Cloudflare Outage Explained: CEO Details How A Single Internal Error Broke The Internet For Five Hours

Cloudflare’s global network went dark on Tuesday, taking down massive parts of the internet with it. Cloudflare CEO Matthew Prince provided details regarding what went wrong during this five-hour outage that plagued several major platforms, including Cloud X, ChatGPT, Canva and others. Now a full report from Cloudflare on what went wrong can be seen below. And here’s the thing—this wasn’t a hack, a DDoS storm or any kind of external threat. It was an internal systems flaw that spiralled across Cloudflare’s infrastructure, making a bad file spread through the network and repeatedly crash core services.

Why Cloudflare Went Down

Prince described the incident as “Cloudflare’s worst outage since 2019.” The chain reaction started with a permissions change in a database system. That small tweak caused Cloudflare’s Bot Management tool to generate a feature file twice as large as the software was built to handle.

ALSO READ: Jio Makes A Big AI Push: Free 18-Month Gemini 3 Pro Plan For All Unlimited 5G Users

Once that bloated file began circulating across the global network, Cloudflare’s proxy software started failing every time it tried to read it.

How the Faulty File Spread

The problematic file came from a ClickHouse database cluster that regenerates the file every five minutes. But the cluster wasn’t uniform — some nodes were updated, others weren’t. That mismatch led to an intermittent propagation of the “bad” file. For engineers, it created a maddening loop: the network would partially fix itself, then break again minutes later.

Why It Looked Like a Cyberattack

The pattern resembled a massive DDoS attack — sudden failures across multiple regions, traffic spikes, and recovery cycles. That sent the investigation in the wrong direction initially. Once engineers traced everything back to the faulty config file, they stopped its distribution, rolled back to a stable version and rebooted the proxy services.

Core traffic stabilised by 14:30 UTC (8:00pm IST), with full recovery at 17:06 UTC (10:36pm IST).

Services Affected

The outage hit several Cloudflare products:

- CDN and security services issued widespread 5xx errors

- Turnstile bot-challenge failed to load

- Workers KV suffered high error rates

- Dashboard access broke because Turnstile was down

- Email security lost access to an IP-reputation source, reducing spam detection accuracy

ALSO READ: Qualcomm Confirms Snapdragon 8 Gen 5 Launch For November 26: Specs, Performance Leaks And Expected Phones

Prince acknowledged the seriousness, saying, “An outage like today is unacceptable,” and promised more resilient systems going forward.

Final Thoughts

Cloudflare’s breakdown shows how fragile the internet can be when a core infrastructure provider hits trouble. A single oversized file ricocheted across a global network and knocked out services used by hundreds of millions of people. Cloudflare says it’s fixing the underlying design flaw, but the outage is a reminder of just how much of the modern web depends on a few critical systems staying stable.