Hi Justin,
Thank you for reviewing.
I have tried to run the below query and could see only less than 5 connections active when I get this error. The total rows I see is only 10 including idle and active sessions for this output.
select pid as process_id,
usename as username,
datname as database_name,
client_addr as client_address,
application_name,
backend_start,
state,
state_change
from pg_stat_activity;
Regards,
Guna
On Wed, May 11, 2022 at 12:59:01AM +0800, Sudhir Guna wrote:
> Dear All,
>
> We have recently upgraded Postgresql 9.4 standalone server to Postgresql
> 11.2 with High Availability (2 servers : Master and Standby).
>
> While trying to test using ETL applications and reports, we observe that
> the ETL jobs fails with below error,
>
> 2022/05/06 16:27:36 - Error occurred while trying to connect to the database
> 2022/05/06 16:27:36 - Error connecting to database: (using class
> org.postgresql.Driver)
> 2022/05/06 16:27:36 - FATAL: Sorry, too many clients already
>
> We have increased the max_connections = 1000 in postgresql.conf file.
>
> It worked ok for a day and later we get the same error message.
>
> Please help to advise on any additional settings required. The prior
> Postgresql 9.4 had the default max_connections = 100 and the applications
> worked fine.
It sounds like at least one thing is still running, perhaps running very
slowly.
You should monitor the number of connections to figure out what.
If you expect to be able to run with only 100 connections, then when
connections>200, there's already over 100 connections which shouldn't still be
there.
You could query pg_stat_activity to determine what they're doing - trying to
run a slow query ? Are all/most of them stuck doing the same thing ?
You should try to provide the information here for the slow query, and for the
rest of your environment.
https://wiki.postgresql.org/wiki/Slow_Query_Questions
--
Justin