# Cloudflare's November 18, 2025 Global Outage: A Detailed Analysis
On the evening of **November 18, 2025 (Beijing time)**, global network infrastructure provider **Cloudflare** experienced a **major outage**.
This incident disrupted many prominent internet services worldwide, including **ChatGPT**, **X (Twitter)**, and **Spotify**.
Industry observers noted:
> “This event absolutely deserves its own chapter in the history of internet disasters.”
---
## Overview of the Incident
### Key Facts
- **Occurrence Time:**
November 18, 2025, **19:20–22:30** (primary impact window)
- **Duration:**
Approximately **3 hours** for core outage; about **6 hours** to full restoration
- **Scope:**
**Global**
- **Root Cause:**
Not a cyberattack. A latent bug caused by **internal database permission changes** led to abnormal ballooning of bot management configuration files, triggering software crashes across Cloudflare’s worldwide nodes.
---
## Event Timeline
- **19:05:** Change deployed to ClickHouse database access control
- **19:28:** Change took effect — failures begin
- **19:32–21:05:** Investigation underway
- **21:05:** First-stage mitigation applied (core issue persists)
- **21:37:** Root cause identified
- **22:24:** Abnormal config generation halted; nodes rolled back to stable files
- **22:30:** Core services restored
- **01:06 (next day):** Full recovery
---
## Impact Scope
This outage cascaded across the internet, affecting nearly half of global online services:
- **AI Platforms:** ChatGPT, Claude, Perplexity
- **Social Services:** X (Twitter), Spotify, Discord, Grindr
- **Gaming Platforms:** League of Legends, Minecraft servers
**Symptoms:**
- “500 Internal Server Error” messages
- Continuous human verification prompts
---
## Official Explanation
Cloudflare’s detailed incident report can be found here:
[https://blog.cloudflare.com/18-november-2025-outage/](https://blog.cloudflare.com/18-november-2025-outage/)
### Technical Summary
- The outage stemmed from **a permissions change in the database system**.
- This triggered excessive entries in the bot management **feature configuration file**, doubling its size beyond acceptable software limits.
- Oversized files propagated globally, causing routing software failures.
---
## How Cloudflare Operates
Cloudflare’s CDN powers **~20% of world websites**, delivering content from the nearest data center with latencies under **50ms** to 95% of the global population.
**Benefits include:**
- Faster site performance
- Reduced server load for operators
- Security against malicious bots and attacks
On outage night, **CTO Dane Knecht** stressed it was *not an attack*, but a vulnerability triggered by routine changes.
---
### Root Technical Cause
1. **Bot Management Feature File**:
Updated every few minutes to defend against threats
- Input: ClickHouse query results
2. **Permission change**
- Allowed retrieval of metadata from additional databases
- Duplicated feature rows added unexpectedly
3. **File size doubled**, exceeding routing software limits
4. Core proxy systems crashed, returning HTTP 5xx errors
5. Workers KV and Access services also impacted
Initially misdiagnosed as a **DDoS attack**, engineers later identified the oversized file as the true cause.
---
## Remediation Measures
Cloudflare plans to:
- Enforce **stricter file size limits**
- Validate auto-generated configuration files more robustly
- Use **staged rollouts** with continuous monitoring
---
## Business Implications for Creators
Cloudflare’s outage highlights the risk of **single points of failure** in internet infrastructure. Platforms like [AiToEarn官网](https://aitoearn.ai/) — an open-source global AI content monetization system — can help creators:
- Publish across major channels (Douyin, Kwai, WeChat, Bilibili, Rednote, Facebook, Instagram, LinkedIn, Threads, YouTube, Pinterest, X)
- Integrate AI content creation, analytics, and model ranking ([AI模型排名](https://rank.aitoearn.ai))
- Maintain visibility and revenue during network disruptions
---
## Cloudflare’s Statement
> This incident was our most serious failure since 2019.
> Outages of this magnitude are unacceptable.
> Our architecture is designed for high fault tolerance; after every failure, we improve resilience.
**Compensation:**
- No plan announced yet
- **Business** and **Enterprise** customers covered by SLA for sub-99.9% availability
- Possible **10–20% account credit** for ~4.5 hours downtime
**Additional:**
X (Twitter) is collecting compensation claims for paid users affected.
---
## Discussion
Were **you** impacted by Cloudflare’s outage? Share your experience and thoughts below.
---
## References
1. [Last night, Cloudflare's global outage took down half the internet!](https://mp.weixin.qq.com/s/XmM9pjejZcMfH3gtO5DyZg?scene=1&click_id=30)
2. [Cloudflare’s worst outage in six years: one line of Rust code halted half the world’s traffic](https://mp.weixin.qq.com/s/Lx2BiBiQPgsA5gbpJlNl3Q?scene=1)
3. [Cloudflare blog: November 18, 2025 outage](https://blog.cloudflare.com/18-november-2025-outage/)
---
*Source compilation: Kuozhanmi AIGC, infoQ, dbaplus community*
