Обсуждение: pgbouncer transaction pool mode issue for prepared statements

Поиск
Список
Период
Сортировка

pgbouncer transaction pool mode issue for prepared statements

От
Durgamahesh Manne
Дата:
Hi PGDG Team 



We’re currently facing an issue with PgBouncer transaction pooling mode, where prepared statements are failing, and we are unable to modify the application code to disable or change how prepared statements are used.

Switching to session pooling mode is not feasible for us because we have a very high number of connections that we cannot fully control. Environment details: PgBouncer version: 1.25.1 PostgreSQL version: 16.11

 The specific error we’re seeing is: 

bind message has 43 result formats but query has 47 columns

ERROR: unnamed prepared statement does not exist Repeatedly logs 

 Has anyone encountered this issue with transaction pooling and prepared statements? Are there any PgBouncer parameters or settings we can use to mitigate this problem without requiring application‑level changes? Any guidance or solutions would be greatly appreciated.

Regards 
Durga Mahesh

Re: pgbouncer transaction pool mode issue for prepared statements

От
"Adrien Obernesser (OBA)"
Дата:
Hi, 

It looks like you have the issue of a query that is referencing the prepared statements of another session. This is expected in transaction pooling mode. Since you are using this version of pgbouncer 
try to look into the  max_prepared_statements parameter, I am not quite sure with the unnamed aspect though, this would need testing on your config. 



Cordialement,  Kind Regards, 

Adrien OBERNESSER
 



From: Durgamahesh Manne <maheshpostgres9@gmail.com>
Sent: Monday, February 16, 2026 2:24 PM
To: pgsql-general <pgsql-general@lists.postgresql.org>
Subject: pgbouncer transaction pool mode issue for prepared statements
 
Hi PGDG Team 



We’re currently facing an issue with PgBouncer transaction pooling mode, where prepared statements are failing, and we are unable to modify the application code to disable or change how prepared statements are used.

Switching to session pooling mode is not feasible for us because we have a very high number of connections that we cannot fully control. Environment details: PgBouncer version: 1.25.1 PostgreSQL version: 16.11

 The specific error we’re seeing is: 

bind message has 43 result formats but query has 47 columns

ERROR: unnamed prepared statement does not exist Repeatedly logs 

 Has anyone encountered this issue with transaction pooling and prepared statements? Are there any PgBouncer parameters or settings we can use to mitigate this problem without requiring application‑level changes? Any guidance or solutions would be greatly appreciated.

Regards 
Durga Mahesh

Re: pgbouncer transaction pool mode issue for prepared statements

От
Adrian Klaver
Дата:
On 2/16/26 05:24, Durgamahesh Manne wrote:
> Hi PGDG Team
> 
> 
> 
> We’re currently facing an issue with PgBouncer transaction pooling mode, 
> where prepared statements are failing, and we are unable to modify the 
> application code to disable or change how prepared statements are used.
> 
> Switching to session pooling mode is not feasible for us because we have 
> a very high number of connections that we cannot fully control. 
> Environment details: PgBouncer version: 1.25.1 PostgreSQL version: 16.11
> 
>   The specific error we’re seeing is:
> 
> bind message has 43 result formats but query has 47 columns
> 
> ERROR: unnamed prepared statement does not exist Repeatedly logs
> 
>   Has anyone encountered this issue with transaction pooling and 
> prepared statements? Are there any PgBouncer parameters or settings we 
> can use to mitigate this problem without requiring application‑level 
> changes? Any guidance or solutions would be greatly appreciated.

A little exploration/searching would have revealed:

https://www.pgbouncer.org/faq.html

5), How to use prepared statements with transaction pooling?

Which leads to:

https://www.pgbouncer.org/faq.html#how-to-use-prepared-statements-with-transaction-pooling

which in turn leads to:

https://www.pgbouncer.org/config.html#max_prepared_statements

Note though the caveat in last link:

"Note: This tracking and rewriting of prepared statement commands does 
not work for SQL-level prepared statement commands, so PREPARE, EXECUTE 
and DEALLOCATE are forwarded straight to Postgres. The exception to this 
rule are the DEALLOCATE ALL and DISCARD ALL commands, these do work as 
expected and will clear the prepared statements that PgBouncer tracked 
for the client that sends this command."


> 
> Regards
> Durga Mahesh


-- 
Adrian Klaver
adrian.klaver@aklaver.com



Re: pgbouncer transaction pool mode issue for prepared statements

От
Durgamahesh Manne
Дата:


On Mon, 16 Feb, 2026, 21:35 Adrian Klaver, <adrian.klaver@aklaver.com> wrote:
On 2/16/26 05:24, Durgamahesh Manne wrote:
> Hi PGDG Team
>
>
>
> We’re currently facing an issue with PgBouncer transaction pooling mode,
> where prepared statements are failing, and we are unable to modify the
> application code to disable or change how prepared statements are used.
>
> Switching to session pooling mode is not feasible for us because we have
> a very high number of connections that we cannot fully control.
> Environment details: PgBouncer version: 1.25.1 PostgreSQL version: 16.11
>
>   The specific error we’re seeing is:
>
> bind message has 43 result formats but query has 47 columns
>
> ERROR: unnamed prepared statement does not exist Repeatedly logs
>
>   Has anyone encountered this issue with transaction pooling and
> prepared statements? Are there any PgBouncer parameters or settings we
> can use to mitigate this problem without requiring application‑level
> changes? Any guidance or solutions would be greatly appreciated.

A little exploration/searching would have revealed:

https://www.pgbouncer.org/faq.html

5), How to use prepared statements with transaction pooling?

Which leads to:

https://www.pgbouncer.org/faq.html#how-to-use-prepared-statements-with-transaction-pooling

which in turn leads to:

https://www.pgbouncer.org/config.html#max_prepared_statements

Note though the caveat in last link:

"Note: This tracking and rewriting of prepared statement commands does
not work for SQL-level prepared statement commands, so PREPARE, EXECUTE
and DEALLOCATE are forwarded straight to Postgres. The exception to this
rule are the DEALLOCATE ALL and DISCARD ALL commands, these do work as
expected and will clear the prepared statements that PgBouncer tracked
for the client that sends this command."


>
> Regards
> Durga Mahesh


--
Adrian Klaver
adrian.klaver@aklaver.com

Hi

Thank you so much for this information 

Regards
Durga Mahesh