Naveen,
You've not given much background to the problem, but the answer to
your problem. May be a lot simpler than tracing.
Ora-01555s normally stem from problems from rollbacks. Either to few
(causing contention) or too small. So increase the number or make
them larger.
Another cause can be SQL doing fetches across commits. Usually
getting the code to commit less often solves this.
BUT, if your database is large (shall we say 100G+) then things can
get more complex=E2=80=A6..
On Wed, 10 Nov 2004 05:36:40 -0600, Nahata, Naveen (US - Hyderabad)
<nnahata_at_deloitte.com> wrote:
> Hi All,
>=20
> I see a few ORA-01555: Snapshot too old errors in the alert log daily.
> The trace file doesn't give much clue about the application/SQL that is
> causing the error.
>=20
> I'm planning to set event 1555 at the system level to get to the bottom
> of the problem.
>=20
> 1. What are the performance implications of setting this event? Is the
> impact high or low?
> 2. Should I use systemstate or errorstack in the event specification?
> What are the performance implications of both?
>=20
> Since I have no experience of setting this event in the past, please
> provide any useful information which I have not sought for in my above
> questions.
>=20
> Regards
> Naveen
>=20
> This message (including any attachments) contains confidential =3D
> information intended for a specific individual and purpose, and is =3D
> protected by law. If you are not the intended recipient, you should =3D
> delete this message. Any disclosure, copying, or distribution of this =
=3D
> message, or the taking of any action based on it, is strictly =3D
> prohibited.
> --
> http://www.freelists.org/webpage/oracle-l
>=20
--=20
David Cheyne
Oracle Database Administrator
BA(hons.)
--
http://www.freelists.org/webpage/oracle-l
Received on Wed Nov 10 2004 - 06:40:02 CST
