they
just want the database to be opened. So if this is a prod database,
well you can do a full EXP then rebuild your database.

2) Event 10231
which causes full table scans to skip corrupt blocks. Had a client
that got a block corruption on the SOURCE$ table (internal table for
DBA_SOURCE view) running on Ebiz 11.5.10. We can't pull the data out
through RMAN or EXPDP, can't even do BLOCKRECOVER because the
archivelogs are already deleted. So we have to set this event, then do
a full EXP, recreate the database, then full IMP. There were some
invalid objects on the APPS schema, luckily they are not using the
modules affected.




- Karl Arao
http://karlarao.wordpress.com



On Fri, Aug 7, 2009 at 12:24 AM, Martin Klier<usn@xxxxxxxxx> wrote:
> Hi list,
>
> I know it's unsecure, unsupported and ugly. But given this situation:
>
> I need a recovery of a tablespace, all archivelogs are there, bu I miss
> one from last night. I want to do a "complete" recovery.
>
> The log that's missing is out of a low activity period, and I am ready
> to acceppt slight data corruption, and I will set up the DB as a new
> afterwards, but I need the data from this morning.
>
> What can one desperate DBA do?
>
> To make it clear: I don't have this situation at the moment, but someday
> this may change, and I would prefer to test this stress situation in
> peacetime (now).
>
> Any hints, advice, procedures?
>
> Thanks a lot
> Martin Klier
> --
> Usn's IT Blog for Linux, Oracle, Asterisk
> http://www.usn-it.de
>
> --
> //www.freelists.org/webpage/oracle-l
>
>
>
--
//www.freelists.org/webpage/oracle-l


Other related posts: