Designing a strategy for periodical block corruption detection.

From: ErikYkema <erik.ykema_at_gmail.com>
Date: Sat, 21 Jul 2007 18:41:05 -0000
Message-ID: <1185043265.444318.31340@22g2000hsm.googlegroups.com>

ErikYkema wrote:
> (Posted this per GoogleGroups few days ago but never found it back in
> the group.)
>
> Hi,
> We were discussing new RMAN features and touched the topic Block
> Change Tracking (BCT).
> A theoretical drawback of BCT could be of that not all blocks are
> checked daily for corruption in a BCT setup.
>
> If I combine BCT with Incrementally Updated image copies, in such I
> way that I will only take incremental backups, how will I find my
> corrupt blocks?
>
> 1 - is there indeed a risk of not being early warned for a bad block
> when switching to BCT+Incremenally Updated
> 2 - what would be a wise mitigation of this risk?
> (periodical "analyze validate structure"? Take a level 0 once in a
> while?)
>
> Regards, Erik Ykema

Is it ethical to continue on my own thread... and I changed the title: "designing a strategy for periodical block corruption detection" for Oracle 10gR2 (on AIX5).

I would appreciate your input for designing a sound strategy for proactive periodical detection of corrupt blocks. See the previous post for the incentive.

I found in http://download.oracle.com/docs/cd/A91202_01/901_doc/server.901/a90117/repair.htm, some methods for corruption detection:

- dbms_repair.check_object
- dbverify (offline)
- analyze validate structure
- DB_BLOCK_CHECKING initialization parameter (default false)

and in the Backup & Recovery Advanced User's Guide (RMAN):

- physical block corruption detection during backup (default enabled)
- logical block corruption detection during backup (default disabled)
- and ad hoc for physical and logical (without making an actual

backup) with the BACKUP VALIDATE command.

What do _you_ use proactively?

Or is block corruption such a small risk that we can approach it reactively? (We never seem to have had a bad block in 8 years). I am struggling with this issue since we intend changing our backup strategy to BCT and hence implicitly are changing our physical corrupt block detection policy.

Any input is appreciated, regards, Erik Received on Sat Jul 21 2007 - 13:41:05 CDT