Message-ID: <2bdca9488627c8a0e855f00d7a1137cfdd01746e.camel@cybertec.at>
Subject: Re: Pgbouncer and Node JS application Query read timeout error
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: KK CHN <kkchn.in@gmail.com>
Cc: pgsql-general <pgsql-general@postgresql.org>
Date: Fri, 10 Apr 2026 09:29:01 +0200
In-Reply-To: <CAKgGyB_zeOcCu+U=_=1OW_iJTcBweX1crEJQfxT-dN-b1_0JdQ@mail.gmail.com>
References: 
	<CAKgGyB_OmPnsgrr7YWOY=w8VDH7pN2hCt4Tsd1nJy=-xM=64eA@mail.gmail.com>
	 <6b3cf69fa569d263a593cc2afe74dc35d29754eb.camel@cybertec.at>
	 <CAKgGyB_zeOcCu+U=_=1OW_iJTcBweX1crEJQfxT-dN-b1_0JdQ@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.58.3 (3.58.3-1.fc43) 
MIME-Version: 1.0
Archived-At: <https://www.postgresql.org/message-id/2bdca9488627c8a0e855f00d7a1137cfdd01746e.camel%40cybertec.at>
Precedence: bulk

On Fri, 2026-04-10 at 12:37 +0530, KK CHN wrote:
> > The only way I can imagine that pgBouncer is leading to timeouts on the=
 client side
> > is if client sessions are waiting, because all connections are in use.
> >=20
> > You can run SHOW POOLS in the pgBouncer console to see if there are any=
 "cl_waiting".
> > If that is the case, you should configure the Node.js pools smaller, so=
 that no
> > connection has to wait.
>=20
> Configuring Node.js=C2=A0 pools smaller ?=C2=A0 I couldn't get the=C2=A0 =
logic here=C2=A0 why advised to reduce the pool size ?=C2=A0
> =C2=A0
> Increasing pool size=C2=A0 more than 10 adversely=C2=A0affects the connec=
tion=C2=A0 establishment from
> Node.js=C2=A0 application=C2=A0?=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Since =
DB=C2=A0 is having=C2=A0 =C2=A0Pgbouncer infront=C2=A0and
> default_pool_size =3D 50 there ,=C2=A0 don't we have the freedom to incre=
ase node.js application
> pool size and it will help the query timeout ?=C2=A0 =C2=A0or any hidden =
facts involved could you elaborate ..

There are no hidden facts.

You asked if pgBouncer could lead to query timeouts.  I answered that the o=
nly way a query
could take significantly longer with pgBouncer is if the client has to wait=
 for a pooled
connections, that is, if the connection pool in pgBouncer is smaller than t=
he number of
clients that want to run a transaction concurrently.

BEFORE you take any action, you should figure out if that is the case by lo=
oking
at SHOW POOLS.  Since you didn't report any findings, I assume that you lau=
nched your
volley of questions before doing the required investigation.  Bad idea.

IF (and only if) there are indeed client sessions waiting for a pooled conn=
ection,
you could react in two different ways:

- increase the pgBouncer pool size
- reduce the application server pool size

Obvious, isn't it?

Increasing the pgBouncer pool size is a good idea only if your database mac=
hine can
stomach the additional load.  If not, reducing the application server pool =
size is
the correct measure.  If the application load is too much for the database,=
 somebody
has to suffer.  If you choose to overload the database, everybody will be u=
nhappy.
If you throttle the load in the connection pool, only some client will be u=
nhappy
(those that have to wait and get a timeout).

You will have to diagnose what the exact problem is.

Yours,
Laurenz Albe