Re: Lost Writes

  • From: Robert Freeman <robertgfreeman@xxxxxxxxx>
  • To: Richard.Goulet@xxxxxxxxxxx, charlottejanehammond@xxxxxxxxx, oracle-l@xxxxxxxxxxxxx
  • Date: Thu, 6 Aug 2009 09:26:57 -0700 (PDT)

RMAN should detect a corrupt block during a backup and report it. Of course, if 
the block is not in use then it would not be an issue.
Did you try a backup validate check logical database from RMAN?
Did you check v$backup_corruption or v$database_block_corruption to see if 
anything was present in that view ?
The corrupt blocks you had, were they actually in use? Or allocated and not 
assigned to any objects?
Did you try to repair the block?


 Robert G. Freeman
Oracle ACE
Oracle Database 11g RMAN Backup and Recovery (Oracle Press) - ON IT'S WAY SOON!
OCP: Oracle Database 11g Administrator Certified Professional Study Guide 
Oracle Database 11g New Features (Oracle Press)
Portable DBA: Oracle  (Oracle Press)
Oracle Database 10g New Features (Oracle Press)
Oracle9i RMAN Backup and Recovery (Oracle Press)
Oracle9i New Features (Oracle Press)
Other various titles out of print now...
The LDS Church is looking for DBA's. You do have to be a Church member in
good standing. A lot of kind people write me, concerned I may be breaking
the law by saying you have to be a Church member. It's legal I promise! :-)

----- Original Message ----
From: "Goulet, Richard" <Richard.Goulet@xxxxxxxxxxx>
To: charlottejanehammond@xxxxxxxxx; oracle-l@xxxxxxxxxxxxx
Sent: Thursday, August 6, 2009 9:25:46 AM
Subject: RE: Lost Writes


    Best idea in this case is to check that all of your data is available and 
the best way that I know of to do that is to do an exp full=yes to /dev/null.  
If that finishes cleanly then your 99% sure that everything is alright.  After 
that I would take another backup.

Dick Goulet
Senior Oracle DBA
PAREXEL International

-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On 
Behalf Of Charlotte Hammond
Sent: Wednesday, August 05, 2009 5:29 PM
To: oracle-l@xxxxxxxxxxxxx
Subject: Lost Writes

Hi All,

We recently had a disk fault on one of our test servers.  After replacing the 
drive I recovered the database using RMAN only to encounter an ORA-600 [3020] 
part way relating to blocks in the system tablespace.  I had to do a 
point-in-time restore to get the database back.  We lost transactions (but 
since it was test not a big deal).

As I understand it this problem is possibly due to lost writes (i/o not going 
to disk but saying it has) due to the disk drive going bad (it may be due to 
other things, but lets assume that's the problem for the sake of discussion).  
What I'm trying to understand is:

1. Will this only manifest itself during a recovery or will I get a similar 
error if I try to update a stale block using SQL?
2. Given I used the same backup but terminated the recovery earlier, presumably 
the stale block is still there, I just avoiding applying recovery to it.   Do I 
now have a "bad" database? - should I exp/imp?  DBV shows nothing amiss but 
presumably these will not be seen as logical corruptions as the blocks are 
simply old, not badly formed.         
3. I see 11g can detect lost writes in conjunction with a standby.  Is there 
anything I can do on stand-alone 10g?  I'm now very paranoid I'll get the same 
thing in production and really loose data.




Other related posts: