Michael Fuhr wrote:
On Thu, Mar 09, 2006 at 11:38:40AM +1100, Noel Faux wrote:
Have a look at this post:
http://archives.postgresql.org/pgsql-general/2006-03/msg00002.php
I have the same problem and we are in the process of fix it.
But we should point out that we're not "fixing" it in the sense of
recovering data. We're wiping out bad blocks; whatever data was
there has been corrupted so we're telling the database to forget
about it.
I haven't tried it, but I wonder how the database would respond to
pulling an old, good copy of the block from a filesystem-level
backup and plugging it into the data file where the bad block is.
I'm sure that could cause problems, but if the data hadn't changed
since before it went bad then I wonder if that might work.
That's a very good idea, but one really needs to have setup a robust / regular backup of the database. Has it been tried before?