Re: Bad Blocks ORA_01578

From: Bob Manieri <bmanieri_at_ford.com>
Date: 1996/05/20
Message-ID: <4npnni$e74_at_eccdb1.pms.ford.com>#1/1


Nina Boss <nina.boss_at_ccmail.adp.wisc.edu> wrote:
>We gave two Oracle servers. Version 7.1.6 and Version 7.2.2.4., on an RS/6000
>with AIX v4.2. The disks are FWD by Seagate. We use the servers mainly in
>client/server mode. With tools ranging from BRIO query to Oracle Card. Both
>servers are mainly datawarehouse applications.
>
>Our problem, we have gotten Oracle BAD BLOCK erros over and over again. We move the
>tables and indexes all over the disks (each server has 12 of them), and drop and
>recreate them in many different tablespaces. STill we get these errors.
>
>The kicker is, many times the file number and block number reported doesn't have a
>bad block. We know because we use Oracle's method for scanning a block by it's HEX
>record id.

Nina: Its probably a little after the fact, but have you set the init.ora parms db_block_checksum and log_block_checksum to TRUE. These will make Oracle to checksums on blocks of data before he writes them out. If the checksums don't match I believe the process about to write will crash. You'll need 7.2 or higher for this, but it can be done thru undocumented init.ora parms or 'events' with older versions.
Just an idea, albeit a little late.

-- 
========================================================================
  Bob Manieri, Database Administrator, Ford Electronics, Lansdale, PA
    SELECT 'These are my opinions and not those of my employer.' 
          FROM SYSTEM.DUAL WHERE DISCLAIMER = 'YES' ;
Received on Mon May 20 1996 - 00:00:00 CEST

Original text of this message