Re: Ora-1555 error

Hi Boodah,

ORA-1555 occurs when a job is holding rows in a rollback segment for read
consistency. When the job finishes, the space is available to other user
jobs. I a query is depending on this for read consistency, then the error
occurs....

The first step is to increase the number of rollback segments (usually double). Be sure to add these to your instance's init.ora file or they won't be brought online when the database is restarted. The interesting

thing is that this usually won't cure the cause of the problem.

The problem is often in the scheduling of the two jobs involved that cause
the error in the first place. Almost always, the error is in the query (duh) so rerunning it later is an option.

And a side note....when setting up Rollback Segments, use multiples of the
Oracle block size for your system (usually 8192--8K), Thus a simple declaration of INITIAL 32K NEXT 32K OPTIMAL 64K can work better that numbers that aren't multiples of the block size.

John

Boodah wrote:

> I have a situation where I have a client running software developed by

> my company. Originally the client had the same Unix platform HP v10.2

> using Oracle 7.2.3 as we did. The client upgraded Oracle to 7.3.4 and

> changed machines and since has been getting an Oracle error during
some
> batch processing. The error is
>
> ORA-1555 snapshot too old (rollback segment too small).
>
> This error did not occur until the upgrade and changing machines. We
> haven't had a chance here to upgrade and release new software. Their
> on-site DBA claims it's a software problem where the commits are
> occuring too fast. This process is in batch mode and commits occur
> every 1000 records.
>
> Has anyone else had any experience with this problem? Other than
> re-compiling are there any thoughts/suggestions?
>
> Thanks,
Received on Mon Aug 17 1998 - 14:40:34 CDT