Re: Weird "ORA-03113: end-of-file on communication channel" during parse

I think you had an example with "value in (-1, 2) and rownum <= {count} - 1" that worked.

If it's an index defect then I'd consider the possibility that the problem is at the end of the -1 range, did you have some comment (I've lost the trail) about getting failures with other "single value" queries? What about two-value queries. The other thing I'd try (since there's a gap between -1 and 2 is to query value in (-1,1) which I think was the two consecutive values.

(I'd have to day that I would be a little surprised if it were a physical index problem as the crash also appears on the explain plan - but maybe there's a dynamic-sampling effect that's causing the crash in that case).

Are there any other indexes on the table that Oracle might be considering as possible alternatives ?

Regards
Jonathan Lewis

On Fri, 26 Mar 2021 at 10:35, Kim Berg Hansen <kibeha_at_gmail.com> wrote:

> A short update:
>
>
> _at_Mark:
> I'm guessing too that an index rebuild or similar could fix the issue.
> At the moment my main interest is to diagnose (mostly in order to be able
> to tell if this might be an issue in other of our environments or tables),
> so at the moment I don't want to "destroy the error-causing scenario" by
> fixing it ;-)
>
>
>

--
http://www.freelists.org/webpage/oracle-l