PHP-FPM and Nginx

PHP-FPM manages a pool of worker processes; Nginx serves static files and forwards PHP requests via FastCGI. Each worker handles one request at a time, so pm.max_children caps concurrency — size it from available memory divided by per-worker memory. Use pm = dynamic or static for steady traffic, enable the status page and slow log, and set request_terminate_timeout to kill runaway requests.

; /etc/php/8.4/fpm/pool.d/www.conf
pm = dynamic
pm.max_children = 40          ; ~ (RAM for PHP) / (avg worker RSS)
pm.start_servers = 10
pm.min_spare_servers = 5
pm.max_spare_servers = 15
pm.max_requests = 1000        ; recycle workers to contain leaks
request_slowlog_timeout = 5s
slowlog = /var/log/php-fpm/slow.log
request_terminate_timeout = 30s
pm.status_path = /fpm-status

server {
    root /var/www/app/public;
    location / { try_files $uri /index.php$is_args$args; }
    location ~ ^/index\.php(/|$) {
        fastcgi_pass unix:/run/php/php8.4-fpm.sock;
        include fastcgi_params;
        fastcgi_param SCRIPT_FILENAME $realpath_root$fastcgi_script_name;
        internal;
    }
    location ~ \.php$ { return 404; }   # never execute other PHP files
}

Common pitfalls

• Setting pm.max_children too high and running out of memory (OOM kills).
• Letting Nginx execute any .php file, including uploads.
• No slow log, making stuck requests invisible.

Key points

• max_children caps concurrency; size by memory
• Only route the front controller to PHP
• Use slow logs and status pages

Further reading: PHP manual: FPM configuration

Go deeper: PHP-FPM and the Shared-Nothing Model in the Request Lifecycle course traces this step of the request, with a production bug and a runnable simulation.