| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> RE: ORA-1578...block corrupted...error is normal...a block...had a NOLOGGING...operation performed against
Thomas Mercadante:
Sorry, I was referring to the generally given advice about backing up after nologging operations in a sarcastic manner, meaning this sort of error can happen. One possibility, as Chris said, is he recovered before backing up at some point long ago, invalidating the block. Another possibility, trying to recover when data loaded in noarchivelog mode... like doing a big db load, then turning on archivelog mode? But I'm wondering if there is some possiblity of doing a nologging index creation, updating one of the index blocks so something is marked not-a-nologging, then backing up and having some other bit waiting to bite later. Pure speculation, of course. Maybe Rajiv's noting of the append hint is what really happened. (Rajiv, as far as the redo records having a range of blocks, remember that the nologging was supposedly long ago.)
See also
http://metalink.oracle.com/metalink/plsql/ml2_documents.showFrameDocumen
t?p_database_id=NOT&p_id=290161.1
Chris Marquez wrote:
>*HOWEVER*, I regularly apply redo logs from and active PROD to a
database >with many indexes defined as LOGGING='NO' to a database I
restored and >recovered from an RMAN backup. I never have errors on the
recovered >database???
Maybe the recovery doesn't touch those datafiles? I'm not saying nologging is never the way to go, nor that backing up shouldn't work after a nologging... unless there is a bug. But whatever it is, we can all agree that nologging situations might (and, it seems, eventually _will_) require something beyond recovery.
Joel Garry joelgarry_at_anabolicinc.com(949) 609-4020
-----Original Message-----
From: Mercadante, Thomas F (LABOR)
[mailto:Thomas.Mercadante_at_labor.state.ny.us]
Sent: Friday, August 19, 2005 11:27 AM
To: Joel Garry; oracle-l_at_freelists.org
Subject: RE: ORA-1578...block corrupted...error is normal...a
block...had a NOLOGGING...operation performed against
Except that he says he performed full backups subsequent to the "create index...NOLOGGING" which is what you said he should have done.
If your assumption is true then a NOLOGGING option is never the correct way to go (if I read your assumption correctly).
His latest reply now is that he may have done some *other* NOLOGGING operations, and they may be the culprit and not the ones he mentioned originally.
-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Joel Garry
Sent: Friday, August 19, 2005 2:19 PM
To: oracle-l_at_freelists.org
Subject: RE: ORA-1578...block corrupted...error is normal...a
block...had a NOLOGGING...operation performed against
>Does the NOLOGGING attribute stay with the index after it has been
created? It doesn't, >correct?
Welllll...
Obviously, getting the error shows that it does. And perhaps reading
something into the 26040 error, it must be tracked by block, maybe
something in the Integrity Section at the beginning of the block... 10g
might be more informative
http://download-west.oracle.com/docs/cd/B14117_01/server.101/b10755/dynv
iews_1055.htm#REFRN30048
I'm also wondering if analyze...validate might tell us anything. Then again, if this is expected behavior, would it be a misfeature if analyze didn't pick it up?
Joel Garry
http://www.garry.to
-- http://www.freelists.org/webpage/oracle-l -- http://www.freelists.org/webpage/oracle-lReceived on Fri Aug 19 2005 - 14:33:14 CDT
 |
 |