From: | Ashutosh Bapat <ashutosh(dot)bapat(dot)oss(at)gmail(dot)com> |
---|---|
To: | Andrei Lepikhov <a(dot)lepikhov(at)postgrespro(dot)ru>, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> |
Cc: | Andy Fan <zhihui(dot)fan1213(at)gmail(dot)com>, jian he <jian(dot)universality(at)gmail(dot)com>, David Rowley <dgrowleyml(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
Subject: | Re: Report planning memory in EXPLAIN ANALYZE |
Date: | 2023-12-04 07:24:25 |
Message-ID: | CAExHW5stHDXdRi1_pvNwq5R=qEseBeUxL6_v2ZRT3Pns2K6Jsw@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Fri, Dec 1, 2023 at 8:27 AM Andrei Lepikhov
<a(dot)lepikhov(at)postgrespro(dot)ru> wrote:
>
> On 30/11/2023 18:40, Alvaro Herrera wrote:
> > Well, SUMMARY is enabled by default with ANALYZE, and I'd rather not
> > have planner memory consumption displayed by default with all EXPLAIN
> > ANALYZEs. So yeah, I still think this deserves its own option.
> >
> > But let's hear others' opinions on this point. I'm only one vote here.
>
> I agree; it should be disabled by default. The fact that memory
> consumption outputs with byte precision (very uncomfortable for
> perception) is a sign that the primary audience is developers and for
> debugging purposes.
That's 2 vs 1. Here's patch with MEMORY option added. Replying to
Alvaro's earlier relevant comments.
> If EXPLAIN (MEMORY) is added, then probably auto_explain needs a
> corresponding flag, and doc updates.
auto_explain does not implement planner_hook and hence can not gather
information about planner's memory usage. So no new flag and doc
update to auto_explain. I have added a comment auto_explain.c but
probably even that's not needed. We are considering this option only
for developers and auto_explain is largely for users. So I didn't feel
like adding an implementation of planner_hook in auto_explain and use
MEMORY option. If we feel that planner's memory usage report is useful
in auto_explain, it should easy to do that in future.
> The EXPLAIN docs (explain.sgml) need an update to mention the new flag
> and the new output, too.
Done.
0001 is as is except explain.out and explain.sgml changes reverted.
0002 following changes. It's a separate patch for ease of review.
a. implements MEMORY option, adds tests in explain.sql and also
updates explain.sgml.
b. show_planning_memory renamed to show_planner_memory to be in sync
with the output
c. small indentation change by pgindent.
--
Best Wishes,
Ashutosh Bapat
Attachment | Content-Type | Size |
---|---|---|
0002-EXPLAIN-reports-memory-consumed-for-plannin-20231204.patch | text/x-patch | 10.7 KB |
0001-EXPLAIN-reports-memory-consumed-for-plannin-20231204.patch | text/x-patch | 11.8 KB |
From | Date | Subject | |
---|---|---|---|
Next Message | Xiang Gao | 2023-12-04 07:27:01 | RE: CRC32C Parallel Computation Optimization on ARM |
Previous Message | Andrei Lepikhov | 2023-12-04 06:50:34 | Re: POC, WIP: OR-clause support for indexes |