Optimizing Nginx + PHP‑FPM: Architecture, Configurations, and 5XX Error Fixes

1. Relationship and Division of Nginx and PHP‑FPM

Nginx is a web server that handles static files, while PHP‑FPM is a PHP FastCGI process manager; they communicate via the FastCGI protocol. This separation prevents slow or faulty PHP scripts from blocking Nginx, preserving high‑concurrency handling.

2. Nginx Working Overview

When a PHP request arrives, Nginx forwards it to a PHP‑FPM worker using fastcgi_pass. During this exchange Nginx applies various timeout, buffer, cache, and keep‑alive mechanisms to ensure efficient cooperation.

Timeout directives : fastcgi_connect_timeout – time to connect to the backend. fastcgi_send_timeout – time to send data. fastcgi_read_timeout – time to receive data.

Buffer directives : fastcgi_buffer_size – size of the response header buffer. fastcgi_buffers – size and number of buffers for response body (e.g., 8 4k).

3. PHP‑FPM Working Overview

PHP‑FPM runs a master process and multiple worker processes. The master listens on a port, accepts connections, and distributes tasks to workers. Workers execute PHP scripts and return results to Nginx via FastCGI.

PHP‑FPM can manage multiple pools; each pool can be static, dynamic, or ondemand. Example configuration:

pm = static|dynamic|ondemand

pm.max_children = 256

pm.max_requests = 1024

In the author’s production environment (12‑core CPU, 16 GB RAM) they run 12 Nginx workers and 256 PHP workers, each consuming ~30 MB, totaling about 8 GB of RAM, achieving sub‑100 ms response times for >90 % of requests.

4. Common 5XX Errors in This Model

4.1 502 Bad Gateway

Typical causes:

PHP‑FPM worker exceeds its execution timeout; the master kills the worker and Nginx returns 502. Relevant settings: max_execution_time in php.ini (default 30 s) and request_terminate_timeout in PHP‑FPM.

Connection backlog is full, preventing new connections from reaching PHP‑FPM, leading to 502. Adjust backlog or system net.core.somaxconn (default 128).

Client disconnects (Nginx logs 499) and PHP later aborts, causing a subsequent 502.

Optimization suggestions: increase PHP‑FPM worker count, raise execution timeout, and set backlog to 1–2 × the number of workers.

4.2 504 Gateway Timeout

Typical causes:

PHP‑FPM workers process requests too slowly, causing Nginx’s accept queue to wait beyond its timeout.

PHP script execution exceeds Nginx’s FastCGI timeout settings.

Client network is poor; Nginx cannot deliver the response within the timeout.

Optimization suggestions: tune Nginx timeout directives ( fastcgi_connect_timeout, fastcgi_send_timeout, fastcgi_read_timeout) and ensure adequate file descriptor and buffer sizes for high‑concurrency scenarios.

5. Illustrative Diagrams