Re: Back-patch change in hashed DISTINCT estimation?

On Wed, Aug 21, 2013 at 2:54 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

>
> What I'm wondering is whether to back-patch this or leave well enough
> alone. The risk of back-patching is that it might destabilize plan
> choices that people like. (In Tomas' original example, the underestimate
> of the table size leads it to choose a plan that is in fact better.)
> The risk of not back-patching is that the error could lead to
> out-of-memory failures because the hash aggregation uses more memory
> than the planner expected.

FWIW I recently investigated an out-of-memory issue in hash aggregation.
That case was because of use of a large temp table which was not manually
analysed and thus lead to a bad plan selection. But out of memory errors
are very confusing to the users and I have seen them unnecessarily
tinkering their memory settings to circumvent that issue. So +1 to fix the
bug in back branches, even though I understand there could be some
casualties on the border.

Thanks,
Pavan

--
Pavan Deolasee
http://www.linkedin.com/in/pavandeolasee