Wow, this is so pleasant now! I can actually see users. Before it was just pages and pages of guests
Apologies if sucking eggs etc, feel free to ignore/delete, sorry.
Those IPs are Cloudflare, so presumably nginx.conf needs set_real_ip_from / real_ip_header.
In /etc/nginx/conf.d/50-cloudflare.conf or wherever for you:
set_real_ip_from 103.21.244.0/22;
set_real_ip_from 103.22.200.0/22;
set_real_ip_from 103.31.4.0/22;
set_real_ip_from 104.16.0.0/12;
set_real_ip_from 108.162.192.0/18;
set_real_ip_from 131.0.72.0/22;
set_real_ip_from 141.101.64.0/18;
set_real_ip_from 162.158.0.0/15;
set_real_ip_from 172.64.0.0/13;
set_real_ip_from 173.245.48.0/20;
set_real_ip_from 188.114.96.0/20;
set_real_ip_from 190.93.240.0/20;
set_real_ip_from 197.234.240.0/22;
set_real_ip_from 198.41.128.0/17;
set_real_ip_from 199.27.128.0/21;
set_real_ip_from 2400:cb00::/32;
set_real_ip_from 2606:4700::/32;
set_real_ip_from 2803:f800::/32;
set_real_ip_from 2405:b500::/32;
set_real_ip_from 2405:8100::/32;
set_real_ip_from 2c0f:f248::/32;
set_real_ip_from 2a06:98c0::/29;
real_ip_header CF-Connecting-IP;
That would also very likely explain the seemingly repetitive slightly odd Crowdsec messages mentioning "
traffic coming from your IP range" - because it's always seeing the exact same set of source IPs which are passed through by nginx.
Excessive 404s can also sometimes cause high php-fpm use, so IIRC there's a way to prevent those from ever being sent to PHP. But not necessarily needed when CF is doing its thang.
Check php-fpm settings are suited to whatever usuable memory it has.
The two web servers also still need 80+443 firewall rules, to block non-CF traffic. This is assuming you don't have other things also proxying traffic & don't accidentally block SSH or whatever.
Usually - not always, obv - a good CF
custom security rule, applying a managed challenge to almost every path bar a few, works fine. 'Under attack' mode isn't usually necessary. UA mode is just a forceful way of applying challenges, without the flexibility of doing the same via custom security rules.
Scrapers using Bright Data (aka Luminati) residential proxies & bypassing CF were my most recent foes. Sometimes they use request consistent predictable paths & can be tarpitted/similar, but on a high-traffic site I imagine they could be awkward to spot in logs.