MIME-Version: 1.0
References: <CAMazQQfp+3f8tD_Q23rCR=O+Jj4jouSRVigbD8OmrTOfHV+8gA@mail.gmail.com>
 <CA+hUKGLu-oDokjt=Hyb8cLPTm19s5HZSq0PY-jXBP6-u2cqMEg@mail.gmail.com>
In-Reply-To: <CA+hUKGLu-oDokjt=Hyb8cLPTm19s5HZSq0PY-jXBP6-u2cqMEg@mail.gmail.com>
From: Riku Iki <riku.iki.x@gmail.com>
Date: Fri, 26 Apr 2024 16:15:35 -0700
Message-ID: <CAMazQQewmq3Zt+mF1JOTTpJyNyhKQT7C_G7z+MjGxs-xRq0Mxw@mail.gmail.com>
Subject: Re: Preallocation changes in Postgresql 16
To: Thomas Munro <thomas.munro@gmail.com>
Cc: pgsql-general@lists.postgresql.org
Content-Type: multipart/alternative; boundary="0000000000005e95650617081579"
Archived-At: <https://www.postgresql.org/message-id/CAMazQQewmq3Zt%2BmF1JOTTpJyNyhKQT7C_G7z%2BMjGxs-xRq0Mxw%40mail.gmail.com>
Precedence: bulk

--0000000000005e95650617081579
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Thank you, I have such a system. I think my task would be to compile PG
from sources(need to learn this), and see how it works with and without
that code block.

On Thu, Apr 25, 2024 at 2:25=E2=80=AFPM Thomas Munro <thomas.munro@gmail.co=
m> wrote:

> On Fri, Apr 26, 2024 at 4:37=E2=80=AFAM Riku Iki <riku.iki.x@gmail.com> w=
rote:
> > I am wondering if there were preallocation related changes in PG16, and
> if it is possible to disable preallocation in PostgreSQL 16?
>
> I have no opinion on the btrfs details, but I was wondering if someone
> might show up with a system that doesn't like that change.  Here is a
> magic 8, tuned on "some filesystems":
>
>         /*
>          * If available and useful, use posix_fallocate() (via
>          * FileFallocate()) to extend the relation. That's often more
>          * efficient than using write(), as it commonly won't cause the
> kernel
>          * to allocate page cache space for the extended pages.
>          *
>          * However, we don't use FileFallocate() for small extensions, as
> it
>          * defeats delayed allocation on some filesystems. Not clear wher=
e
>          * that decision should be made though? For now just use a cutoff
> of
>          * 8, anything between 4 and 8 worked OK in some local testing.
>          */
>         if (numblocks > 8)
>
> I wonder if it wants to be a GUC.
>

--0000000000005e95650617081579
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Thank you, I have such a system. I think my task would be =
to compile PG from sources(need to learn this), and see how it works with a=
nd without that code block.</div><br><div class=3D"gmail_quote"><div dir=3D=
"ltr" class=3D"gmail_attr">On Thu, Apr 25, 2024 at 2:25=E2=80=AFPM Thomas M=
unro &lt;<a href=3D"mailto:thomas.munro@gmail.com">thomas.munro@gmail.com</=
a>&gt; wrote:<br></div><blockquote class=3D"gmail_quote" style=3D"margin:0p=
x 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On=
 Fri, Apr 26, 2024 at 4:37=E2=80=AFAM Riku Iki &lt;<a href=3D"mailto:riku.i=
ki.x@gmail.com" target=3D"_blank">riku.iki.x@gmail.com</a>&gt; wrote:<br>
&gt; I am wondering if there were preallocation related changes in PG16, an=
d if it is possible to disable preallocation in PostgreSQL 16?<br>
<br>
I have no opinion on the btrfs details, but I was wondering if someone<br>
might show up with a system that doesn&#39;t like that change.=C2=A0 Here i=
s a<br>
magic 8, tuned on &quot;some filesystems&quot;:<br>
<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 /*<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* If available and useful, use posix_fall=
ocate() (via<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* FileFallocate()) to extend the relation=
. That&#39;s often more<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* efficient than using write(), as it com=
monly won&#39;t cause the kernel<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* to allocate page cache space for the ex=
tended pages.<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0*<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* However, we don&#39;t use FileFallocate=
() for small extensions, as it<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* defeats delayed allocation on some file=
systems. Not clear where<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* that decision should be made though? Fo=
r now just use a cutoff of<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* 8, anything between 4 and 8 worked OK i=
n some local testing.<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0*/<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0 if (numblocks &gt; 8)<br>
<br>
I wonder if it wants to be a GUC.<br>
</blockquote></div>

--0000000000005e95650617081579--