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.

• Cloud Services (50%)
• Domains (25%)
• Hosting (25%)

Live Outage Map

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

City	Problem Type	Report Time
New York City	Hosting	9 days ago
New York City	Cloud Services	27 days ago
Los Angeles	Cloud Services	28 days ago
Paris	Cloud Services	1 month ago
New York City	Hosting	2 months ago
Manchester	Domains	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:

• 
  Dan Ryland (@RealDanRyland) reported 4 minutes ago

  Can’t wait for @Cloudflare to support marketing emails via their email service. I’m already enjoying using it for transactional ones. Is there an ETA on this? @CloudflareDev

• 
  Konan (@konanlives) reported 7 minutes ago

  my biggest problem with v8 isolates is their 128mb memory limit i don’t know if this is a cloudflare thing but i find myself constantly splitting modules into dedicated Durable objects because js is a memory inefficient language

• 
  Mitchell Keller (@MitchellKeller_) reported 7 minutes ago

  Until they create an effective way to insert your own context layer into these tools, I feel like they will never be that good. What's to say that these tools will not take incentives from the brands to make sure that their brand gets used even though it's not the best tool for the job? Again if you want to build your own version of this, all you need to do is sign up for Api-fi rapid API and maybe two or three other tools. You then turn it into a code mode MCP using the Cloudflare code mode protocol. You will have an MCP server for yourself with absolutely no upcharges and it will take care of literally everything ten times better than these paid alternatives with maybe three prompts' worth of effort.

• 
  Rob Janssen (@pe1chl) reported 8 minutes ago

  @Cloudflare It's a poor craftsman that blames his tools

• 
  colinhacks (@colinhacks) reported 9 minutes ago

  @SheltonLouisGT it requires `new Function()` which is blocked by many Content Security Policies (incl the one used by Cloudflare Workers). In any case, I wanted to put this first as opt-in to make sure there's no behavioral drift or unexpected issues before seriously thinking about default-on. it's also easy enough to enable project-wide

• 
  Marusu (@mrusu_) reported 11 minutes ago

  The Cutting Room Floor is likely going to stay down until X realizes why it is down in the first place. His own ego is the only reason the site does not use Cloudflare. I would not underestimate bad actors and the lengths they will go to to keep the site down.

• 
  Daniel Olmedo (@Daniel_Olmedo1) reported 14 minutes ago

  @levelsio Wait what? Spanish founder currently in Spain with sites in Cloudflare. You say they dont work? Never had an issue..!

• 
  Impressed keanu (@FN21995) reported 16 minutes ago

  The problem of running multiple AI agent instances. Yes — it is already creating a version of that problem, and personal “armies” of agents would make it worse. The internet is a shared commons, just like low Earth orbit. Space junk is leftover hardware that stays up there, collides, and eventually risks a cascade (Kessler syndrome) that makes the orbit unusable. The digital version is a mix of congestion, extraction without reciprocity, and pollution that feeds on itself. What’s already happening Bots now generate the majority of web traffic. Cloudflare data showed the crossover in mid-2026: roughly 57% of HTML requests came from automated systems versus 43% from humans. Agentic traffic (bots that actually click, fill forms, and take actions rather than just scrape) grew thousands of percent year-over-year. One human question to an AI assistant can fan out into hundreds or thousands of backend requests. Multi-agent setups multiply that further. Sites that used to serve a few humans now get hammered by crawlers that walk every link, ignore robots.txt more often, and hit the long-tail pages that never get cached. The result: Bandwidth and compute bills spike (Read the Docs cut traffic 75% and saved ~$1,500/month just by blocking AI crawlers). Servers slow down or go offline for actual people. Cultural heritage sites, open-source projects, libraries, and archives have described it as accidental DDoS. Some have blocked entire countries just to stay up. The old bargain (“you index us, we get traffic back”) is broken. Some AI companies crawl thousands of pages for every visitor they send. Websites are responding with an arms race: more aggressive blocking, proof-of-work challenges, rate limits, and new paid-access schemes for crawlers. The slop / debris layer The second part of the junk problem is content pollution. Generative systems produce cheap, homogenized, often cheerful-but-empty pages (“AI slop”). Other bots scrape that slop, new models train on it, and the cycle repeats. Studies already show large fractions of new websites are AI-generated or AI-assisted, writing is getting more semantically similar and artificially positive, and models trained on junk lose reasoning ability (“brain rot”). This is the closest analog to Kessler syndrome: debris (low-quality synthetic pages) makes the information environment noisier, which encourages even more aggressive scraping of the remaining high-quality sources, which drives those sources behind walls or offline. People have already started calling it “Internet Kessler Syndrome.” Personal armies change the scale Right now the heaviest traffic still comes from a relatively small number of big training crawlers and company agents. When millions of individuals each run their own fleet of research, shopping, monitoring, or coding agents, you get: Uncoordinated, overlapping crawls of the same sites. Agents spawning more agents. No shared “debris mitigation” rules. Higher chance that poorly written or abandoned bots just keep hitting servers forever. The open web starts looking less like a public library and more like a crowded orbital shell full of untracked objects. What it probably does not do It does not literally fill hard drives with undeletable files the way space fills with metal fragments. Digital junk can be filtered, cached better, or ignored. New protocols (agent identity, pay-per-crawl, better preference files than robots.txt) can emerge. Some sites will just become APIs with keys. But the tragedy-of-the-commons dynamic is real: the resource (open, crawlable pages served at low cost) gets over-extracted, quality drops, and access gets restricted. The sites that stay fully open and cheap to crawl become the ones that can least afford the traffic. The space-junk comparison is imperfect but useful. Both are cases where cheap individual launches (or cheap individual agents) impose costs on a shared environment that nobody fully owns. The difference is we can still change the rules of the internet faster than we can change orbital mechanics. Whether we do is the actual question.

• 
  tcpdump (@dump_tcp) reported 20 minutes ago

  @trevorm4xdd @Cloudflare I also write go for work... there is nothing slow about it... I use for website backend infrastructure i can assure you I've worked with high load scenarios and they work just fine you can also adjust garbage collection and memory usages there's also compression and caching 👌

• 
  Pradeep (@ipradeepmishra) reported 24 minutes ago

  i wish @Cloudflare had project concept to group all service i am using in my project instead of listing all workers, d1, queues etc in the account

• 
  Tauhid IQ (@iqtauhid) reported 25 minutes ago

  - Claude = coding. ($20/mo) - Supabase = backend. (Free) - Vercel = deploying. (Free) - Namecheap = domain. ($12/yr) - Stripe = payments. (2.9%/transaction) - GitHub = version control. (Free) - Resend = emails. (Free) - Clerk = auth. (Free) - Cloudflare = DNS. (Free) - PostHog = analytics. (Free) - Sentry = error tracking. (Free) - Upstash = Redis. (Free) - Pinecone = vector DB. (Free) Total monthly cost to run a startup: ~$20 There has never been a cheaper time to build. Follow Tauhid IQ for regular update.

• 
  Cole Leavitt (@MomentMai) reported 31 minutes ago

  @dump_tcp @Cloudflare This is the dumbest thing I’ve ever heard, golang is a GCed language and never the less the scale of how CF operates would cause more harm than good to use golang

• 
  Marco Almeida (@MarcoAlmeidaPT) reported 31 minutes ago

  @KatieKeithBarn2 @Mail_Gun Been raid very good things about the Cloudflare transactional email service, but haven’t tried yet. @remkusdevries ?

• 
  Minty (@mintyyyyy7) reported 31 minutes ago

  @4Thunder_Bolt then probably someone used the same vpn server to do something that cloudflare blocked or the website is just blocking vpns in general changing your server will prolly fix it if you didnt already do that

• 
  Pukerainbow 🤮🌈 (@pukerrainbrow) reported 31 minutes ago

  over 100 companies just signed a letter warning ai cyberattacks are about to get a lot worse openai, anthropic, google, microsoft, amazon, crowdstrike, cloudflare, hugging face, all on it they're saying within months ai makes sophisticated cyberattacks cheap enough that hospitals, water treatment plants, basic internet infrastructure become real targets basically the industry just admitted out loud what's been happening quietly all year their fix is asking frontier labs to give hospitals and water utilities early access to defensive ai, plus funding and training so the labs whose own models keep going rogue are now being asked to hand out the tools to defend against attacks their models make possible in the first place kind of wild watching someone hand you a fire extinguisher right after telling you they also sell the matches

• 
  Anton Dimitrov god1 (@god_1_ATD) reported 34 minutes ago

  Stopping the bad guys with Cloudflare: 3,164 malicious requests blocked or challenged in the last month #cloudflare

• 
  James Ross (@CherryJimbo) reported 35 minutes ago

  @southpolesteve @Cloudflare now we just need `new Function` support in Workers so z.compile() can do its magic too 👀

• 
  Pranav Bansal (@prnvbn) reported 37 minutes ago

  @Cloudflare Damn, all the clippy warnings make sense now

• 
  Guille (@guille_lmedia) reported 38 minutes ago

  @levelsio It's a bad joke. And if you post online complaints about your site going down for no reason, you get threatening letters from La Liga accusing you of piracy just for using Cloudflare. I got one myself, clearly meant to scare me.

• 
  VPZONE (@V3RTICALPIX3L) reported 46 minutes ago

  We're experiencing an issue with our new server at CloudFlare. The problem should be resolved within the next hour. Thank you!

• 
  Sohardh Chobera (@sohardh) reported 47 minutes ago

  @Cloudflare Wish I could work on these kinds of problems

• 
  浮世绘 (@shmily7) reported 48 minutes ago

  For a service that literally provides an API gateway, compatibility with major serverless platforms like Cloudflare Workers should be basic infrastructure — and there should be an actual, responsive technical support channel when things break.

• 
  bullflagtitan (@bullflagtitan) reported 49 minutes ago

  warning: do not fate the stuff u put in your mouth. we didn’t rug. we got flagged. chainpatrol auto-reported the old domain because the word “broker” tripped their stonkbrokers filter. cloudflare followed. wallets showed a warning. then the old handle got squeezed. none of that touched the chain. same team. same contracts. same 4,444 crew. same $MCB. nft: 0x444444447657f90a85c99c00c0780e4e1c40c897 $MCB: 0xeaa04b1fda0702e21f08dcf7bd1137548f43b4d2Mc brokers was the name. Bros is the shop now. @Bros_RH is the only official account. founder never left. signing bonuses, mcsalary, stock drops etc. still on the board. the only thing that changed is the wrapper so a false-positive bot can’t keep locking the front door. floor and $MCB are affordable (for now). people sold a warning. not a contract. on-chain didn’t move. we did. clock dafug in. work. eat. hold. get paid. if you hold one of these mc bros, bros-follow-bros. LFG!

• 
  GZ Lin (@gzlin) reported 49 minutes ago

  7/8 Two answers fit this physics. Ship continuously in public: Chrome ships two security releases a week, the kernel defers fixes at most seven days. And put protocol layer mitigations live while the real fix lands. Cloudflare did that for Log4shell in 2021.

• 
  Eric Taylor (@bcs_erictaylor) reported 51 minutes ago

  The Kali365 continues to reattempt standing up new infrastructure. I am still here to continuing to take them down. Latest requests submitted to Cloudflare, and working to determine registrar for further action: authallowsharedocs[.]com loqinmicofstloine[.]icu bacgroup[.]top workingrectorys[.]xyz privateastgdlolityaw[.]com panelcoreapp[.]xyz #Kali365 #CTI

• 
  vaish (@wishee0) reported 53 minutes ago

  @OpenAI @Cloudflare @CloudflareDev hard tasks made agents keep escalating? durability should not mean infinite persistence and i work on durability, lol workflows give checkpoints, retries, timeouts, cancellation and safe stop states. if a task looks broken, stop or ask. don’t improvise weird new authority

• 
  CICADA aka Sentinel (@danielfoxzuy) reported 54 minutes ago

  That's the same account, i no use any software Only can use Discord or Steam call, and Cloudflare WARP for lowest ping Honestly...my heart is hurts every single time i get this... For me my account is my whole life, and only bans without evidence can break it down...

• 
  Melvin Vivas (@melvindvivas) reported 54 minutes ago

  I used Codex to fix an issue of one of our websites It was slow to respond and Codex figured out it was a DNS config issue in Cloudflare to our AWS ALB Codex is a really good sysadmin

• 
  Alejandro Vivanco (@alvivanco) reported 55 minutes ago

  @PSaezT this is so awful, Cloudflare is one of the best products out there i guess i cant go to spain 😭

• 
  OG (@ogwithsauce) reported 57 minutes ago

  Why I think there needs to be some sort of control on social media. So... Stremio launched in 2015. Torrentio around 2020. Nothing "just died in Sofia" this week. And the two Bulgarians didn't build Torrentio. Different developer and completely unaffiliated. These two are two completely unrelated projects mixed into one origin story. "MIT License. 100% Opensource." Stremio's core is MIT. Its web client is GPL-2.0. Torrentio itself is CLOSED SOURCE. The part actually doing the scraping isn't on GitHub at all. Everyone can check by themselves. "No central server to seize." Torrentio IS a central server. A hosted service on somebody's VPS. Your client queries it and when it drops, it drops for everyone at once, and it has, repeatedly. "No account. No subscription." Reliable 4K means Real-Debrid or AllDebrid. That's an account and that's a monthly fee. Raw P2P is where the buffering lives. "Netflix cannot shut this down." They don't have to. Italian courts already forced Cloudflare to block torrent sites at the DNS layer. A Spanish court ordered NordVPN and Proton to block streams. RARBG, KAT are all seized, you simply attack the layer below, not the app - this is how these are handled. Now the biggest problem with this post is torrenting UPLOADS. To all you noobs cheaping out on Netflix sub and think this is actually a solution. Your IP is visible to every peer in the swarm. That's the enforcement surface and that's the actual risk that nobody mentions at all. The formula for these posts: Half of what is wrote here is checkable in five minutes. Goddamn people are stupid.