# The entire internet just crashed... again Source: https://www.youtube.com/watch?v=tF_4baiIUiQ Recap page: https://rapidrecap.app/video/tF_4baiIUiQ Generated: 2025-11-19T18:38:36.66+00:00 --- ## Quick Overview The recent widespread internet outage on November 18, 2025, which affected nearly every website relying on Cloudflare for DNS hosting and DDoS protection, was caused by a latent bug in the bot mitigation service triggered by an overly large, auto-generated configuration file. **Key Points:** - The massive internet outage on November 18, 2025, impacted nearly every website relying on Cloudflare for DNS hosting and DDoS protection. - The root cause was identified as a latent bug in the bot mitigation service that crashed after processing an auto-generated configuration file exceeding the system's pre-allocated size. - Cloudflare CEO Dane Knecht apologized, stating the incident was due to a routine configuration change that cascaded into a broad network degradation, clarifying it was not an attack. - The incident caused widespread service degradation across Cloudflare sites and services, including Access, Firewall, Network, and Workers. - The video highlights the extreme dependency modern internet services have on centralized infrastructure like Cloudflare, contrasting it with the decentralized ARPANET of 1973. - Sentry, a monitoring service, is shown using its AI code reviewer to catch similar logic errors and typos before they hit production, suggesting proactive monitoring solutions. ![Screenshot at 00:08: Cloudflare's internal network status page showing a red 'Cloudflare Error' in the Atlanta region, indicating the source of the widespread service degradation.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-08.png) **Context:** The video discusses a major, widespread internet outage that occurred on November 18, 2025, affecting countless websites, including major AI platforms like ChatGPT and Claude. The narrative shifts from the initial panic and broad impact to the official explanation provided by Cloudflare's CEO, Dane Knecht, detailing the internal software failure that led to the global disruption. ## Detailed Analysis The video investigates the widespread internet outage of November 18, 2025, which caused services like ChatGPT, Claude, and even DownDetector to fail or show errors. Initial panic reactions, including conspiracy theories about stress-testing for a 'digital obedience system,' are contrasted with the official explanation. Cloudflare CEO Dane Knecht later confirmed the failure was internal, caused by a latent bug in the bot mitigation service. This bug was triggered when an auto-generated configuration file, intended to manage threat traffic, grew beyond the system's pre-allocated size, leading to a crash and broad degradation across the Cloudflare network affecting DNS, firewalls, and workers. The video emphasizes the fragility of the modern, centralized internet infrastructure by displaying the historical, decentralized ARPANET map from 1973 for comparison. Finally, the video pivots to suggest preventative measures, showcasing Sentry's AI code review feature, which identifies and suggests fixes for bugs, typos, and logic errors in pull requests before they reach production, demonstrating proactive error catching. ### The Outage Event - Panic mode triggered by widespread internet outage on November 18, 2025 - Services affected included ChatGPT, Claude, and X (Twitter) - DownDetector itself failed to load. ### Official Explanation (Dane Knecht) - Incident was due to a latent bug in the bot mitigation service - Triggered by a routine configuration change that created an unexpectedly large file - Confirmed it was an internal service degradation, not an external attack. ### Infrastructure Dependency - Visual comparison between the decentralized ARPANET map (1973) and modern centralized infrastructure providers like AWS, Azure, and Cloudflare, highlighting modern fragility. ### Proactive Error Catching (Sentry) - Sentry's AI code reviewer is introduced as a solution to catch issues like typos and logic errors in PRs (e.g., a bug in a PATCH handler consuming a request body stream twice) before deployment. ### AI Code Review Example - Sentry bot flagged a critical bug concerning incorrect foreign key usage in a GitHub organization relationship, suggesting a fix and providing a detailed analysis. ![Screenshot at 00:00: Title screen displaying 'YESTERDAY...' followed by a 'Panik' meme, setting a tone of sudden crisis.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-00.png) ![Screenshot at 00:02: Image of a server rack with shattered hard drives scattered on the floor, visually representing catastrophic data loss or system failure.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-02.png) ![Screenshot at 00:08: Cloudflare error page showing 'Cloudflare Error' in the Atlanta region, confirming the source of the widespread issue.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-08.png) ![Screenshot at 00:16: Cloudflare logo displayed after showing 'NOT AWS' and 'NOT Microsoft Azure,' emphasizing Cloudflare as the provider experiencing the outage.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-16.png) ![Screenshot at 00:29: A screenshot of the Claude.ai login page blocked by a Cloudflare security challenge, showing the immediate impact on AI services.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-29.png) ![Screenshot at 00:44: Guillermo Rauch's tweet suggesting the cause was a 'nasty useEffect,' pointing towards a software/coding issue.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-44.png) ![Screenshot at 00:57: Cloudflare System Status page showing incident reports peaking around 9:00 AM to 11:00 AM Eastern Time on Nov 18, 2025.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-00-57.png) ![Screenshot at 01:22: A graph showing request volume spiking dramatically and then falling off, with the question 'WHY?' overlaid, focusing on the unexpected nature of the event.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-01-22.png) ![Screenshot at 01:51: A YAML configuration file for Bot Management, showing the file growing past its expected size, the root cause of the system failure.](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-01-51.png) ![Screenshot at 02:33: A baseball player making an incredible catch, used as a metaphor for Sentry's AI reviewer catching broken code before it hits production \(pull requests\).](https://ss.rapidrecap.app/screens/tF_4baiIUiQ/00-02-33.png)