| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | David Rowley <dgrowleyml(at)gmail(dot)com> |
| Cc: | Onder Kalaci <onderk(at)microsoft(dot)com>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Assertion failure with LEFT JOINs among >500 relations |
| Date: | 2020-10-09 02:06:25 |
| Message-ID: | 2670122.1602209185@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
David Rowley <dgrowleyml(at)gmail(dot)com> writes:
> Are you worried about the costs above the join that triggers that
> coming out as NaN with that fix? It appears that's the case.
[ pokes at that... ] Yeah, it looks like nestloop cost estimation
also has some issues with inf-times-zero producing NaN; it's just
not asserting about it.
I notice there are some other ad-hoc isnan() checks scattered
about costsize.c, too. Maybe we should indeed consider fixing
clamp_row_estimate to get rid of inf (and nan too, I suppose)
so that we'd not need those. I don't recall the exact cases
that made us introduce those checks, but they were for cases
a lot more easily reachable than this one, I believe.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Kyotaro Horiguchi | 2020-10-09 02:12:01 | Re: [Patch] Optimize dropping of relation buffers using dlist |
| Previous Message | Tom Lane | 2020-10-09 01:41:55 | Re: Expansion of our checks for connection-loss errors |