MIME-Version: 1.0
References: <CAGCUe0Lwk3C0qdkBa+OLpYc7yXwW=pbaz8Sju4xMXEQAmyp+5g@mail.gmail.com>
 <CAH2-Wzk6-mzP4utr9=sKT9XQi+NgonD_X30z1sf_hgnKQvPpGg@mail.gmail.com>
In-Reply-To: <CAH2-Wzk6-mzP4utr9=sKT9XQi+NgonD_X30z1sf_hgnKQvPpGg@mail.gmail.com>
From: Siddharth Kothari <sidkot@google.com>
Date: Wed, 29 Apr 2026 20:29:23 +0530
Message-ID: <CAGCUe0L+M113dXmW44LWDmqSHa8kMcwAKgW+hXdTwVd-8dz8jA@mail.gmail.com>
Subject: Re: Fix size estimation for parallel B-Tree scans with skip arrays
To: Peter Geoghegan <pg@bowt.ie>
Cc: pgsql-bugs@lists.postgresql.org, Vaibhav Jain <jainva@google.com>, 
	Madhukar <madhukarprasad@google.com>, Xun Cheng <xuncheng@google.com>
Content-Type: multipart/alternative; boundary="000000000000817a5006509a9814"
Archived-At: <https://www.postgresql.org/message-id/CAGCUe0L%2BM113dXmW44LWDmqSHa8kMcwAKgW%2BhXdTwVd-8dz8jA%40mail.gmail.com>
Precedence: bulk

--000000000000817a5006509a9814
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Thank you Peter!

On Wed, Apr 29, 2026 at 8:14=E2=80=AFPM Peter Geoghegan <pg@bowt.ie> wrote:

> On Wed, Apr 29, 2026 at 2:54=E2=80=AFAM Siddharth Kothari <sidkot@google.=
com>
> wrote:
> > Root cause:
> >
> > In src/backend/access/nbtree/nbtree.c, the loop in
> btestimateparallelscan assumes that every index column might require a sk=
ip
> array and adds sizeof(int) to the estimated size:
> >
> > However, every skip array actually needs space for its slot in the
> btps_arrElems array AND space to store its scan key's sk_flags.
>
> Your diagnosis looks correct to me. As you said, the problem is that
> we only add btps_arrElems space overhead for input scan keys -- we
> neglect to do the same for any skip array scan key that might be
> output by nbtree preprocessing later on.
>
> I'll commit this patch later today.
>
> Thanks!
> --
> Peter Geoghegan
>

--000000000000817a5006509a9814
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Thank you Peter!</div><br><div class=3D"gmail_quote gmail_=
quote_container"><div dir=3D"ltr" class=3D"gmail_attr">On Wed, Apr 29, 2026=
 at 8:14=E2=80=AFPM Peter Geoghegan &lt;<a href=3D"mailto:pg@bowt.ie">pg@bo=
wt.ie</a>&gt; wrote:<br></div><blockquote class=3D"gmail_quote" style=3D"ma=
rgin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:=
1ex">On Wed, Apr 29, 2026 at 2:54=E2=80=AFAM Siddharth Kothari &lt;<a href=
=3D"mailto:sidkot@google.com" target=3D"_blank">sidkot@google.com</a>&gt; w=
rote:<br>
&gt; Root cause:<br>
&gt;<br>
&gt; In src/backend/access/nbtree/nbtree.c, the loop in btestimateparallels=
can assumes that every index column might require a skip array and adds siz=
eof(int) to the estimated size:<br>
&gt;<br>
&gt; However, every skip array actually needs space for its slot in the btp=
s_arrElems array AND space to store its scan key&#39;s sk_flags.<br>
<br>
Your diagnosis looks correct to me. As you said, the problem is that<br>
we only add btps_arrElems space overhead for input scan keys -- we<br>
neglect to do the same for any skip array scan key that might be<br>
output by nbtree preprocessing later on.<br>
<br>
I&#39;ll commit this patch later today.<br>
<br>
Thanks!<br>
-- <br>
Peter Geoghegan<br>
</blockquote></div>

--000000000000817a5006509a9814--