MIME-Version: 1.0
References: <DDPQ1RV5FE9U.I2WW34NGRD8Z@jeltef.nl>
 <DFJ3E1X9169E.WV892I950LYW@jeltef.nl>
 <DFY0RHAZTTXJ.15AD95RE6A5N5@jeltef.nl>
 <CAN4CZFOw2v9G_9vuP0vnNH8K9X_UeSOyq=gWJp7bdGZ5HKNTDA@mail.gmail.com>
 <DG9X4XC66EMC.2S34SS1SZ8K3K@jeltef.nl> <DGBMBCDXXNBG.3CNHEZ9QVB4GL@jeltef.nl>
 <CAN4CZFMLpakQb8qohDAnrOoQ1-UrvmzJewZss95Y+Mf_J+iNeA@mail.gmail.com>
 <DH1JZEYZOBSW.1QEY6MI45FCIL@jeltef.nl>
 <f97528e4-5766-415e-b998-678da925ce05@vondra.me>
In-Reply-To: <f97528e4-5766-415e-b998-678da925ce05@vondra.me>
From: Jim Nasby <jnasby@upgrade.com>
Date: Mon, 23 Mar 2026 17:34:35 -0500
Message-ID: 
 <CAMFBP2rSvyj6DYAw0EEBGR2UUP85ByKAC_D0djJm7rdypuMvdQ@mail.gmail.com>
Subject: Re: Add GoAway protocol message for graceful but fast server
 shutdown/switchover
To: Tomas Vondra <tomas@vondra.me>
Cc: Jelte Fennema-Nio <postgres@jeltef.nl>,
 Zsolt Parragi <zsolt.parragi@percona.com>,
	PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>,
 Dave Cramer <davecramer@gmail.com>,
	Jacob Champion <jacob.champion@enterprisedb.com>,
 Heikki Linnakangas <hlinnaka@iki.fi>
Content-Type: multipart/alternative; boundary="0000000000005c2977064db8a485"
Archived-At: 
 <https://www.postgresql.org/message-id/CAMFBP2rSvyj6DYAw0EEBGR2UUP85ByKAC_D0djJm7rdypuMvdQ%40mail.gmail.com>
Precedence: bulk

--0000000000005c2977064db8a485
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Mar 20, 2026 at 2:20=E2=80=AFPM Tomas Vondra <tomas@vondra.me> wrot=
e:

> * Does it make sense to tie this to smart shutdowns? I realize it's just
> an example, and it probably makes sense to send the GoAway message
> before a shutdown. But isn't this a bit similar to cancel/terminate of a
> backend? Why not to have a pg_goaway_backend() function, that'd send the
> message to a single backend? It might be useful for load-balancing, if
> we could pick a "heavy" backend and ask it to reconnect / move to a
> different replica. (Could that be handled by a middleware?)
>

+1. Another scenario that comes to mind is asking for a reconnect based on
backend memory consumption, since there's a bunch of internal structures
(relcache, etc) that can grow in an unbounded fashion.

--0000000000005c2977064db8a485
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr">On Fri, Mar 20, 2026 at 2:20=E2=80=AFPM T=
omas Vondra &lt;<a href=3D"mailto:tomas@vondra.me">tomas@vondra.me</a>&gt; =
wrote:</div><div class=3D"gmail_quote gmail_quote_container"><blockquote cl=
ass=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid=
 rgb(204,204,204);padding-left:1ex">* Does it make sense to tie this to sma=
rt shutdowns? I realize it&#39;s just<br>
an example, and it probably makes sense to send the GoAway message<br>
before a shutdown. But isn&#39;t this a bit similar to cancel/terminate of =
a<br>
backend? Why not to have a pg_goaway_backend() function, that&#39;d send th=
e<br>
message to a single backend? It might be useful for load-balancing, if<br>
we could pick a &quot;heavy&quot; backend and ask it to reconnect / move to=
 a<br>
different replica. (Could that be handled by a middleware?)<br></blockquote=
><div><br></div><div>+1. Another scenario that comes to mind is asking for =
a reconnect based on backend memory consumption, since there&#39;s a bunch =
of internal structures (relcache, etc) that can grow in an unbounded fashio=
n.=C2=A0</div></div></div>

--0000000000005c2977064db8a485--