Automated Catalogue Requests and PHP-FPM Saturation: A 17,000-Product WooCommerce Case Study
This technical report documents a production investigation into PHP-FPM saturation, unusual automated catalogue requests and targeted server-side protection on a managed VPS hosting large WooCommerce websites. The investigation involved a WooCommerce catalogue containing approximately 17,000 products, alongside other WordPress installations sharing the same server infrastructure. The initial investigation established genuine server capacity and availability problems: PHP-FPM saturation: Two affected WooCommerce pools reached their configured limits of five child processes each. Request backlog: A listen queue of 129 requests was observed against a configured backlog length of 128. FastCGI timeouts: Requests encountered waits reaching 300 seconds. Service disruption: HTTP 503 and 504 responses occurred across affected websites. Suspicious traffic: A later one-hour sample identified 137 unusual cart/pagination requests originating from 136 distinct IP addresses. The investigation examined automated-looking requests combining WooCommerce add-to-cart operations with deep product pagination, alongside other potentially expensive dynamic requests. Targeted ModSecurity restrictions were introduced to reject selected unwanted request patterns before PHP execution, while preserving legitimate public product discovery. A controlled verification request returned HTTP 403 in approximately 14 milliseconds. Additional corrective work resolved conflicting Apache ErrorDocument directives associated with internal redirect loops (AH00124). Subsequent observations recorded an empty PHP-FPM request queue, a successful HTTP 200 homepage response and no new AH00124 errors during the sampled verification interval. The report distinguishes confirmed server saturation, observed automation patterns, implemented protections and verified individual test outcomes from unproven causal explanations. Importantly, the evidence does not establish that the 137 suspicious requests caused the original PHP-FPM exhaustion, that those requests originated from verified AI crawlers, or that the interventions produced a quantified, sustained improvement in whole-server performance. The investigation presents an evidence-based engineering methodology for diagnosing expensive WooCommerce request patterns, managing PHP-FPM capacity, applying narrowly scoped security controls and preserving machine-readable catalogue accessibility. It also identifies the additional logging, request-cost measurements and controlled comparisons required for stronger future causal attribution. Author: Keith Rowley, Sydney Business Web. Publication: Engineering Research Report 11, Version 1.0.0, 14 November 2026.
Authors
- Keith Rowley
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24213061
- Primary Topic
- Web Application Security Vulnerabilities
- Type
- article
- Field-Weighted Citation Impact
- 0.00