Longpolling issue - Multiple Docker instances on server with multiple workers for each instance

I have multiple dockers installed on my server with each docker having multiple workers. The issue is my longpolling doesn`t work.

What should I change in my setup to make it work?

The image below shows all of the dockers running.

Below is my Nginx config file for one of the sites....

server {
	root /var/www/html;
	index index.html index.htm index.nginx-debian.html;
	server_name website.co.za www.website.co.za;

	location / {
		proxy_read_timeout 720s;
        proxy_connect_timeout 720s;
        proxy_send_timeout 720s;
		proxy_pass http://127.0.0.1:8069;
		proxy_redirect off;
		proxy_set_header Host $host;
		proxy_set_header X-Real-IP $remote_addr;
		proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
		proxy_set_header X-Forwarded-Proto $scheme;
	}

    listen [::]:443 ssl ipv6only=on; # managed by Certbot
    listen 443 ssl; # managed by Certbot
    ssl_certificate /etc/letsencrypt/live/www.website.co.za/fullchain.pem; # managed by Certbot
    ssl_certificate_key /etc/letsencrypt/live/www.website.co.za/privkey.pem; # managed by Certbot
    include /etc/letsencrypt/options-ssl-nginx.conf; # managed by Certbot
    ssl_dhparam /etc/letsencrypt/ssl-dhparams.pem; # managed by Certbot

    add_header Strict-Transport-Security "max-age=31536000" always; # managed by Certbot

}

server {
	listen 80;
	listen [::]:80;
	root /var/www/html;

	index index.html index.htm index.nginx-debian.html;
	server_name website.co.za www.website.co.za;

	location / {
		proxy_read_timeout 720s;
        proxy_connect_timeout 720s;
        proxy_send_timeout 720s;
		proxy_pass http://127.0.0.1:8069;
		proxy_redirect off;
		proxy_set_header Host $host;
		proxy_set_header X-Real-IP $remote_addr;
		proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
		proxy_set_header X-Forwarded-Proto $scheme;
	}
}

Start with the posted Nginx server block and the Docker setup containing multiple instances and workers. Trace how longpolling requests are routed for each site and compare that with the container and worker configuration. Done means identifying and documenting a setup change that makes longpolling work across the shown instances.