As someone else pointed out, you don't have block corruption just rman
re-reading a block. You might want to consider putting into place a
periodic rman "backup validate check logical database" command "if you
are able to" (depends on size of databases and open windows of
opportunity to run these commands without impacting system throughput),
followed by a check of v$database_block_corruption.
The oracle product documentation is still somewhat weak in this area.
Try on a test system forcing bad blocks ( dd is what you use on unix to
do this ) and do some experimentation to make sure you know what you
are looking at and how to address it. Depending on hardware and dasd
configuration these types of areas are (thankfully) usually pretty rare
(depends).
Received on Thu Dec 23 2004 - 18:08:41 CST
