From: finite9 on 15 Dec 2008 10:01
On Oct 17, 10:06 am, Yves <yve...(a)gmail.com> wrote:
> did your solution work ? Because we are having the same problem since
> we upgraded one of our databases from 10.2.0.3 to 10.2.0.4. (value too
> large notifications on programs that haven't changed. always
> Yves Van Wert
Did anyone get this resolved? I am in the same situation, using an
large in-house application suite that has worked (in this scenario)
without problems on 8i, 9i (all patch levels) and 10.2.0.1. We have
more than 60 customers running this software. Since patching some
customers to 10.2.0.4, we have experienced this issue, where they have
not changed the application at all. Our developers have simulated
test runs with 10,000 records where only a handful have generated
ORA-12899 and the other 9000+ have gone through OK. It is not even
the same field that generates the ORA-12899 despite the fact that the
same rows are being processed in a batch job with the same trigger.
Sometimes it is a date field, but other times, it can be a text field
that holds the userid of the session, that is always a fixed length.
There are several triggers of the same triggeringevent that exist on
the table, but this has not been an issue in the past [pre 10.2.0.4].
We do not have any non-default NLS settings. This seems lika bug in
10.2.0.4 that has not yet been resolved or even reported, as I cannot
find anything useful on Metalink.