MIME-Version: 1.0
References: <19373-aac0a0ee0aac6a8b@postgresql.org>
In-Reply-To: <19373-aac0a0ee0aac6a8b@postgresql.org>
From: surya poondla <suryapoondla4@gmail.com>
Date: Fri, 16 Jan 2026 16:09:02 -0800
Message-ID: <CAOVWO5oKxGOdaGSS5QDH8WnpNqjiZ=W5PjP8_Uk4b2x6=EfUUA@mail.gmail.com>
Subject: Re: BUG #19373: One backend hanging in AioIoUringExecution blocking
 other backends
To: michael.kroell@gmail.com, pgsql-bugs@lists.postgresql.org
Content-Type: multipart/alternative; boundary="000000000000909bc306488a44a4"
Archived-At: <https://www.postgresql.org/message-id/CAOVWO5oKxGOdaGSS5QDH8WnpNqjiZ%3DW5PjP8_Uk4b2x6%3DEfUUA%40mail.gmail.com>
Precedence: bulk

--000000000000909bc306488a44a4
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Michael,

Thank you for the detailed report.

Even though there was a *global* statement_timeout=3D61s configured, backen=
ds
> accessing the same table were hanging with ``LWLock AioUringCompletion``
>

The statement_timeout not interrupting and not erroring out looks weird,
and this part could be a postgres bug in itself.


> Restarting the cluster did not go through until the hanging leader PID wa=
s
> ``SIGKILL``ed
>
Am I understanding this correctly as "Normal shutdown (SIGTERM or pg_ctl
stop) did not complete, and postmaster remained waiting on
until AioIoUringExecution was force killed" ?

I=E2=80=99m interested in digging into this and am wondering about the belo=
w
1. What filesystem and storage was this instance running on
2. Was this a parallel sequential scan, was any index access involved?
3. By any chance do you have a reproducible test case?
4. Can you share what shared_preload_libraries you are using?

Regards,
Surya Poondla

--000000000000909bc306488a44a4
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr"><div dir=3D"ltr">Hi Michael,<br><br>Thank=
 you for the detailed report.</div><div dir=3D"ltr"><br></div></div><div cl=
ass=3D"gmail_quote gmail_quote_container"><blockquote class=3D"gmail_quote"=
 style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);p=
adding-left:1ex">
Even though there was a *global* statement_timeout=3D61s configured, backen=
ds<br>
accessing the same table were hanging with ``LWLock AioUringCompletion``<br=
></blockquote><div><br></div><div>The statement_timeout not interrupting an=
d not erroring out looks weird, and this part could be=C2=A0a postgres bug =
in itself.</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D=
"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-le=
ft:1ex">
Restarting the cluster did not go through until the hanging leader PID was<=
br>
``SIGKILL``ed<br></blockquote><div>Am I understanding this correctly as &qu=
ot;Normal shutdown (SIGTERM or pg_ctl stop) did not complete, and postmaste=
r remained waiting on until=C2=A0AioIoUringExecution=C2=A0was force killed&=
quot; ?<br><br>I=E2=80=99m interested in digging into this and am wondering=
 about the below</div><div>1. What=C2=A0filesystem and storage was this ins=
tance running on<br>2.=C2=A0Was this a parallel sequential scan, was any in=
dex access involved?</div><div>3. By any chance do you have a reproducible =
test case?<br>4. Can you share what=C2=A0shared_preload_libraries you are u=
sing?<br><br>Regards,<br>Surya Poondla</div></div></div>

--000000000000909bc306488a44a4--