Cloudflare Broke the Internet - One Config File, Two Hours, Millions of Sites Down.
November 18, 2025: Cloudflare—the company that keeps millions of websites online—went down for 2 hours, taking ChatGPT, X, Shopify, Discord, LinkedIn, Fortnite, and thousands of other sites offline. The cause? A database permissions change that doubled the size of a configuration file. One small change. Global consequences. This is what happens when internet infrastructure becomes too concentrated.
🌐 THE OUTAGE AT A GLANCE:
2 hours of global disruption (11:20-13:20 UTC)
Thousands of websites unreachable simultaneously
ChatGPT, X, Shopify, Indeed, LinkedIn, DoorDash all down
Cloudflare protects millions of sites—all affected
Cause: Database permission change → oversized config file
Distributed globally within minutes → network-wide crash
E-commerce disrupted 10 days before Black Friday
ChatGPT unavailable during business hours (Europe/Africa)
Pattern: Cloudflare's 3rd major incident in 2025
⏱️ TIMELINE OF FAILURE:
November 18, 2025
11:20 UTC (6:20 AM ET) - Outage begins globally
11:25 UTC - Engineering team alerted
11:35 UTC - Root cause identified (oversized feature file)
11:40 UTC - Fix deployed
11:45-13:20 UTC - Gradual recovery, some regions slower
Total: ~2 hours complete unavailability, longer degradation for many
⚙️ WHAT ACTUALLY BROKE:
Simple cascade with massive consequences:
**Step 1**: Database permissions change (routine operation)
**Step 2**: Database outputs duplicate entries to Bot Management feature file
**Step 3**: Feature file doubles in size unexpectedly
**Step 4**: Oversized file propagates to ALL Cloudflare edge servers globally
**Step 5**: Software reading file overwhelmed, crashes everywhere
**Step 6**: Traffic routing fails across entire Cloudflare network
**Result**: Millions of websites unreachable despite servers running fine
The irony: Cloudflare's distributed architecture—designed for resilience—became mechanism for distributing failure globally and simultaneously.
🏢 WHO WAS AFFECTED:
*Major Services Down:*
ChatGPT - AI assistance unavailable for millions
X (Twitter) - Social media silent
Shopify - 4M+ online stores couldn't process orders
Discord - Gaming/community communication lost
LinkedIn - Professional networking halted
Fortnite - Matches interrupted globally
DoorDash - Food delivery orders stuck
Indeed - Job searches blocked
Plus thousands of small businesses entirely dependent on online sales
*Why They All Failed Together:*
These services use Cloudflare for CDN, DDoS protection, traffic management. When Cloudflare crashes, they become unreachable even though their own servers work fine.
Dependency stack: Your service → Cloudflare → Users
When middle layer fails, connection breaks.
💰 THE TIMING PROBLEM:
November 18 = 10 days before Black Friday/Cyber Monday
E-commerce stores preparing for busiest shopping period lost 2 hours of critical testing, setup, and early sales. Small businesses entirely dependent on Shopify: helpless.
ChatGPT down during midday Europe/Africa business hours = productivity halt for millions using AI tools for work.
📊 THE 2025 PATTERN:
Cloudflare wasn't alone—major infrastructure providers all had significant 2025 outages:
**June 2025**: Google Cloud, 7+ hours (Gmail, Docs, Drive, Maps down)
**October 2025**: AWS, 15+ hours (Slack, Atlassian, Snapchat offline)
**November 2025**: Cloudflare #1, ~2 hours (this incident)
**December 2025**: Cloudflare #2, another disruption
Pattern reveals: Internet infrastructure more concentrated and fragile than it appears.
🔔 SUBSCRIBE for infrastructure failure analysis
💬 COMMENT: Were you affected by the Cloudflare outage?
👍 LIKE if you've experienced infrastructure dependency
📤 SHARE with anyone who builds on cloud infrastructure