Issue nginx "worker_connections are not enough" causing HTTP 500 error pages

[Azurel]

Server operating system version

AlmaLinux 9.8

Plesk version and microupdate number

18.0.80#5

Hi, I'm seeing thousands of nginx errors in proxy_error_log like:

[alert] 8372#0: *40621122 1024 worker_connections are not enough while connecting to upstream, client: ....

This is new. Before upgrading from AlmaLinux 8.10 to 9.8, users never reported this kind of issue.
The server runs AlmaLinux 9.8 with Plesk Obsidian 18.0.80 #5 and has 24 CPU cores.

Current nginx settings:

/etc/nginx/nginx.conf: worker_processes 1;
/etc/nginx/nginx.conf: worker_connections 1024;
/etc/nginx/ulimit.global_params: worker_rlimit_nofile 40000;

/etc/nginx/ulimit.global_params is from from 2020-12-15
/etc/nginx/nginx.conf is dated 2026-08-10. This date is a bit strange, as there was no system or Plesk update on that day.

# systemctl show nginx -p LimitNOFILE
LimitNOFILE=4096

# grep -i "Max open files" /proc/$(cat/run/nginx.pid)/limits
Max open files            4096                 4096                 files

What is the Plesk-supported and update-safe way to increase nginx capacity here?
In particular:

• Is worker_processes 1 expected on a 24-core server?
• Should worker_processes be set to auto?
• Should worker_connections be increased?
• Should LimitNOFILE also be increased?

Thanks.

 

[Azurel]

Additional information:

# stat /etc/nginx/nginx.conf
Modify: 2026-08-10 14:58:55.000000000 +0200
Change: 2026-08-15 03:46:22.269847289 +0200
Birth:  2026-08-05 14:20:54.922420007 +0200

The installed package is:

# rpm -qf /etc/nginx/nginx.conf
sw-nginx-1.30.4.1-2.redhat.9+p18.0.80.3+t260813.2046.x86_64

It was installed on August 15:

# rpm -q --qf '%{NAME} %{VERSION}-%{RELEASE} installed: %{INSTALLTIME:date}\n' \
  $(rpm -qf /etc/nginx/nginx.conf)

sw-nginx 1.30.4.1-2.redhat.9+p18.0.80.3+t260813.2046 installed: Sat 15 Aug 2026 03:46:22 AM CEST
Also, rpm -V reports no modification of /etc/nginx/nginx.conf, so the current file matches the one shipped by the Plesk sw-nginx package.

Timeline:

• 2026-08-05: AlmaLinux 8 → 9 upgrade
• 2026-08-15: Plesk update to 18.0.80 #3 / installation of the above sw-nginx package

I found an additional important detail.

There is an /etc/nginx/nginx.conf.rpmsave created during the AlmaLinux 8 → 9 upgrade on August 5:

Modify: 2020-08-08 18:07:47
Change: 2026-08-05 13:56:27

My previous nginx configuration contained "worker_processes auto;" not "worker_processes 1;"

So during the AlmaLinux 8 → 9 upgrade, my nginx.conf was saved as "nginx.conf.rpmsave" and the package-provided configuration restored worker_processes 1;. This would also explain why the worker_connections are not enough issue only appeared after the OS upgrade.

How can I set this to "worker_processes auto;" so that some Plesk update doesn't change it again?