Re: Block Corruption in empty pages in Oracle 11g
Date: Sat, 7 Aug 2010 10:39:53 -0700 (PDT)
Message-ID: <486b32d6-4cca-4172-97e8-42cb18607b42_at_w30g2000yqw.googlegroups.com>
On Aug 7, 5:54 am, John Hurley <hurleyjo..._at_yahoo.com> wrote:
> Mladen:
>
> > The command is:
>
> > ALTER SYSTEM DUMP DATAFILE <file_id> BLOCK MIN <min block>
> > BLOCK MAX <max block>;
>
> > On a real OS, one can have rather satisfactory results by doing:
>
> > dd if=<data file> bs=8k count=<count>|od -x
>
> > If the block size is not 8k, the corresponding value needs to be
> > substituted into the dd command.
>
> Some people actually take the time to ( on a test system ) introduce
> damage and test out their ability to find problems and recover from
> them. Shocking eh?
>
> Of course a hex editor can let you introduce just little tiny bits of
> corruption ... especially with tools that show one the structure of
> the blocks.
Thanks to all for provding your input; very useful. I do know how to corrupt a blok using dd command. In fact, several years ago when I took training on Oracle 9i or 10g; one of the classroom exercises was to corrupt a block and recover it using rman block recover command. I tried it on my database and I it seems to take same amount of time as full recover; because it scans entire backup to serach for the blocks specfied in the block recver command.
I am still at loss why this block corruption is happening; I have seen it twtice on this system. It is hard to convince UNIX Administartors that something is wrong in disk or disk controller; I even don't know how they will find that. All they can do is to replace disks and disk controllers. Received on Sat Aug 07 2010 - 12:39:53 CDT