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 (41%)
• Cloud Services (25%)
• Hosting (18%)
• Web Tools (11%)
• E-mail (5%)

Live Outage Map

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

City	Problem Type	Report Time
Noida	Hosting	12 days ago
Jewar	E-mail	13 days ago
Braga	Web Tools	13 days ago
Noida	Cloud Services	14 days ago
Paris	Cloud Services	14 days ago
Prievidza	Domains	14 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:

• 
  X-Ray Tango (@xraytangooscar) reported 6 minutes ago

  App: I used to go the full Cloudflare route (Workers, D1, etc) but the Wrangler part was kind of a pain. These days I push Node code to Render and build on PostgreSQL which is hosted on Supabase. For **** with a small number of users, I'm far from hitting the free tier limit.

• 
  Peter Garety (@PeterGarety) reported 8 minutes ago

  @Cloudflare We built our own because you didn't have this - now we support nearly 80 models with clear billing attribution.

• 
  VC Intern (@the_vc_intern) reported 10 minutes ago

  VC Concept Simplified: Operating Leverage Some companies get more profitable as they get bigger. That sounds obvious, but it is not automatic. Operating leverage is what happens when revenue grows faster than operating costs. A startup may need to spend heavily upfront on product, engineering, infrastructure, brand, or sales. But if those costs do not rise at the same speed as revenue, the business starts to change shape. The same company that looked expensive at $10M revenue can look much cleaner at $100M revenue. This is one reason software businesses became so attractive to venture. The first version of the product is expensive to build. But once the product works, serving the next customer can cost much less than building the product again. Shopify is a useful example. In 2024, Shopify kept growing revenue while expanding free cash flow margin through the year, reaching 22% in Q4. That is the kind of pattern investors like: more scale, more cash generation. Duolingo shows a different version. The company kept growing users and paid subscribers, while Q4 2024 delivered record revenue, record Adjusted EBITDA, and 42% free cash flow margin. The product, brand, and distribution engine started throwing off more economics as usage scaled. Cloudflare is still investing aggressively, but investors watch the same question there too: Can revenue keep growing while operating margins improve over time? That is the founder lens: Growth is good. Growth that makes the business easier to run is better. Operating leverage is the moment scale stops only making the company bigger. It starts making the company stronger.

• 
  Hey...Juan (@JuanAG_IT) reported 11 minutes ago

  @seangeng Local inference on an airplane your only use case? For me personally it was realizing how fickle our infrastructure could be. aws and cloudflare issues last year made me realize that I don’t wanna be pigeonholed by single points of failure and not have access to the tools I’m becoming accustomed to

• 
  Alvin De Cruz (@decruz) reported 12 minutes ago

  @pilcrowonpaper Typically for caching, firewall support, and does the heavy lifting of speeding up the site via compression. And tacking on Cloudflare is free.

• 
  Alan | Broadcaster | WhatsApp for trades (@alanefuller) reported 12 minutes ago

  @uglyrobot @KatieKeithBarn2 For instance I use complex products like Amazon Web Service, Cloudflare, Google Cloud, and my method of operation now is ask an LLM how to do something,.

• 
  Jovi Ebi | Graphic Designer🎨& Brand Strategist💻 (@boyjovi_) reported 13 minutes ago

  @AirtelNigeria why cant i use my airtel oud unlimited for youtube and x without a private network connection like cloudflare warp. i dont undertand you guys.

• 
  Tanner Scadden (@TannerSDev) reported 14 minutes ago

  @awscloud please help us asap your team suspended our account at the same time we got an email saying we have two weeks to address a problem with our root account. Cloudflare is down for us, and EC2 won't provision new instances. Our app is down. We serve hospitals. Please

• 
  esyx (@esyx0) reported 21 minutes ago

  TIL @digitalocean blocks SMTP ports on your VPS 🫠 i was looking for a mail service provider and ended up choosing Purelymail, they have one $10/year (yea, YEAR) for unlimited mails (transactional, they prohibit marketing ones). And now i found out that DO blocks the SMTP ports So now im looking at either Cloudflare (probably favorite since i already use Cloudflare) or Amazon SES but it's so annoying to pick/pay/setup the whole thing only to found out it doesn't work because of this stupid thing

• 
  BuyTheDiplo (@BuyTheDiplo) reported 21 minutes ago

  My top 10 SaaS watchlist: $MSFT Microsoft The king of enterprise software. Office, Teams, Azure, GitHub, Copilot. Not pure SaaS, but it owns the business software stack. $NOW ServiceNow Runs workflows for large companies. IT, HR, customer service, automation. Boring product, elite business model. $CRM Salesforce The customer relationship management giant. Sales, marketing, service, data, AI agents. The question is growth reacceleration. $ADBE Adobe Creative software monopoly. Photoshop, Acrobat, Creative Cloud, Firefly AI. Huge margins, but AI disruption risk is real. $CRWD CrowdStrike Cybersecurity SaaS. Protects companies from hacks. Cyber is not optional spending anymore. $DDOG Datadog Cloud monitoring and observability. Helps companies see what is breaking across apps, servers, AI workloads, and cloud systems. $SNOW Snowflake Data cloud. Helps companies store, organize, and use massive data sets. Big AI upside if enterprise data spending accelerates. $NET Cloudflare Internet infrastructure layer. Security, speed, edge computing, developer tools. Expensive, but strategically important. $WDAY Workday HR and finance software for large companies. Payroll, hiring, employee data, planning. Sticky because replacing it is painful. $MDB MongoDB Database software for modern apps. Developers like it, AI apps need flexible data, but valuation and competition matter.

• 
  Phil | Rentier Digital Automation (@rentierdigital) reported 24 minutes ago

  your claude.md is already wrong you write a CLAUDE.md once and assume it's done. then your codebase evolves. dependencies shift, ports change, schemas multiply. the file sits there looking authoritative while everything around it drifts turns out /init isn't a generator. on repos with existing docs, it's an auditor. reads your CLAUDE.md, cross-references against lockfiles and configs, surfaces what stopped matching while the code moved forward i ran it on 4 repos. repo 1 had npm commands documented but was running on bun.lock. wrong port hardcoded. both had been true once, both silently false now. nothing broke bc npm commands mostly work on bun projects anyway. you just try the wrong URL then the right one, forget about it repo 3 was undocumented. multi-frontend catalog system, Astro frontends, Hono backend, SQLite. no README. /init did 4 parallel tool passes and reconstructed the entire delivery chain: SQLite schema, export pipeline, Astro build under Node 22 (not Bun bc of SSG renderer constraints), rsync to nginx, Cloudflare Workers routing. one pass. specific enough to navigate without asking repo 4 had 64 lines of CLAUDE.md. /init found 5 silent lies. package manager mismatch. schema states (3 documented vs 7 real). routing layer that was actually a full processor. undocumented mode switches. port number wrong in 3 places documentation drift doesn't announce itself like a build error. it's corruption in a save file. the game keeps running, the numbers look fine. then the boss fight hits and the stats don't add up the work isn't generating boilerplate. it's finding the delta between what you wrote and what the code actually became i build and ship daily with Claude Code. SaaS, tools, automations. ⭐ if AI can build it, I've probably broken it first. what works → link in bio

• 
  teej dv 🔭 (@teej_dv) reported 25 minutes ago

  @just_cromer i'm still working on a lot of ideas of what the runtime will look like. i am really interested in actors because i think they are cool and then i can get rid of a bunch of shared memory parallelism problems without having to introduce all the modes oxcaml has. and it works nice for things like serverless deploy/multi-computer workfows/cloudflare workers/DOs, etc so i'm not sure yet. a lot to play with there right now

• 
  James Stevens 🏴󠁧󠁢󠁥󠁮󠁧󠁿🇬🇧 (@TheFrogDies) reported 26 minutes ago

  @Unionbuster @NoFarmsNoFoods the startup I worked for re-launched the dot-IO domain space 2006 to 2017, then sold to a US company who took all the intellectual property and completely shutdown the UK operation, all jobs lost I urged the owner not to sell & continue building, but his company, his choice - I guess $70M cash on the table is very tempting I also worked for Nominet (runs dot-UK) who also did cyber security for the UK gov, but when we bid for a contract our bid was a mess & CloudFlare just came in with such a low offer the government couldn't turn it down So the cyber division of Nominet was pretty much shut down, big job losses, and the UK lost that cyber security expertise forever. I think it was a very short sighted decision, but the bid was a mess - Nominet had a big problem with Main Character Syndrome amongst ppl who had the yrs of service, but not the technical capability or expertise

• 
  Alec (@alecsiemerink) reported 28 minutes ago

  Public ingress is via Cloudflare Tunnel. No router port forwards. No random “I opened this for testing and forgot” holes. Private/admin paths stay behind the Tailnet. Again: not because this is enterprise. Because future-me is absolutely capable of being an idiot.

• 
  Lord Waffleman (@LordWaffleman) reported 29 minutes ago

  @maietta @jpschroeder Yeah. That’s why think it’ll get worse. The massive amount of attacks, vulnerabilities we are seeing I think are driven by AI and … “another issue.” Cloudflare issues for the last couple years have set a few fires, so I can’t imagine it getting better.

• 
  Long John Little (@LongJohnLittle) reported 29 minutes ago

  @Cloudflare I am user that cannot open a support and I am facing a very weird issue, Routing blackhole from OTE/Cosmote (possibly all ISPs in Greece) to Cloudflare IP 104.21.43.120", can you forward this to your network support?

• 
  N1Stock (@N1Stock) reported 31 minutes ago

  @drpezeshkian Prerequisite: Make sure your network can reach Cloudflare and Google (ICMP or TCP connectivity). This indicates that you have access to the international internet and can communicate with external servers outside of Iran.

• 
  Sahil Nawaz (@sahilyaps) reported 32 minutes ago

  ChatGPT can think. Claude can reason. Grok can search. None of them can buy a $5 API. That's the bottleneck. Everyone is obsessed with AI intelligence. Almost nobody is talking about AI payments. But that's where the next trillion-dollar market is forming. @awscloud just launched AgentCore Payments with @coinbase infrastructure. @Cloudflare and Coinbase are pushing x402. The entire premise is simple: (1) An AI agent discovers a service. (2) An AI agent pays for that service. (3) An AI agent consumes that service. No human in the loop. For 30 years, HTTP 402 ("Payment Required") was basically ignored. Now it's becoming the payment layer for machine-to-machine commerce. Most founders still think AI is a content tool. The biggest opportunity may actually be AI becoming an economic actor. Question: What happens when there are more agents spending money online than humans?

• 
  Your Private Proxy (@YourPrivateProx) reported 36 minutes ago

  Cloudflare and Datadome aren't the same problem. Cloudflare: TLS fingerprint + JS challenge. Fix your JA3, done. Datadome: persistent behavioral model. Clean TLS on a residential IP still flags if request timing is machine-uniform. Different layers. One fix doesn't cover both.

• 
  Jeni Farnsworth (@farnsjennifer) reported 36 minutes ago

  @KansasOz45 April 2020 “Guo has claimed not only that his Hong Kong assets were frozen by the CCP, but that his operations have been shut down there. Yet Guo Media’s Cloudflare account shows that ..."i

• 
  Will (@war59312) reported 36 minutes ago

  @BinTube DNS issue on Google and Cloudflare. Working fine on OpenDNS.

• 
  Tanner Scadden (@TannerSDev) reported 37 minutes ago

  @AWSSupport DMd. 19:27 we got the notice, saying take action by 06/22/2026. At 19:29 our production environment went down due to cloudflare being disabled. Please help us asap

• 
  Whoa! Shut It Down (@rightsofrefusal) reported 38 minutes ago

  @CultLaser @HoffmanTactical Cloudflare told me the website was down last night when I went to order one and now it says it's all sold out. 😭

• 
  Augustus (@IndieBeaverHere) reported 39 minutes ago

  @vpdn @cesaralvarezll During the last cloudflare outage, RC was down for quite some time. I clearly remember that those who used RC paywalls were in trouble as it didn't work and those who used their own paywalls with RC integration more or less worked. I bet you guys put safeguards for this now.

• 
  j (@jordanrstout) reported 43 minutes ago

  @dillon_mulroy 1. What are you instructing it to do to give you these call stacks then also convert to basic contract implementations 2. In cloudflare, specifically with alchemy-effect, how are people testing things like DO. They don’t support miniflare and their testing utilities are only good for basic e2e stuff. Trying to like alchemy but it looks better as a vibe code to “production” tool but I prefer to AI code with TDD and not just test api responses.

• 
  Balogun Hammed (@bhalloinfraguy) reported 44 minutes ago

  Infrastructure Concept — explained: What is Anycast Routing? When you query Google's DNS at 8.8.8.8, where exactly is that server? The answer: it depends on where YOU are. 8.8.8.8 isn't one server. It's hundreds of servers around the world, all advertising the same IP address. Your request automatically goes to the nearest one due to anycast routing. How Anycast works: Multiple servers in different locations all announce the same IP to the internet using BGP. When a user makes a request, BGP routing naturally sends it to the closest server based on network topology. The same IP. Different physical locations. No DNS tricks. No load balancer redirects. The internet itself does the routing. Real-world uses: DNS resolvers: (8.8.8.8, 1.1.1.1), fast lookups globally. CDNs: Cloudflare uses Anycast for edge servers. DDoS protection: absorbs attacks across many locations. Time servers (NTP): accurate time globally. Critical infrastructure where speed matters. Anycast vs other casting types: Unicast: One sender to one receiver (most internet traffic) Multicast: One sender to a defined group of receivers Broadcast: One sender to everyone on a network. Anycast: One sender to the closest receiver. This is why public DNS is fast, no matter where you are. The "server" you're talking to is literally next door.

• 
  Bull Theory (@BullTheoryio) reported 45 minutes ago

  BREAKING: Anthropic is expected to release Claude Mythos tomorrow, the same model it said was too dangerous to make public. A "Mythos 1" tag was briefly spotted inside the Claude Code UI last week before being pulled, signaling a public release is imminent. In a restricted preview, Mythos found 271 vulnerabilities in Firefox alone, including a 15-year-old bug in Mozilla's HTML engine and a 20 year old flaw in its XML processor that years of human auditing had completely missed. Mozilla went from patching 21 security issues per month to 423 in a single month. When Mythos was first leaked in March: CrowdStrike fell -7% Palo Alto fell -6% Zscaler fell -4.5% Okta and Netskope fell -7% Tenable crashed -9% Cloudflare fell -13% Thomson Reuters fell -19% RELX fell -15% LegalZoom crashed -20%. The S&P 500 Software and Services Index fell 2.6% in a single session and is now down 12% since January.

• 
  Dave Lui ☀️🪝 (@madave_lui) reported 51 minutes ago

  @cosmostation888 @HugoPhilion Just based off what they say and who they're connected to. You have to dissect the facts from the noise, especially those who use "prophecy" as an investment criteria. But, if they succeed at building strong rails inside of ACH payments that directly connect to banks they could see a pretty great increase in price. I think they have a shot, but I could lose all my investment with them. They're currently in my mid-conviction tier. I do think they can do pretty alright, though, and maybe they'll even surprise us all. There's a lot of folks who are, in my opinion, overly optimistic with the dots they've connected. I'm a bit more balanced and tend to lean on what's actually been said/announced... For instance they call themselves a Ripple strategic partner. They also said Ripple invested in them. To date I can't find a single Ripple announcement that confirms that. What I did find is that Ripple granted them a lot of xrp to put on Stronghold'd exchange. At the time Xpring was solely focused only on expanding the xrp market, so that made sense. And also at the time, my understanding is that Stronghold removed the xlm/shx pair on that exchange and replaced it with xrp/shx. BUT, that exchange is dead now, the old URL resolves to their homepage. Further, they're not listed anywhere that I can find as a true Ripple investment; meaning Ripple gave them capital. There's no Form D filed with the SEC between Ripple and Stronghold (or Action Factory (Stronghold's founding company owned by Tammy)) and no official Ripple announcement. They're not even listed on the RippleX site where Ripple disclose's their investment portfolio. Does it mean that Ripple didn't invest? No, they could have in a more obscure way, but it could also mean that Stronghold is inflating the relationship and calling the xrp they received an investment/partnership... They could have been Ripple's partner in expanding xrp, but that deal could have been one-sided in Ripple's favor, and Stronghold only benefited from the exposure. Getting strong talking points is a valid growth strategy so who knows. The point is that without announcements from both sides (otherwise called "confirmation") or a paper trail that you can validate the claims yourself, you're leaving it to faith, and imaginations can run wild when you blindly believe what's claimed. Even still, Stronghold shows an impressive number of investors and advisors on their site. But again, that's coming from them. So what I have been able to track down is that Stronghold is based in SF, they have about 30+ people on LinkedIn who claim to work there full time (predominantly software engineers who live in New Zealand and a few other places), Stronghold is a Delaware entity (normal for privacy and taxes) that is owned by Action Factory, and their trademark is in fact registered and live. Action Factory and Stronghold both trace back to the same address in San Francisco. Both Action Factory and Stronghold's URL use identical CloudFlare name servers and that's a strong connection because those are unique to Cloudflare users. So the same person is managing both domains under one account. So at a minimum we know that Action Factory and Stronghold are the same entity both sitting at the same address at the time of inception. Doesn't mean they don't have a different corporate location now. There's no issue with that. Most recently they called their listing on Uphold as a partnership for institutions and tagged Uphold in the post. That's a brave move if not completely true, but those claims again have to be taken on faith. @UpholdMarkets (Uphold's marketing twitter account for their listings) did post about $shx, but they didn't mention institutions in that post. So we're back to faith there. I have never been able to confirm any of the investments or partnerships they claim outside of the now dead xrp moment with Ripple on their now dead trade[dot]stronghold[dot]co exchange. It's all Stronghold saying it, there's no confirmation from the other side. And the one everyone loves is the IBM trial in a closed loop environment. That was real, but it's long over now. Stronghold has moved away from Stronghold USD and their focus has shifted completely to leveraging other existing networks. IBM was a great name to tout for marketing, but the strength of that moment mattering has long passed for me. So my criteria won't let me over invest on what "might happen." I'd rather miss some upside than be completely wrong if they miss their shots or if they're "faking it until they make it" but never actually make it.

• 
  luna (@ImLunaHey) reported 58 minutes ago

  @llmDestructor cost in terms of what? i have 100s of sites on cloudflare and never got past the free limits lol

• 
  IntelliData Solutions (@Intellidsi) reported 1 hour ago

  1/ The "cloud is always up" assumption took three hits in five weeks last fall. The dates, the causes, the blast radius — then one question for your ops team: 2/ Oct 19–20, 2025: AWS US-East-1 goes down. Cause: a DynamoDB DNS race condition. Recovered ~6 p.m. ET Oct 20. 3/ Oct 29, 2025: Azure, 8+ hours. Cause: a faulty tenant configuration deployment. Alaska Airlines reported key systems affected. 4/ Nov 18, 2025: Cloudflare. Blast radius: ~40M live sites, 19M+ in the U.S. 5/ Different providers, different causes, same lesson: one region or one config push can be one point of failure. The question isn't if it recurs — it's what your team does in hour one. What's your plan? CTA — Reply prompt: "What's your hour-one plan?"