| From: | Michael Christofides <michael(at)pgmustard(dot)com> |
|---|---|
| To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
| Cc: | stepan rutz <stepan(dot)rutz(at)gmx(dot)de>, Matthias van de Meent <boekewurm+postgres(at)gmail(dot)com>, Tomas Vondra <tomas(dot)vondra(at)enterprisedb(dot)com>, pgsql-hackers(at)postgresql(dot)org |
| Subject: | Re: Detoasting optionally to make Explain-Analyze less misleading |
| Date: | 2024-07-08 18:13:07 |
| Message-ID: | CAFwT4nDhfSujstTsaezDm9xJHNrdwwRTw4GGokTExaz=85654w@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
> I'm not sure there's a need for it. When a query runs under
> auto_explain, the output values will be sent to the client,
> so those cycles should be accounted for anyway, no?
>
Yes, great point, the total duration reported by auto_explain includes it.
Explicit serialization stats might still be helpful for folks when it is
the bottleneck, but less useful for sure (especially if nothing else causes
big discrepancies between the duration reported by auto_explain and the
"actual total time" of the root node).
(Perhaps the auto_explain documentation should mention this?)
>
I'd value this. I notice the folks working on the other new explain
parameter (memory) opted to add a comment to the auto_explain source code
to say it wasn't supported.
Thanks again,
Michael
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Melanie Plageman | 2024-07-08 18:16:13 | Re: Confine vacuum skip logic to lazy_scan_skip |
| Previous Message | MARK CALLAGHAN | 2024-07-08 17:57:37 | Re: debugging what might be a perf regression in 17beta2 |