| From: | Michael Paquier <michael(at)paquier(dot)xyz> |
|---|---|
| To: | Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> |
| Cc: | Julien Rouhaud <rjuju123(at)gmail(dot)com>, gkokolatos(at)pm(dot)me, Ranier Vilela <ranier(dot)vf(at)gmail(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>, depesz(at)depesz(dot)com |
| Subject: | Re: Showing I/O timings spent reading/writing temp buffers in EXPLAIN |
| Date: | 2022-04-08 02:34:17 |
| Message-ID: | Yk+fKeLkA7NuzxTS@paquier.xyz |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Thu, Apr 07, 2022 at 09:48:02PM +0900, Masahiko Sawada wrote:
> Oops, the results are opposite:
>
> HEAD: 5367.234 ms
> Patched: 5418.869 ms
I have been playing with external sorts & friends after running an
instance on scissors (fsync=off, PGDATA on tmpfs, etc.), even forcing
a compilation of the code with gettimeofday(), and I can see a
tendency of a ~1% impact with the patch in this configuration. So
there is a tendency, while this is also rather close to the usual
noise range. Anyway, this is going to be helpful for debugging when
temp file I/O is a bottleneck, so.. Applied 0001 with some tweaks to
the docs and some indentation fixes.
Now, 0002 is straight-forward but I need more coffee and lunch..
--
Michael
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Thomas Munro | 2022-04-08 02:56:15 | Re: Windows now has fdatasync() |
| Previous Message | Robert Haas | 2022-04-08 02:26:18 | Re: pg14 psql broke \d datname.nspname.relname |