MIME-Version: 1.0
References: 
 <CAA5-nLCdhiJW=zEBexw=NeDGirNGdKVWb-TEhRALM6+DCDRs6w@mail.gmail.com>
 <44166282b3516f33696f025ad9f5271244f78107.camel@cybertec.at>
 <CAA5-nLB9x9SW4KoPcukxAMj9N=wARJLHEBfkkGW+0Hh2FarnPA@mail.gmail.com>
 <CAA5-nLBjwDeXTbsMxL-fwbfQ8asoCG0uG6hCMzPt2MLysdwFwQ@mail.gmail.com>
 <f393ed522208b1253be393eaf7c66b4637e4d059.camel@cybertec.at>
In-Reply-To: <f393ed522208b1253be393eaf7c66b4637e4d059.camel@cybertec.at>
From: Fabrice Chapuis <fabrice636861@gmail.com>
Date: Tue, 3 Mar 2026 09:27:29 +0100
Message-ID: 
 <CAA5-nLB-grG6pcnx0q-Sd5X+fNzw2ymGkpm8GHzXOuxyQdz4EQ@mail.gmail.com>
Subject: Re: problem during postgres restore: max_connections = 100 is a lower
 setting than on the primary server, where its value was 350
To: Laurenz Albe <laurenz.albe@cybertec.at>
Cc: Pgsql-admin <pgsql-admin@lists.postgresql.org>
Content-Type: multipart/alternative; boundary="0000000000000fbdf2064c1a7aff"
Archived-At: 
 <https://www.postgresql.org/message-id/CAA5-nLB-grG6pcnx0q-Sd5X%2BfNzw2ymGkpm8GHzXOuxyQdz4EQ%40mail.gmail.com>
Precedence: bulk

--0000000000000fbdf2064c1a7aff
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

pooling is a good point.

Regards,

Fabrice

On Mon, Mar 2, 2026 at 6:42=E2=80=AFPM Laurenz Albe <laurenz.albe@cybertec.=
at>
wrote:

> On Mon, 2026-03-02 at 17:13 +0100, Fabrice Chapuis wrote:
> > I think the value of max_connections must be aligned with primary becau=
se
> > the goal is to maintain this sizing in case of a failover and promotion
> of the standby.
>
> Not every standby is for failover.
> The technical reason is that the process array on the standby has to be a=
t
> least
> as big as on the primary (if you are setting "hot_standby =3D on").
>
> > Why to be conservative with `max_connection` and not with shared buffer=
?
> > If we're performing recovery on a machine with significantly fewer CPU
> and RAM
> > resources than the original server, lowering these parameters could be
> an option
> > because they reserve memory at starting.
>
> Perhaps, but you cannot do it.  If that is a requirement, use a connectio=
n
> pool.
>
> Yours,
> Laurenz Albe
>

--0000000000000fbdf2064c1a7aff
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">pooling is a good point.=C2=A0<div><br></div><div>Regards,=
</div><div><br></div><div>Fabrice</div></div><br><div class=3D"gmail_quote =
gmail_quote_container"><div dir=3D"ltr" class=3D"gmail_attr">On Mon, Mar 2,=
 2026 at 6:42=E2=80=AFPM Laurenz Albe &lt;<a href=3D"mailto:laurenz.albe@cy=
bertec.at">laurenz.albe@cybertec.at</a>&gt; wrote:<br></div><blockquote cla=
ss=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid =
rgb(204,204,204);padding-left:1ex">On Mon, 2026-03-02 at 17:13 +0100, Fabri=
ce Chapuis wrote:<br>
&gt; I think the value of max_connections must be aligned with primary beca=
use<br>
&gt; the goal is to maintain this sizing in case of a failover and promotio=
n of the standby.<br>
<br>
Not every standby is for failover.<br>
The technical reason is that the process array on the standby has to be at =
least<br>
as big as on the primary (if you are setting &quot;hot_standby =3D on&quot;=
).<br>
<br>
&gt; Why to be conservative with `max_connection` and not with shared buffe=
r?<br>
&gt; If we&#39;re performing recovery on a machine with significantly fewer=
 CPU and RAM<br>
&gt; resources than the original server, lowering these parameters could be=
 an option<br>
&gt; because they reserve memory at starting.<br>
<br>
Perhaps, but you cannot do it.=C2=A0 If that is a requirement, use a connec=
tion pool.<br>
<br>
Yours,<br>
Laurenz Albe<br>
</blockquote></div>

--0000000000000fbdf2064c1a7aff--