Another fail in Internet
Cloudflare Outage: When a Single Glitch Shakes the Internet
When Internet was created (1969, it’s named Arpanet) the goal was to build a network that could continue to function even if many of its nodes, the transmission stations let’s say, failed. The Internet Protocol (IP) and the Packet Transmission were two building blocks of that architecture.
It seems it’s so no longer.
Three days ago we have experienced another fail in Internet, after the Azure (Microsoft Cloud) and AWS (Amazon Cloud) ones.
The recent widespread outage at Cloudflare on November 18, 2025, served as a potent reminder of the fragility of the internet’s centralized infrastructure. When a key “gatekeeper” stumbles, millions of websites and services worldwide can be temporarily taken offline.
What is Cloudflare and Why It Matters
Cloudflare, Inc., is an American cloud services and cybersecurity company that acts as an essential, high-speed middle layer between a website’s host server and its end users. It provides critical services that form a key part of the modern internet’s nervous system.
Cloudflare’s importance lies in its ability to deliver security, performance, and reliability at scale, leveraging its massive global network of data centers.
Content Delivery Network (CDN): Cloudflare caches (stores copies of) website data on its global network. When a user requests a site, the data is served from the closest Cloudflare server, dramatically reducing latency and making the website load faster.
DDoS Protection: It acts as a shield, protecting websites from Distributed Denial-of-Service (DDoS) attacks by filtering out malicious traffic before it can overwhelm the origin server.
Security & Web Application Firewall: It provides robust security, including firewalls, to block threats like bots, spammers, and malicious requests.
DNS Services: It runs one of the world’s most popular and critical Domain Name System (DNS) resolvers, functioning as the “phone book” that directs internet traffic to the correct digital addresses.
With estimates suggesting that Cloudflare provides services to roughly one in five of the world’s websites, an internal failure at the company can cause a massive domino effect across the digital world.
What Happened During the November 18, 2025 Outage
The outage, which began around 11:20 UTC on November 18, 2025, was not the result of a cyberattack but rather an internal system degradation triggered by a subtle, cascading configuration error.
The Root Cause: The incident was traced back to a latent bug related to the Bot Management systemconfiguration. A change to database permissions on a ClickHouse database cluster—part of a security improvement update—caused an underlying query to suddenly return duplicate rows when generating a “feature configuration file.”
The Crash: This resulting configuration file became far larger than anticipated. When the new, oversized file was rapidly propagated across Cloudflare’s global network, it exceeded a hard-coded limit within the core proxy software that handles customer traffic. This triggered the software to “panic” and crash, resulting in a flood of HTTP 5xx errors (Internal Server Error) for millions of end-users attempting to access customer websites.
The Effect: Cloudflare’s system experienced unusual behavior, with traffic swinging between “good” and “bad” configurations every five minutes as different versions of the faulty file were generated and propagated. Engineers resolved the issue by halting the generation of the bad file and rolling back the configuration. The core impact was resolved after about three hours, with full recovery taking longer.
Because Cloudflare sits in the critical path for traffic flow, its failure resulted in millions of users being unable to access a wide swath of the internet. Key websites and services that were visibly affected by HTTP 5xx errors were X, the main AI Platforms like ChatGPT and Gemini, entertainment services like Spotify and League of Legends, Uber, Canva.
Last 5 Years Major Internet Infrastructure Outages
The November 2025 Cloudflare outage is the latest example of a single-point-of-failure event at a major infrastructure provider. Here is a list of significant, widespread outages involving major internet backbones and services in the last five years:
Nov 2025 - Cloudflare; Bug in Bot Management configuration file size
Oct 2025 - Amazon Web Services (AWS); Unspecified malfunction in the US-EAST-1 cluster
Sep 2025 - Cloudflare; Dashboard and API outage due to internal error (did not affect cached content delivery)
Oct 2024 - Verizon; Major routing issue/technical problem: Widespread internet connectivity issues across parts of the US
Mar 2024 - CrowdStrike; Faulty update in its security software: Services from airlines to healthcare and finance involved (via Windows)
Oct 2021 - Meta (Facebook); Faulty configuration change to network backbone
July 2021 - Akamai: Bug in configuration deployment (triggered by a customer update): Global websites like FedEx, PlayStation Network, Fidelity involved
June 2021 - Fastly; Software bug triggered by a single customer configuration: Major news sites (NYT, CNN, BBC), Reddit, Twitch, and Amazon involved
Nov 2020 - AWS; Networking and connectivity issues in US-EAST-1 region: Hundreds of major sites and services including Adobe, Roku, and major banking applications involved
Jun 2019 - Cloudflare; Faulty Border Gateway Protocol (BGP) announcement (related to Verizon): Amazon, Google, Facebook, Discord, and many other sites involved
Source Links for Cloudflare Outage Coverage
These sources provide initial reporting, technical analysis, and the official statement from Cloudflare regarding the incident (provided from Gemini AI):
1. Official Technical Post-Mortem
Cloudflare Blog: “Cloudflare outage on November 18, 2025”
Why it’s important: This is the primary source, providing the definitive technical explanation for the outage (a latent bug in the Bot Management configuration file generation leading to a system crash).
Link: Cloudflare
2. Major Technology News Coverage
TechRadar: “A major Cloudflare outage took down large parts of the internet - X, ChatGPT and more were affected, but all recovered now”
Why it’s important: Provides a live-blog format of how the incident unfolded, confirming the affected services in real-time.
Link: Techradar
Tom’s Hardware: “Cloudflare outage hints at nefarious ‘traffic spike’ as service impact spreads — X, McDonalds go down, company issues statement amidst chaos”
Why it’s important: Covers the initial confusion over the cause (whether it was an attack or an internal issue) and the widespread impact.
Link: Tom’s Hardware
3. General Global News Coverage
The Guardian: “Cloudflare outage causes error messages across the internet”
Why it’s important: Discusses the broader implications of relying on a few key infrastructure providers and includes user reports.
Link: The Guardian
The Independent: “Cloudflare down latest: ‘Fix’ update issued after X, ChatGPT and more websites suffer mass outage”
Why it’s important: Offers expert commentary on the surprising nature of the global failure, given Cloudflare’s redundancy design.
Link: The Independent