'maxcorrupt' and corruption detection by RMAN

'maxcorrupt' and corruption detection by RMAN

Post by Ravi » Sun, 21 Oct 2001 02:52:55



Environment details:

RDBMS Version: 8.0.6.1
Operating System and Version: Solaris 5.8
Product (i.e. SQL*Loader, Import, etc.): Recovery Manager (RMAN)
Product Version: 8.1.5

Problem:
One of our production database's datafile was corrupted and dbverify showed
this error.

DBVERIFY - Verification starting : FILE = /blah/unix/cen/oracle/
datafiles/vc1merk/VC1MERK.1.G1MP01.dbf
kdbchk: xaction header lock count mismatch
trans=1 ilk=1 nlo=2
Page 22966 failed with check code 8

DBVERIFY - Verification complete

Solution:
We exported the data, dropped the tablespace, recreated the tablespace and
imported the data.

Now to the most important part.
Question/s:

1. Why did not R-man backups detect this corruption? I am 90% sure noone
changed the 'maxcorrupt' parameter so it should be set to zero. Are there
certain corruptions that RMAN doesn't detect?

2. How do I check my present maxcorrupt value to be 100% sure if I had to issue
the 'set maxcorrupt to 0' command in RMAN?

3. Is the 'set maxcorrupt to 0' need to be issued everytime I do R-man backup?
Or once its set, its set forever, until we change it again by another 'set
maxcorrupt to n' command?

Thanks a lot in advance for your responses.