Undo Memory Block Cocrruption - DB Crashes

Our production database (Oracle 9.2.0.4.0 on Solaris 10) has crashed 6-7 times 
in the last two months due to error below.  The file involved is always a file 
in the Undo tablespace.  
   
  Corrupt block relative dba: 0x37039d6e (file 7381, block 236910)
Bad check value found during preparing block for write
Data in bad block -
 type: 2 format: 2 rdba: 0x37039d6e
 last change scn: 0x0533.57fb9135 seq: 0x2 flg: 0x04
 consistency value in tail: 0x91350202
 check value in block header: 0x0, computed block checksum: 0xa5fd
 spare1: 0x0, spare2: 0x0, spare3: 0x0
***
Tue Feb 13 11:55:59 2007
Errors in file /d1000/app/oracle/admin/dwprod/bdump/dwprod_dbw2_9274.trc:
ORA-00600: internal error code, arguments: [kcbzpb_1], [922983790], [3], [1], []
, [], [], []
   
   
    We have an SR open with Oracle and their recommendation is to enable a 
couple of init parameters, but when we did that the performance overhead was 
unacceptable (db 2-3X slower).  These parameters are:
   
  DB_BLOCK_CHECKING =TRUE
_DB_BLOCK_CACHE_PROTECT =TRUE
   
  The other suggestion is to have the SA do a "low-level" memory diagnostics to 
check for memory problems.  We have this scheduled in a few days.
   
  Has anyone encountered this before or has any suggestions?  

Thanks,
  Abdul

 
---------------------------------
Now that's room service! Choose from over 150,000 hotels 
in 45,000 destinations on Yahoo! Travel to find your fit.

Other related posts: