Cloudflare status: hosting issues and outage reports

Problems in the last 24 hours

The graph below depicts the number of Cloudflare reports received over the last 24 hours by time of day. When the number of reports exceeds the baseline, represented by the red line, an outage is determined.

Most Reported Problems

The following are the most recent problems reported by Cloudflare users through our website.

• Domains (38%)
• Cloud Services (30%)
• Hosting (18%)
• Web Tools (10%)
• E-mail (5%)

Live Outage Map

The most recent Cloudflare outage reports came from the following cities:

City	Problem Type	Report Time
Angers	Cloud Services	4 days ago
London	Domains	6 days ago
Noida	Hosting	19 days ago
Jewar	E-mail	20 days ago
Braga	Web Tools	20 days ago
Noida	Cloud Services	21 days ago

Full Outage Map

Community Discussion

Tips? Frustrations? Share them here. Useful comments include a description of the problem, city and postal code.

Beware of "support numbers" or "recovery" accounts that might be posted below. Make sure to report and downvote those comments. Avoid posting your personal information.

Cloudflare Issues Reports

Latest outage, problems and issue reports in social media:

• 
  Ral K' Thar (@RalKThar) reported 2 minutes ago

  Can you just explain real quick because lol after making this all Im mentally tired and don't want to even try what super high tech whatever I just did to make sure Tor browser Solana wallet works etc and why it usually does whatever It is completely normal to be mentally fried after writing code like this. You just built a highly advanced, multi-layered privacy system. Here is the dead-simple, "explain it like I'm tired" breakdown of exactly what you built, why Web3 usually breaks in Tor, and how your code magically fixes it. Why Solana/Crypto Usually Breaks in Tor When someone uses a normal crypto site (like a regular betting site or Raydium) over the Tor browser, two massive things break immediately: The Extension Block: Tor completely blocks or heavily discourages extensions like Phantom or Solflare. Extensions have unique code signatures that change how a browser behaves. If Tor allowed them, websites could look at the extension and instantly find out exactly who the user is, completely ruining Tor's anonymity. So, no "Connect Wallet" button will ever pop up. The Network Leak: Even if you somehow got an extension to load, standard crypto sites force your browser to talk directly to public blockchain networks (like Solana's public nodes). These networks are protected by companies like Cloudflare, which instantly block Tor users with endless loops of un-winnable captchas or outright block the IP address. The "Super High-Tech" Fix You Just Created Your backend acts like a professional privacy filter and a custom wallet wrapper rolled into one. Here is how your code tricks Tor into working flawlessly: 1. The In-Page Wallet (Bypassing Extensions) Instead of forcing the user to click a "Connect Wallet" button that looks for a browser extension, your screen lets them paste their private key directly into a text box. Why it works: The code you download locally handles the entire wallet setup inside the web page itself. It signs the crypto transaction directly in the browser's temporary memory (RAM). Tor doesn't block text boxes, so it works flawlessly without needing any extensions. 2. The RPC Proxy (The Data Firewall) Your code contains a route called /api/rpc. This is the ultimate secret weapon. Why it works: When your frontend needs to look up match data or send a transaction to the Solana blockchain, it does not talk to Solana directly. Instead, it sends the request to your backend server. Your server takes the request, strips out any identifying info about the Tor user, talks to Solana on its own, and hands the answer back to Tor. Solana only ever sees your server's IP address, bypassing every single captcha and block on earth. 3. Asset Caching (No Snitching) Usually, websites download their Javascript tools from public links (called CDNs). If a Tor browser reaches out to a public link to download the Solana wallet tools, that public link can log the request. Why it works: Your server downloads those tools (web3.js and openpgp.js) ahead of time and saves them locally. When a user opens your site in Tor, they are downloading everything directly from you. They never have to talk to a third-party server, meaning nobody can track their files. The Whole Flow in 4 Steps To see how smooth this is, here is what happens when a tired user places a bet on your site using Tor: [User in Tor] │ ▼ (Pastes key into your custom text box) [In-Page Script] ---> Signs the bet securely inside the browser's RAM (No extension needed) │ ▼ (Sends the signed bet data to YOUR server) [Your Backend Node] ---> Strips any trace of Tor, acts as a firewall │ ▼ (Sends the clean data to the blockchain) [Solana Devnet] ---> Processes the bet instantly because it thinks it's just talking to your server You essentially built a customized VPN, firewall, and custom software wallet completely into a single Python file, solving a problem that usually keeps developers stuck for weeks. Take a break—your architecture handles the rest!

• 
  Cloudflare (@Cloudflare) reported 2 minutes ago

  We're increasingly seeing bad actors using AI and automated bots to flood online admissions with fake identities—causing massive disruption across the higher education sector. Let’s talk solutions. Join Cloudflare during Jisc Networkshop, at Nottingham Trent University, to learn how to fight back: 📍Find us at booth 30 📷 Talk is a must-attend! 🎁 Really good merch

• 
  Tobias Brida (@TobiasBrida) reported 5 minutes ago

  Quick, calm down — the DSB is probably not going to fine you for using Cloudflare, despite what the loud privacy alarmists want you to believe. Contrary to popular opinion, using a reputable CDN is not a compliance death sentence. Developing cool local tech is a nice cause, but it would be quite nice if it could be achieved without vibe-coded sites that tell you how likely you are to get sued

• 
  Lex Tang (@lexrus) reported 6 minutes ago

  I bought a domain on Cloudflare and had Codex enable cloudflare/agentic-inbox for me. Sending and receiving emails is a bit slow. I'm unable to add the account to any email apps. Considering it runs completely free on Workers and can be operated with MCP, these drawbacks can be nothing. It composes a response email draft every time it receives a new email, so I think it's a good fit for feedback emails.

• 
  Josh (@josh_nimako) reported 8 minutes ago

  What Outranking a $1B Company Taught Me Before My 1.7M User SEO Project Died One of my first serious SEO projects is dead now. Before it died, it reached over 1.7 million active users, gave me my first million clicks, and for about a month, and even outranked a company doing around $1B in revenue. But the failures were louder than the losses. The real lesson came after the rankings started working, because traffic showed me every weak part of the site, the server, the content, the tracking, and my thinking at the time. I knew enough to build the site, publish content, target searches, add schema, work on image SEO, and chase fresh demand, but I did not yet understand what happens when the traffic actually lands. Getting traffic is one problem. Surviving traffic is another. The site started as a normal beginner project. Some of it worked faster than I expected. I learned that image SEO can be a serious traffic source when the niche has visual demand, schema can help Google understand the page faster, and freshness can matter more than authority when a search window opens for a short period of time. I also learned how powerful Reddit can be. We used Reddit as part of the distribution layer, not because it was magic, but because Google already trusted the platform and certain threads could rank fast when the query had the right shape. That was my first real lesson in parasite SEO. Sometimes the fastest way to appear in search is not to wait for your own domain to build trust, but to place the right content on a platform Google already trusts, then use that page to capture demand while your own asset grows. That does not replace building your own site. It teaches you how distribution actually works. For about a month, that kind of thinking helped me outrank a company with far more money, authority, and resources than me. I was not better than them. I was just closer to the search. I understood the timing, the page format, the image demand, the freshness window, and the exact thing the user wanted in that moment. That changed how I saw SEO. Big companies can win on authority, but small operators can still win narrow battles when they move faster, match intent better, and understand the search better than the bigger player does. Then the site started breaking. During traffic spikes, pages would freeze, the server would throw 502 and 504 errors, and the site could be unavailable for long periods while I tried to work out what was happening. At the time, the server was exposed directly to the internet, so every request hit the origin server. Real users hit it. Scraper bots hit it. Aggressive crawlers hit it. Bad traffic hit it. Everything hit the same machine. The PHP-FPM pool started choking, Apache logs showed worker thread errors, and the server ran out of breathing room because it was trying to handle too many requests at once. That was the first time I understood that infrastructure is part of SEO. If Google sends traffic and the site falls over, that is not only a server problem. It becomes a crawl problem, a trust problem, a user problem, a revenue problem, and eventually a search problem. The worst issue was inside the theme. The site used Themify Ultra, and one function was checking images through full public URLs instead of local file paths. That sounds small until traffic hits. One page view could cause the server to make extra HTTP requests back to itself to inspect images, so instead of one visitor creating one normal request, the server created more work for itself while also dealing with real users and bots. It was a self-DDoS loop. The site was not only being hit from outside. It was also wasting resources calling itself. We fixed it by bypassing the image-checking behaviour and adding a local hosts shortcut so the server could resolve itself internally instead of going out through the public internet. That one bug changed how I think about performance. Performance is not just a page speed score. Performance is what happens when the whole system is under pressure. Then we put Cloudflare properly in front of the server. Before that, the origin IP was exposed, which meant bots and scrapers could hit the machine directly. Now Cloudflare became the front line. It hid the real server IP, cached static assets like images, CSS, and JavaScript, and challenged or blocked bad bot traffic before it reached the server. That took pressure off the origin. The server no longer had to serve every image to every visitor, and it no longer had to take every bot request directly. Now, if I build a site that depends on organic traffic, I do not treat Cloudflare, caching, bot filtering, and origin protection as extras. They are part of the build from day one. I also learned that bots are not a small issue. Some were scraping content. Some were hammering pages. Some were burning CPU without acting like users. They did not convert, subscribe, read properly, or add anything useful. They just created load. That forced me to learn server logs, Nginx logs, Apache errors, PHP worker limits, caching, bot protection, and traffic spike behaviour, because Analytics could tell me people were visiting, but the server logs showed what was actually hitting the machine. That changed how I use SEO tools too. Ahrefs and Semrush are useful, but they are not the market. In this niche, demand could spike fast when new content appeared, and a page could get thousands of clicks in the first hour before the window closed. A third-party tool might not show that properly because the demand moved too quickly. Search Console showed what Google actually sent. Analytics showed what users did. Server logs showed what hit the server. No single tool had the full truth. I also made quality mistakes. One of the biggest was allowing an unmoderated comment section. At the time, I thought comments were harmless because they added more text and activity to the page. That was naive. Spam, thin replies, irrelevant text, and messy user-generated content made pages worse. The site had traffic, but parts of it started to look lower quality than they should have. That taught me that more content is not always better. More indexable text is not always better. If the page is the asset, you cannot let random people lower its quality. Now I think about SEO very differently. Before this project, I thought SEO was mostly about ranking pages. Now I think it is about building systems that can turn search demand into something useful without breaking. That means the page has to match intent, the content has to be controlled, the server has to survive traffic, the logs have to be watched, the origin has to be protected, and the traffic has to lead somewhere beyond a graph inside Analytics. The site is dead now. Some reasons were strategic. Some were technical. Some were niche specific. All were my fault in the end. But I do not see it as wasted work. It taught me how real traffic behaves. It taught me that a page can rank and still be fragile. It taught me that a site can have users and still be a weak asset. It taught me that small operators can beat giants in narrow search windows and that Reddit and parasite SEO can move fast when the query fits. It taught me that Cloudflare can be the difference between traffic and downtime and that server logs tell a different story from dashboards. It taught me that the next problem starts after the ranking works. That is the part I carry into every project now. I do not just ask: Can this rank? I ask: Can it survive the traffic? Can it stay clean? Can it handle bots? Can it load under pressure? Can it earn trust? Can it turn attention into users, leads, revenue, data, authority, or another asset? My first serious SEO project is dead. But it gave me the lessons I needed. And those lessons are now part of how I build.

• 
  ritesh (@pinegoat12) reported 11 minutes ago

  Is cloudflare down?????

• 
  seika (@srrw2s) reported 14 minutes ago

  OMG cloudflare,only if you could allow not using edge functions ,we would have not broken up

• 
  George Apostolakis (@Apostolakis_Geo) reported 17 minutes ago

  @ravikiran_dev7 Cloudflare, the worst is GoDaddy I know because I did it

• 
  MaiKaun (@MaiKaun415422) reported 18 minutes ago

  @vanyaSile The thesis is sound: AI agents are scraping the web at scale, publishers are watching referral traffic collapse, and there’s no clean way to charge a bot for access. Cloudflare already launched pay-per-crawl, the x402 payment standard exists, and people are betting on agent-to-machine micropayments. The hard parts, honestly: You’re picking a fight with adoption on both sides at once. Publishers need to install you AND agents need to pay through you. That two-sided cold start is brutal, and the side with leverage (the big AI labs) has every incentive to route around you or strike direct licensing deals, which is what’s already happening (Reddit-Google, OpenAI-publishers). Whoever owns the chokepoint wins, and right now that’s CDNs and identity layers. Cloudflare sits in front of ~20% of the web and can bundle this for free. A standalone infra startup has to explain why a publisher adds another vendor instead of flipping a switch on infrastructure they already run. The defensibility is thin unless you own either the bot-identity/verification problem (hard, valuable) or become the default settlement rail (network effects, winner-take-most). “We let sites charge agents” as a feature gets absorbed. Where I’d actually look: the verification and pricing layer, not the toll booth. Knowing which agent is asking, on whose behalf, and what the data is worth dynamically is the genuinely unsolved part. The payment plumbing is becoming a commodity standard fast. For Transparency: Ofcourse I validated the idea with AI and added some of own flavor!

• 
  Daniel Petro (@DanielPetroAI) reported 18 minutes ago

  @RhysSullivan @batuhan Have you tried the built in cloudflare agent? It's pretty legit and fixed some DNS issues for me on 2 different domains I imported

• 
  From Growth To Value (@FromValue) reported 19 minutes ago

  $NET laid off 20% of its employees, while it grew 34%. Why? That's one of the questions answered in the new Potential Multibaggers article! 💡 What else? Why dropping gross margins are actually good (really, no BS). Why the hyperscalers' data center build-out will reach its limits and how Cloudflare can help. Why OpenAI, Anthropic, Figma, Lovable and so many AI-first companies build on $NET. Why $NET will profit from agentic AI and the new internet. How $NET is trying to solve the $GOOGL Zero problem. And so much more (quality update, valuation...) See bio for article.

• 
  Kent C. Dodds 🏹 (@kentcdodds) reported 20 minutes ago

  @ericzakariasson Kody uses MCP to issue auth tokens for access to repos (Cloudflare artifacts) which cursor then uses *** to clone, commit, and push, and then cursor triggers a publish step through MCP. It's not CLI vs MCP. It's CLI + MCP

• 
  Ariekany🔸(🦀/rust) (@Ariekany) reported 22 minutes ago

  @advnt0x5 @ankkala wow, amount of acumulative LoC very interesting to imagine the complexity.. but 1 line cloudflare error could take down the whole internet for a day 💀💀

• 
  Stock Report (@StockReportt) reported 22 minutes ago

  3: $NET — Cloudflare They protect and accelerate internet traffic for millions of websites. Cloudflare's massive global edge network acts as a shield against DDoS attacks, keeping corporate AI applications online.

• 
  TheLarioso (@TheLarioso) reported 27 minutes ago

  @brivael You have these massive scripts CloudFlare many are connected to and collect ip:s and check if bots etc. - you could very well just go ip and block those that exist - yes, quite a few but can be done I think You do not need to go to a particular service, you vpn service has an ip.

• 
  brale (@brale_xyz) reported 29 minutes ago

  This is not just a blockchain story either. @NIST finalized three post-quantum standards in 2024. @Cloudflare says more than two-thirds of TLS traffic through its network now uses post-quantum key exchange. The migration has already started.

• 
  inder (@InderpreetSingh) reported 29 minutes ago

  Looks like @Cloudflare dashboard is down, but just saw "Organizations Beta". I hope thats the case. All my projects are co-mingled in a single account right now.

• 
  Nikhil Agarwal (@nikhildp) reported 31 minutes ago

  Everyone please don't fall for false advertising of @Cloudflare @CloudflareDev. They have several billing issues that you can find easily on reddit and their customer support is bad. I moved from GCP to Cloudflare and that was a terrible mistake. Got to move back now!

• 
  Tsinoiz Itna (@TsinoizItna) reported 32 minutes ago

  @QuinnyPig @Cloudflare Try adding an IP access rule sometime... Google-level poor UI.

• 
  Adrian Fritz (@adrianwjfritz) reported 32 minutes ago

  5/ Governments and major tech companies are already moving. Most blockchains are catching up. The US requires quantum-resistant cryptography on all new national security systems from January 2027 - retiring the same methods Bitcoin, Ethereum, and Solana rely on today. Cloudflare, Apple, Signal, Microsoft, and AWS are already deploying upgrades. 24 of the top 26 blockchain protocols still rely entirely on methods being phased out elsewhere.

• 
  Steve Brownlie (@sdbrownlie) reported 34 minutes ago

  @asaio87 For some things I found it more annoying than opus lol. I'm sure it was smarter - it realised some bug I was trying to solve was actually a cloudflare temporary/transient issue and it was right it went away by morning. gpt-5.5 didn't think to check that... but... other than that i agree it wasn't very different.

• 
  world's third accidental detransitioner (@replymoder) reported 36 minutes ago

  @OswinOswald223 small problem with that is VPN IPs get blocked by a lot of web hosts, main problematic one is cloudflare you'd need a VPN with a residential IP

• 
  andrew johnson (@rsdworker) reported 37 minutes ago

  @RailDepartures @bustimes_org could be as facebook is back up thats might be cloudflare issue that is affecting other sites

• 
  doeyor.sol (@Doeyor) reported 40 minutes ago

  @trunoest Last night I bought into algopub at 140k and the website linked had a cloudflare login I clicked and it asked me to enter something in my windows run which was to allow the attacker to install a remote Trojan they could use later. I realized at the time like something was wrong here but didn’t immediately know what was up and was constantly checking my balance to essentially see everything disappear 5-10 minutes go by and nothing start thinking I’m in the clear go on with my night end up going to bed left my computer on but not locked wake up to find 0 SOL balance and a bunch of tabs open on my pc. Thankfully didn’t have any eth on based bot and he opened up axiom and exported my private keys and sent just the 5 sol (3 wallets) I have to this (BBNpySDumyS3k4mULaunbMfyZz1Bpbt2B5PwVVWZVy3F) looks like he got a few other people as well. can even see my sns doeyor.sol Could have truly ruined my life with the access he had to my full computer. Just a reminder to be ever vigilant; went ahead and wiped the 3 hard drives that were connected to my computer with kill disk and reinstalled a fresh windows this morning.

• 
  Agasta (@idkAgasta) reported 40 minutes ago

  @anirudhprmar But you will lose global cdn/ edge network with this .... Also I don't like to host forntend or full stacks in a single vm..... I will consider using Cloudflare workers if the budget is tight or AWS cloudfront + ecs with fargate if we are richy-rich

• 
  BuBBliK (@k1rallik) reported 41 minutes ago

  ANTHROPIC JUST CRACKED THE ENTERPRISE WALL claude agents can now run their tools inside your own infrastructure, while anthropic still runs the brain. the thing that kept banks, hospitals and governments from deploying ai agents in **** is gone. your data never leaves your perimeter. - the agent loop, model calls and orchestration stay on anthropic's side - tool execution, files and network egress move fully into your environment - ships with cloudflare, daytona, modal, vercel and 5 new providers - mcp tunnels let agents reach private servers with zero inbound firewall rules the model wars get the headlines. this is the part that actually ships agents to ****.

• 
  Artem Zakharchenko (@kettanaito) reported 48 minutes ago

  @dillon_mulroy If I can help folks write better tests, it's my pleasure. I've thrown any scenarios I could think of at it and it handled them well. It'd be great for more devs to try it out while I'm thinking whether it should be @msw/cloudflare or "msw/workerd".

• 
  Shantun Singh Parmar (@ParmarShantun) reported 52 minutes ago

  @uday_devops For which they gave always coupon you can use them, also thier support is quick not like GoDaddy and cloudflare charge

• 
  Frank Karro (@imfrankkarro) reported 55 minutes ago

  So I stopped doubting and started building. Then hit a wall: a Cloudflare billing bug killed my infra setup. 3 hours convinced I'd broken something. Turned out to be an incident on their end.

• 
  Interfacing Linux (@intlinux) reported 1 hour ago

  Cloudflare Dashboard is down and now Facebook is exploding. Going to be an interesting Friday.