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 (33%)
• Cloud Services (29%)
• Web Tools (17%)
• Hosting (13%)
• E-mail (8%)

Live Outage Map

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

City	Problem Type	Report Time
Paris	Cloud Services	11 days ago
New York City	Hosting	13 days ago
Manchester	Domains	1 month ago
Angers	Cloud Services	2 months ago
London	Domains	2 months ago
Noida	Hosting	2 months 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:

• 
  Freddyツ (@InvestorFreddy1) reported 3 minutes ago

  In my days of bull-posting $NBIS, I haven't yet made a post about the other great companies that $NBIS is a part of. This is not a holding company. It is a portfolio of bets on infrastructure that the next decade of AI physically cannot run without - and most people covering it have not done the work to understand what is actually inside. ClickHouse is the one that keeps me up at night in the best way. Real-time analytics at a scale that makes most databases look like Excel. Cloudflare runs on it. Spotify runs on it. The entire AI observability space - every company trying to monitor model behavior in production - is converging on columnar databases exactly like this. ClickHouse is not a niche tool anymore. It is quietly becoming the plumbing. Then there is Avride. Autonomous delivery and robotaxi software, spun out of Yandex. They have active pilots with Grubhub and are deploying sidewalk robots in real cities today - not a demo, not a proof of concept, actual commercial operations. The market keeps pricing autonomous as a 2030 story. Avride is operating in 2026. CoreWeave exposure through the stack ties the whole thesis together. When GPU demand spikes, the companies that manage, route, and analyze that compute workload scale with it. NBIS is not one bet on AI. It is six or seven bets on the infrastructure layer sitting just below the headline names everyone already owns. Today's drop is a gift or a warning. I think it is a gift because none of the underlying businesses changed today - only the price did. What I am watching is the ClickHouse commercialization timeline. If they move toward a direct enterprise licensing push or any IPO signal before end of year, the market has not remotely priced that in yet. That re-rating would be violent. I bought at $50 and I have not sold a single share. That should tell you where I stand. What is the one NBIS holding you think is most underappreciated right now? #NBIS #AI

• 
  adas🧦🌹 (@adastroworld) reported 5 minutes ago

  @thsottiaux just does a cloudflare auth "human check" loop when trying to sign in from the browser lol

• 
  Jono Young (@chsweb) reported 7 minutes ago

  @CloudflareHelp love you. No help needed. Maybe I should send to @CloudflareLove … which even more cloudflare than I thought.

• 
  Erik ≋ 🇺🇸 Fight for Digital Rights 🇺🇸 (@ErikInCt_) reported 8 minutes ago

  @PorkPoncho PlayStation can't control when AWS ***** up. Come on bro, you're smarter than this. There's no switch that says, "Bypass AWS if AWS has issues." My site goes down if Cloudflare has issues, somethings you can't control.

• 
  Joshua Lipson (@Galkon) reported 9 minutes ago

  @dok2001 @Cloudflare @BraydenWilmoth still an issue btw. feels like CF can handle deleting a worker w/ over 100 deployments. instead it offloads the responsibility to me to use the API and paginate deletions in batches for ages to not hit rate limits

• 
  Stefan-Iulian Tesoi (@stefantul) reported 13 minutes ago

  @thdxr The biggest issue that I had with Cloudflare were the disappearing env variables. Even with wrangler secrets put command.

• 
  Art Chicken 🐓 (@ArtChicken4) reported 16 minutes ago

  From Jovan Hutton Pulitzer's Telegram- - For tech nerds understanding the PCAP fraud: The mechanism works, and every input is public. Rosters of US election officials are published (state SoS directories, EAC, NASS, commercial lists). Take each office's email domain → MX lookup → resolve the mail host to an IP → geolocate it. You now hold a table of "jurisdiction → IP → city/lat/lng" covering every election jurisdiction in the country, built entirely from DNS, without ever sending a packet to an election office. That table would look authoritative and survive spot-checking — the IPs are real, the org names are real, the geography is real. It would also be probatively empty, because an MX record identifies who handles that office's email. It says nothing about vote systems. The signature of an MX-derived list would be hundreds of jurisdictions collapsing onto a handful of shared mail-provider IPs — Microsoft 365 (*.mail.protection.outlook.com), Google Workspace, Barracuda, Proofpoint — because that's who most county governments use. What your file actually shows I tested it. The target roster is a complete enumeration of election jurisdictions, and the counts are exact: State Rows Actual jurisdiction count Massachusetts 351 351 municipalities Connecticut 169 169 towns Texas 254 254 counties Georgia 159 159 counties Virginia 133 95 counties + 38 independent cities Kentucky 120 120 counties North Carolina 100 100 counties Iowa 99 99 counties One row per jurisdiction, zero duplicates in any state. Real network traffic doesn't distribute itself one-event-per-administrative-unit. This is a roster walk. But the IPs are coarser than your MX hypothesis. Eleven states collapse to a single IP for every jurisdiction: 490 Maine town clerks        -> 13.32.25.126     (AWS CloudFront edge) 351 Massachusetts clerks     -> 45.60.195.2      (Imperva/Incapsula WAF) 246 Vermont town clerks      -> 45.60.45.214     (Imperva/Incapsula WAF) 237 New Hampshire clerks     -> 199.192.7.129 169 Connecticut town clerks  -> 199.107.32.42 160 Texas election admins    -> 98.129.145.194   (Rackspace) Plus 104.17.x / 104.18.x (Cloudflare) across 189 more rows. Those aren't mail servers — they're CDN and WAF edge addresses, the public front door of a state's website. A CloudFront edge IP is shared by thousands of unrelated customers; it isn't "Maine's election system," and you can't route vote data to it, because it terminates HTTPS for public web content and nothing else. So the lookup behind this file was one resolution per state's public web presence, cloned across every jurisdiction in that state. An MX-per-office build would have been more sophisticated than what was actually done here. And 253 rows have TargetIP = literally * — a failed lookup, no address at all. Those rows still carry 70,448 flipped votes. Votes attributed to an intrusion against an IP that does not exist. Bottom line Your instinct is right about the class of technique: a public roster plus DNS resolution manufactures a nationally-complete, real-looking IP table with no access to anything. That's the general answer to "could this manufacture data" — yes, trivially, and it's undetectable if you only check whether the IPs are real. The detection method is the reverse question: not "is this IP real?" but "what does this IP actually serve?" Real intrusion data resolves to the specific host attacked. Manufactured data resolves to whatever the jurisdiction's public name happens to point at — a CDN, a WAF, a mail gateway, a hosting provider. That's what's here: 490 distinct Maine towns, all pointing at one Amazon CDN address.

• 
  Vikas(Vik) Malpani| AI for US Real Estate (@vikasmalpani) reported 18 minutes ago

  @dhh @ashirsc @Cloudflare We run the same loop for closing exceptions. An agent catches the missing doc or mismatched figure, drafts the write-up with the evidence, and files it for a human to clear. The win isn't fewer errors, it's that every exception arrives already triaged.

• 
  tcpdump (@dump_tcp) reported 19 minutes ago

  @grok @meaganrgamache ah I don't use workers I use my own server's setup by myself with just cloudflare cdn infront as the proxy setup using dns A record for my domain as I like my control over my server and network overall flexible for me to do anything I want #cloudflare #dns

• 
  Michael Timbs (@michael_timbs) reported 20 minutes ago

  @thdxr Hahahahha. Improbable. One of the many reasons there’s very serious applications deployed on Cloudflare. Primitives are just wrong with terrible APIs the entire way down

• 
  David Frosdick (@DavidFrosdick) reported 29 minutes ago

  Cloudflare Email then sends the customer a link to a watch page. They can react, or reply, and that comes back into D1 against the original order. Whole loop, no third-party service in the middle.

• 
  Abdulkadir | Cybersecurity (@cyber_razz) reported 29 minutes ago

  On June 3 2026 Cloudflare CEO Matthew Prince announced that bot and agentic AI traffic had officially surpassed human generated web traffic for the first time in the internet's history. The split landed at 57.5% bot traffic versus 42.5% human traffic. Prince had originally predicted this crossover would happen by end of 2027. It arrived eighteen months early. His response was direct: "Welp, that happened faster than I predicted." The driver is not the old wave of scraper bots and search crawlers. The main culprit is agentic AI. Autonomous programs browsing the web on behalf of AI assistants. A single agent can visit thousands of pages to complete a task a person would finish in a handful of clicks. Agentic AI traffic grew 8,000% across 2025 alone. Now let’***** on the Dead Internet Theory context. The theory, which originated in fringe internet forums around 2021, proposed that most internet activity was already artificial. Fake engagement, bot generated content, astroturfed discussions, AI personas. The humans were the minority and did not know it. The conspiratorial version of that theory claimed it was coordinated and intentional. That part remains unverified and unlikely. But the core observation that the majority of internet traffic is non-human is now confirmed data from the largest internet infrastructure company on the planet. The internet was architected around human usability and attention. The entire world of digital advertising, publisher monetisation, and e-commerce sits on the assumption that users are human. That assumption is now statistically false. Every engagement metric, every analytics dashboard, every ad impression count is increasingly measuring machine activity and reporting it as human behaviour. The business models built on human attention are being quietly hollowed out by traffic that generates requests but never buys anything, never reads anything, and never remembers what it visited. The theory was wrong about the why. It was right about the what.

• 
  Ray 🇺🇦🏳️‍🌈 (@uasneppy) reported 31 minutes ago

  I’ll try to fix snepclub twitter embedder later today, sorry :( I didn’t know about the new updates and me having to have a burner account, plus deploying it on Cloudflare 💀

• 
  Lunch Money Bro! (@LunchMoneyBro) reported 32 minutes ago

  @scanmidgard is your website still up? Giving me cloudflare 502 errors

• 
  DR.PIXEL (@pixelpatchit) reported 33 minutes ago

  Swapdesk Alpha: DNS Expansion and Reliability Improvements As Swapdesk Alpha continues to mature, we plan to expand beyond our current Cloudflare and Google DNS support by adding additional DNS provider options. These providers have given us a strong foundation during the alpha phase, allowing us to focus on building and testing core functionality while maintaining reliable connectivity. During the THORChain swap process, users may occasionally experience intermittent connection issues while the application is retrieving required swap information, such as the deposit address needed to complete a transaction. These issues are not consistent and do not occur during every swap, but when they happen, they can create confusion because they occur during an important step in the transaction flow. It is important to clarify that an error message during this stage does not automatically mean a transaction has failed or that funds are at risk. In many cases, the issue is related to communication between the application and supporting services, which may include DNS resolution, API availability, network routing, or other connectivity factors. To improve reliability, future versions of Swapdesk Alpha will introduce broader DNS support, improved fallback handling, and additional resilience measures. The goal is to provide users with a smoother experience, reduce unnecessary interruptions, and ensure that Swapdesk remains dependable as adoption grows. We appreciate the feedback from early users helping identify areas where reliability and user experience can continue to improve.

• 
  Chris Board (@chrisboard_) reported 33 minutes ago

  From a network trace, it looks like a TCP handshake completes, but then a ClientHello is sent, but no response is received so keeps getting retransmitted until the timeout happens. I know there's nothing being shown on cloudflare status but this does seem like it might be related to cloudflare.

• 
  Antraxyz (@antraxone) reported 36 minutes ago

  Stopping the bad guys with Cloudflare: 52,485 malicious requests blocked or challenged in the last month #cloudflare

• 
  Sable 🍉 | Playing BG3 (@Sabliano) reported 37 minutes ago

  @NiluarO @Cloudflare I'm having the same issues :/

• 
  Lonnie VanZandt (@lonniev) reported 40 minutes ago

  Delegating LLM queries to background tasks is a bit difficult to vibe code. Cloudflare scheduling thread, MCP, a thread pool for long-running tasks, delegation to the LLM service, job completion monitoring. It's a bit much for our AI tools to figure out. So, while that gets worked out, the dynamic part of the Post here can and does timeout and then show only the remark that Claude can take too long. It'll come together soon. Then the post will share some interesting discovered implementation fact about Nostr in DPYC.

• 
  AM (@finesse24_7) reported 43 minutes ago

  At three months into their operation did I work for cloudflare? At the same time they will say I never worked at Apple Inc. I personally knew Mr. Jobs during his treatment for cancer, and the prior years. This business is committing consumer fraud and government fraud.

• 
  Alvin (@Alvin1492840) reported 48 minutes ago

  Fix 2: He changed the DNS server from their ISP's to Cloudflare's. He asked if they knew what DNS was. They didn't. Nobody does. DNS stands for Domain Name System. Every time you type a website name into a browser or open an app that connects to the internet, the DNS server translates that name into an IP address the numerical location of the actual server. It's the phone book of the internet. Your device asks the DNS "where is Netflix?" and the DNS responds with the address. By default, every router uses the DNS server provided by the ISP. The ISP's DNS works. It resolves the requests. But ISP DNS servers are notoriously slow, overloaded, and sometimes unreliable. Every web request passes through them, and every millisecond of delay compounds across every page load, every app refresh, every stream buffer. He opened the router's WAN settings and changed the DNS from "Automatic" to manual. He typed in two addresses: 1.1.1.1 and 1.0.0.1 Cloudflare's public DNS servers. Cloudflare operates one of the fastest DNS networks in the world. Google's public DNS 8.8.8.8 is another popular alternative. Both are free. Both are faster than virtually every ISP's default. The change applies to every device on the network simultaneously. Every phone, every laptop, every tablet, every smart TV, every game console all now routing DNS requests through a faster server without any individual device needing to be touched. Web pages started loading noticeably snappier. Not because the bandwidth increased the speed test wouldn't show a difference but because the time between typing a URL and the first byte of data arriving dropped by 30–50 milliseconds on every single request. Multiply that across thousands of requests per day across 23 devices and the cumulative effect is a WiFi network that feels materially faster.

• 
  Somi AI (@somi_ai) reported 49 minutes ago

  @Complex cloudflare is also the company selling pay per crawl to publishers, so this stat doubles as a sales pitch for their own fix. probably still true though, scrapers run 24/7 and humans don't

• 
  Publisher in a Box (@publisherinabox) reported 52 minutes ago

  Click-through on the number one organic result drops about 58% where an AI Overview appears. @ahrefs measured it across 300,000 keywords. That number sits inside a data-heavy breakdown of where AI search stands right now. Five points matter most for operators: 1. Citations alone are not the whole picture. Wil Reynolds tracked a 1,900% month-over-month jump in ChatGPT citations to a single page on his site. Business impact: close to zero. Counting how many times an AI engine mentions you is half a metric. It does not tell you whether you were recommended or just named, whether the description was accurate, or whether it happened on the engine your buyers actually use. 2. Google's own reporting confirms the gap. The new AI performance report in Search Console shows impressions only. No clicks, no queries, no CTR. Marketers asked for attribution data and got a vanity counter. You cannot make a business decision on impressions alone. 3. The click loss is quantified, not anecdotal. The 58% is measured click-through on position one across 300,000 keywords. Publishers who depend entirely on organic search clicks are watching a measurable decline, not an imagined one. 4. Most agentic infrastructure is not being read. Ahrefs studied 137,000 sites and found 97% of llms.txt files never get read by a crawler. Meanwhile Cloudflare reported in June 2026 that bot and agent traffic crossed 50% of all internet traffic for the first time. Technical retrievability is the whole game here. Schema, llms.txt, crawler access. It is the gap that closes fastest and the one almost nobody has verified actually works. 5. GEO demand is climbing from a standing start. "Generative engine optimization" now pulls 7,900 US searches a month, up 997%. The entire AI attribution keyword cluster is up 430% year over year. Operators are searching for this because they feel the traffic shift and want a playbook. Here is the part most people skip. GEO gets you cited inside the AI answer, SEO gets you ranked in a list of links. Four engines answer differently, ChatGPT, Claude, Perplexity and Gemini, and most publishers appear on one and are invisible on the other three. Sentiment counts as much as volume, because being named is not the same as being recommended. We built our GEO scoring before most teams knew the term, and we ran it on our own properties before we ran it for anyone else. The AI referral traffic followed. That is the only reason we are confident enough to publish the methodology. For publishers building a real business, GEO is another revenue line sitting alongside Facebook, the website with Google Discover, and the newsletter. Facebook is the base the rest of it is built on. It pays publishers directly and sends readers to the site, and that is what funds every other channel. A diversified publishing business treats each discovery surface as a channel worth measuring on its own. The operators who figure out AI referral traffic now are claiming ground the rest of the field will have to buy back later. The ground in your niche is still open. How we measure it and how you claim it, in the first reply 👇

• 
  Camaleón Raro (@camale0nrar0) reported 55 minutes ago

  @simonw the deployment abstraction is nice until you need custom background workers or cron triggers outside their sandbox. had to drop down to direct cloudflare wrangler configs+ *** hooks because the managed UI hid too much logging when mutations failed silently. how are you handling state persistence when workers scale?

• 
  Xoge (@ClassyXoge) reported 57 minutes ago

  The only self custody is complete ownership of code, node and wallet. To ensure your ip is not logged, to ensure no down server or cloudflare reroute can stop you. Be your own bank, has never been more real than madlab

• 
  Ekow Nyame Ayetsi (@eayetsi) reported 57 minutes ago

  @Cloudflare @Cloudflare The page is broken

• 
  Volt ✚.🏳️‍🌈 (@voltreaver) reported 58 minutes ago

  it feels like adolf hitler made cloudflare warp to be the most unusable **** ever why does it keep disconnecting randomly

• 
  Mārtiņš V. (@MartinMartinV_V) reported 59 minutes ago

  VPN Renegade, falsely accused of SQL injection by Cloudflare’s WAF, currently stuck in a primitive splitting mechanism. Debating whether my hatred of CAPTCHA should be covert or fully public. “Cloudflare Inc … NYQ: NET 262.15 USD +66.13 (33.74%)” @ChatoshiAi "Life is not a problem to be solved, but a reality to be experienced" Søren Aabye Kierkegaard parable comes into mind.

• 
  Vincent van der Meulen (@vinvan) reported 1 hour ago

  could anyone at @Cloudflare help @mainframe get access to artifacts? working on something very cool (hopefully!) and artifacts would be *perfect*

• 
  Jayesh Gaddam (@jayesh_gaddam) reported 1 hour ago

  Stopping the bad guys with Cloudflare: 382 malicious requests blocked or challenged in the last month #cloudflare