RE: dbv - Found block already marked corrupted

  • From: "Bobak, Mark" <Mark.Bobak@xxxxxxxxxxxxxxx>
  • To: <oracle-l@xxxxxxxxxxxxx>
  • Date: Mon, 2 Feb 2004 10:12:42 -0500

Actually, there is a utility that can do what you want.  It's called
bbed.  But, since it's possible, even easy, to not only shoot yourself
in the foot, but actually blow an entire leg off, that's all I'll say
about it.  ;-)

-Mark

Mark J. Bobak
Oracle DBA
ProQuest Company
Ann Arbor, MI
"Imagination was given to man to compensate him for what he is not, and
a sense of humor was provided to console him for what he is."  --Horace
Walpole


-----Original Message-----
From: Kaing, Leng [mailto:Leng.Kaing@xxxxxxxxxxxxxxxx]=20
Sent: Sunday, February 01, 2004 8:55 PM
To: oracle-l@xxxxxxxxxxxxx
Subject: Re: dbv - Found block already marked corrupted
Importance: High


Mladen,

You're of course right again! I've done all I can - rebuilt the index, =
=3D
dropped the index, created the index, yet the block is still reported as
=3D
already marked corrupted. And this time when I do a cdba it points to an
=3D
empty block. Why isn't there a utility to unmark an empty block that was
=3D
previoiusly "marked corrupted"!

I've almost lost faith in this utility now (or maybe I already have!) =
=3D
We've just scheduled a weekly job to do a dbv to detect coruption. Now =
=3D
I'm getting tons of blocks marked corrupted even though they're clean  =
=3D
(I know because we do a drop and import from production every night!) It
=3D
would take me forever to make a list and say "if this block for this =3D
database comes up, ignore it" ARGH!!!

Yes, I'm waiting for the unflag utility too.=3D20

Leng.

----------------------------------------------------------------
Date: Fri, 30 Jan 2004 16:02:41 -0500
From: Mladen Gogala <mladen@xxxxxxxxxxxxxxx>
Subject: Re: dbv - Found block already marked corrupted

On 01/30/2004 03:46:04 PM, Tim Gorman wrote:

> One further question, though...
>=3D20
> >> Of course, if someone knows a utility to un-flag a block I'm all
> ears.
>=3D20
> Why would anyone want to "unflag" a block that has been flagged as
> corrupt?
> I'm all ears for that...  :-)
>=3D20
> -Tim


Actually, not all corrupt blocks are the same. The definition of a =3D20
corrupt block is "a block with a bad checksum". These bad checksums are
mostly caused by software. I have a ton of experience with "corrupt =
=3D20
blocks" which came into being after copying production database to =3D20
development database using BCVs. This means hot backup and instance =
=3D20
recovery. For the reasons of speed, the tables with the computed data
=3D20
were created with "NOLOGGING" attribute and when access was attempted
=3D20
in the cloned instance, oracle started screaming about "corrupted =3D20
blocks". When the table was truncated  and the data computed again, the
=3D

"corruption" was gone. In other words, so called "corrupt blocks" can
simply be flagged "honest" and reused, if they're empty.



----------------------------------------------------------
Leng Kaing
Email: leng.kaing@xxxxxxxxxxxxxxxx
Phone: +61-3-9203-7589
Mobile: +61-417-371-348

----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to:  oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to:  oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------

Other related posts: