public inbox for [email protected]  

help / color / mirror / Atom feed

From: Mark Kirkwood <[email protected]>
To: Tom Lane <[email protected]>
Cc: David Rowley <[email protected]>
Cc: [email protected]
Subject: Re: Unexpected planner choice in simple JOIN
Date: Thu, 8 Jan 2026 17:55:06 +1300
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>
References: <[email protected]>
	<[email protected]>
	<CAApHDvqLR27kZkK5zaK2-6LQo_RB_wnVg8DHfHj17cTbr+RHtQ@mail.gmail.com>
	<[email protected]>
	<CAApHDvoGwHEx1cFQyy-p_ExDMbkYZQFxE4h0qt1ubkd=CHrioA@mail.gmail.com>
	<[email protected]>
	<[email protected]>
	<[email protected]>
	<[email protected]>

Right makes sense - as I noted...the 'wrong' plan is still pretty fast...

On 08/01/2026 17:51, Tom Lane wrote:
> Mark Kirkwood <[email protected]> writes:
>> A point comes to mind - this is not a particularly unusual setup (i.e
>> relatively small parent table with big child one), so maybe the defaults
>> are not ideal here?
> Very probably.  To my mind, the default costs for parallel query and
> JIT are both unduly optimistic and tend to drive the planner to use
> those features when you'd be better off without.  The reason there's
> not been more argument about them is that the downside of using those
> features on a too-small query is bounded, while the upside of using
> them on very-big queries isn't.  So nobody's invested the effort to
> gather enough evidence to back choosing a different set of defaults.
>
> 			regards, tom lane

view thread (10+ messages)

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients using the --to and --cc options:
  reply via email

  To: [email protected]
  Cc: [email protected], [email protected], [email protected], [email protected]
  Subject: Re: Unexpected planner choice in simple JOIN
  In-Reply-To: <[email protected]>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox