Observed Signal · May 24, 2026 · Technical Guide · Source: DEV Community · Impact: 2/5 · Sentiment: Positive
PHP-FPM Tuning: 5 Settings That Decide p99
A technical guide explains five php-fpm configuration settings that heavily influence p99 latency for PHP web apps: pm.max_children, pm.max_requests, request_terminate_timeout, pm.process_idle_timeout (ondemand only), and listen.backlog. The author explains how to calculate pm.max_children from available RAM and average worker RSS, why pm.max_requests protects against memory growth, why request_terminate_timeout is the reliable kill switch, when ondemand is appropriate, and how listen.backlog interacts with the kernel (net.core.somaxconn) during bursts. A real-world c5.large Laravel 12 tuning example shows p99 improving from ~4800ms to ~380ms after switching to static pools, lowering max_children and increasing backlog and recycle settings. The article also summarizes when to consider Laravel Octane as an architectural alternative after tuning.
Practical server-side tuning guidance can materially reduce p99 and prevent OOMs for web applications, benefiting engineering teams running customer‑facing services, but this is operational best-practice rather than industry-shifting news.
Track Amazon Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Article lists five php-fpm settings: pm.max_children, pm.max_requests, request_terminate_timeout, pm.process_idle_timeout, listen.backlog.
- Recommendation: set pm to static for single-purpose API servers; compute pm.max_children = (total_RAM - reserved_RAM) / avg_worker_RSS.
- Example c5.large (4 GB RAM, 2 vCPU) tuning: switching from pm.dynamic (pm.max_children=50) to pm.static with pm.max_children≈20, pm.max_requests=800, request_terminate_timeout=35s and listen.backlog=8192 reduced p99 from ~4800ms to ~380ms and eliminated OOMs.
- listen.backlog should be raised (examples: 4096 or 8192) but is capped by kernel net.core.somaxconn; adjust kernel sysctl and restart php-fpm so socket picks up new backlog.
- Advice: set pm.max_requests (e.g., 500–1000) to recycle workers and mitigate memory growth; pm.process_idle_timeout is only relevant for pm=ondemand.
Connected Companies & Entities
2 Entities mappedOntology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Tuning PHP-FPM for Laravel Performance
This technical guide explains how to tune PHP-FPM for Laravel applications, covering process manager modes (static, dynamic, ondemand), how PHP-FPM worker processes handle requests, and the key settings that determine throughput and stability. It provides a formula for calculating pm.max_children from available RAM and average worker memory, guidance on PHP memory_limit, pm.max_requests for process recycling, request timeouts, and slow-log configuration for diagnosing slow requests. The article includes recommended example configurations for 1GB, 2GB, 4GB and 8GB servers (Deploynix-managed), monitoring tips using the PHP-FPM status page and Deploynix integration, and validation/restart/test commands (php-fpm8.4 -t, systemctl restart, wrk/ab). Practical recommendations emphasize measuring real worker memory, leaving headroom, and iterating based on observability metrics.
Laravel Queue Stopped — Root Cause Was PHP-FPM
A developer debugging a Laravel application on a cPanel server discovered that recurring queue stoppages and Make.com request timeouts were caused by infrastructure issues rather than the queue itself. The queue worker had been started with nohup and disappeared after server restarts; moving execution to a cPanel Cron job with /bin/flock and --stop-when-empty made the worker lifecycle more resilient. Separately, slow or queued HTTP requests were traced to an overly conservative PHP-FPM configuration (pm.max_children = 5, pm.max_requests = 20, pm.process_idle_timeout = 10) on a 65 GB / 20-core server. The author recommends tuning PHP-FPM to match server capacity and diagnosing each layer independently.
Latency Percentiles Explained for Node.js Apps
This technical article explains what latency percentiles (p50, p95, p99, p99.9) represent for user experience and why averages can hide bad experiences. It highlights that Node.js' single-threaded event loop amplifies tail (p99) latency, provides an Express middleware example to measure runtime percentiles, and gives recommended operational practices: set timeouts, size connection pools, and design retry/backoff around p99 rather than p50. The author includes a table of realistic p50/p95/p99 numbers for common dependencies (Postgres, Redis, MongoDB, S3, Stripe, OpenAI) and links a utility package, slowdep, to simulate production-like latency distributions for testing.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.