The global Cloudflare outage of 18 November 2025: what happened, its impact and why La Coderia was not affected

On 18 November 2025, a massive Cloudflare outage brought much of the internet to a standstill for several hours. As Cloudflare is one of the pillars of the world’s web infrastructure, its failure set off a domino effect across thousands of services. In this summary we look at what happened, what caused it, which services were affected and why La Coderia did not suffer any disruption.
1. What happened to Cloudflare on 18/11/2025?
Start and duration of the incident
- The failure began at around 11:20 UTC (12:20 in mainland Spain).
- From that moment on, Cloudflare started failing badly, preventing its network from delivering traffic normally.
- Users saw internal errors such as “Internal server error”, a sign that the problem lay within Cloudflare itself.
- The company acknowledged the incident shortly afterwards and confirmed that it was investigating.
How the outage unfolded
- During the afternoon (European time), services fluctuated: outages, partial recoveries and fresh disruptions.
- After around four hours of unstable operation, Cloudflare declared the incident “resolved” at around 16:00 CET.
- However, full stability did not return until approximately 17:06 UTC, according to the technical report.
- In total, the main outage lasted several hours and affected all of Cloudflare’s critical systems.
2. The cause: an internal bug, not a cyberattack
Cloudflare confirmed that it was not an attack or malicious activity. The problem was entirely internal:
The technical origin
- A change to the permissions of a database cluster caused a faulty query.
- This generated a configuration file (a “feature file”) used by the bot management system… but with invalid data and an excessive size.
- Every five minutes, this file is distributed globally across Cloudflare’s network.
- As it exceeded the maximum size the software could process, the servers simply failed.
Why the outage was intermittent
- Only some nodes generated the faulty file; others generated a valid one.
- Depending on which one was distributed in each five-minute cycle, the network:
- recovered temporarily, or
- went down again immediately.
The fix
- Cloudflare stopped the faulty file from spreading.
- It reverted to a valid version and restarted the affected systems.
- Traffic began to return to normal at around 14:30 UTC, once the root cause had been contained.
Cloudflare’s response
- The company described the incident as “unacceptable”.
- Its CEO apologised and promised new safeguards:
- stricter limits on configuration files,
- systems to quickly switch off problematic configurations,
- internal reviews to prevent similar failures.
3. Services affected and global reach
As Cloudflare handles around 20% of the world’s web traffic, the impact was enormous. Thousands of websites and apps were affected.
Services affected
Social media
- X (Twitter) stopped loading content or would not let users post.
- Forocoches, Menéame and other community services showed errors.
Artificial intelligence
- ChatGPT, Claude, Perplexity AI and DeepSeek went down almost completely.
- Only services that do not depend on Cloudflare (such as Gemini or Copilot) remained stable.
Online gaming
- League of Legends and Valorant suffered massive disruptions.
- Platforms such as Battle.net recorded serious connection failures.
Popular web services
- Canva stopped working for hours.
- Bet365, Grindr and some banking services were also affected.
Transport apps
- Uber and delivery services suffered outages in certain countries.
Monitoring sites
- Even tools such as Downdetector failed, because they rely on Cloudflare.
Geographical reach
The outage was global, affecting users in Europe, Latin America, North America and Asia. Any region with services that use Cloudflare felt the impact.
Internal systems affected
The internal products affected included:
- CDN and security services (widespread 5xx errors),
- Turnstile (verification system),
- Workers KV (global storage),
- the admin dashboard,
- Email Security,
- Cloudflare Access (Zero Trust authentication).
4. Reactions: confusion, memes and transparency
Users and the technical community
- Social media filled up with posts asking whether “the internet was down”.
- With X itself partly down, it was even harder to spread information.
- There was no shortage of memes recalling earlier Cloudflare outages.
Affected companies
- Many companies explained that their problems were due to third-party failures.
- Cloudflare kept updating its status page and posting progress reports.
- Once it was resolved, its CEO shared a lengthy technical report and apologised publicly.
5. Impact and lessons: the internet is more fragile than it seems
This incident reopened the debate about the centralisation of the internet. When one of its pillars fails, the impact is instant and global.
Recent context
- It happened just weeks after major AWS and Azure outages.
- How close together these incidents were highlights our dependence on a handful of key providers.
The main lesson
Design systems with redundancy and resilience. Relying 100% on a single provider (Cloudflare, AWS, etc.) can be a major risk.
6. Why La Coderia did not go down: we do not use Cloudflare
While thousands of websites were collapsing, La Coderia and our clients stayed up without a single problem. The reason? We do not depend on Cloudflare.
Why do we avoid Cloudflare?
Although it offers benefits such as a CDN, caching, SSL and DDoS protection, we avoid it for a reason that is essential for our users:
LaLiga blocks Cloudflare IP addresses during match broadcasts.
In particular:
- Spain’s LaLiga has rolled out a wave of preventive blocks against pirate streams.
- When illegal streaming is detected on a Cloudflare IP address, they order that IP to be blocked.
- The result: legitimate websites that share the IP (such as blogs or forums) become inaccessible to many users, without having done anything wrong.
This has affected:
- technical blogs,
- forums,
- personal websites,
- any site that shares the “wrong” IP on Cloudflare.
There have even been cases of:
- websites receiving legal warnings because of other users on the same Cloudflare IP,
- internet providers being ordered by the courts to block shared IPs.
Our decision
At La Coderia our priority is that nobody is blocked unfairly. That is why:
- We do not use Cloudflare.
- We do not share IP addresses with potential pirate broadcasters.
- We control our own caching and security solutions.
Thanks to this:
- We avoid blocks related to football matches.
- And on 18/11/2025 we suffered no impact at all from the massive outage.
Conclusion
The global Cloudflare outage of 18 November 2025 showed how fragile the internet can be when so much infrastructure depends on a few players. A single badly generated file caused massive failures across social media, gaming, financial services, AI tools and more.
Cloudflare resolved the incident and took responsibility. But the event leaves a clear lesson: do not depend on a single provider for critical services.
La Coderia remained fully operational because we do not use Cloudflare, a decision we made for very practical reasons and which has now proved to be the right one.
We will keep sharing infrastructure analysis and technical decisions designed to offer the best possible experience to our community.

