Cloudflare Outage – What Went Wrong on November 18, 2025

On November 18, 2025, Cloudflare experienced a major global outage that caused thousands of websites and online services to become unreachable. Users were met with widespread 5xx server errors, cutting off access to platforms such as chat systems, social media sites, productivity tools, and various business applications that rely on Cloudflare’s infrastructure.

The issue originated from Cloudflare’s internal systems rather than an external attack. A change in permissions within one of their internal databases caused a query to return duplicate data. This resulted in the creation of a feature configuration file that grew far larger than intended. The oversized file was then automatically distributed across Cloudflare’s global network.

This configuration file plays a key role in Cloudflare’s core proxy systems. Once the file exceeded the expected size, it triggered software crashes within the proxy processes responsible for routing internet traffic. As the faulty version propagated, more systems began to fail, creating a cascade effect that slowed or halted traffic across many Cloudflare-dependent services.

Recovery required halting the rollout of the problematic configuration file, reverting to a known stable version, and restarting affected proxy services worldwide. Most traffic returned to normal within a few hours, and full restoration was achieved later the same day after Cloudflare engineers confirmed global stability.

The incident highlighted the enormous influence of centralized infrastructure providers. A small internal error was enough to disrupt essential services across the internet. It also underscored the need for stronger safeguards against cascading failures, such as strict size validation for configuration files, quicker kill-switch mechanisms, and more isolated failure boundaries within core routing systems.

Cloudflare acknowledged the scope of the disruption and committed to implementing new preventative measures to ensure that a similar technical failure cannot propagate so widely across its network in the future.

References and Sources: blog.cloudflare.com, theverge, reuters