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 (42%)
• Cloud Services (24%)
• Hosting (18%)
• Web Tools (11%)
• E-mail (4%)

Live Outage Map

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

City	Problem Type	Report Time
Noida	Hosting	9 days ago
Jewar	E-mail	10 days ago
Braga	Web Tools	10 days ago
Noida	Cloud Services	11 days ago
Paris	Cloud Services	11 days ago
Prievidza	Domains	11 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:

• 
  ShopOS (@shoposai) reported 11 minutes ago

  We ran Big Head's GEO audit on India's top 10 DTC brands 🤯 Only 3 showed up in AI search at all. And even those 3 had critical gaps that were costing them citations. Here's what the audit found: → 7 out of 10 had robots.txt blocking AI crawlers entirely → 6 out of 10 had weak or missing brand entity signals → Even the 3 "visible" brands had no content mapped to actual customer prompts → None of the 10 had a repeatable system to track or improve their AI visibility These brands are spending serious money on ads and SEO. ChatGPT barely knows they exist. Perfect for DTC brands and agencies still relying on Google rankings and wondering why AI never recommends them. If you're publishing content, running ads, doing SEO — but your brand only shows up in 1 out of 4 AI engines when someone searches "best [your category] brand" — you have a gap costing you sales right now. This checklist eliminates the entire loop: → Phase 0: Remove blockers (robots.txt, Cloudflare, Bing index, page speed) → Phase 1: Audit your baseline AI visibility score → Phase 2: Restructure content so AI engines can actually extract it → Phase 3: Fix entity and brand signals so AI knows who you are → Phase 4: Map the exact prompts your customers are typing and build pages for them → Phase 5: Build topical depth and off-site presence → Phase 6: Track citations weekly and iterate No guessing why AI ignores your brand. No wasting budget on content AI can't cite. No watching competitors get recommended instead of you. What you get: → 32 prioritised signals with CRITICAL / HIGH / MEDIUM labels → The exact structural fixes proven to increase AI citation rate by 25–40% → A phase-by-phase sequence you can hand to your content team today → A reusable audit framework you run every quarter We put together the full checklist — all 32 signals, exactly what these brands were missing, and how to fix each one. Want it for free? Like this post Comment "GEO" And we'll send it over (must be following so we can DM)

• 
  Aspals Legal (@AspalsLegal) reported 14 minutes ago

  Have you noticed that more and more web pages are blocking access to anyone connecting via VPN? Why Pages Are Blocking VPNs Shared IP Flagging: Because thousands of Proton VPN users share the same outgoing server IPs, a site sees unusually heavy traffic from a single address. This often triggers automated security software (like Cloudflare or Akamai) to block it. Fraud and Bot Protection: IP addresses tied to commercial VPN data centers are routinely categorized as "anonymous" or "risky," leading websites to restrict access to protect against spam, credential stuffing, and fraud. Regulatory and Legal Pressure: New regional regulations, such as age-verification requirements in the UK and European data laws, force sites to actively restrict users attempting to bypass geographic and legal content filters. Advertising Revenue: Because VPNs mask user locations, they interfere with targeted advertising. Some websites also actively block the ad-blocking technologies built into VPNs to protect their revenue models. How to Bypass These Blocks To regain access, you can employ a few strategies to conceal your VPN footprint or route around the restrictions: ♦ Switch Servers: Simply disconnecting and connecting to a different server changes your exit IP address, which may not yet be blocklisted. ♦ Use Stealth VPN: Proton VPN includes a custom Stealth protocol designed specifically to bypass standard VPN blocks by making your encrypted traffic look like regular HTTPS data. ♦ Split Tunneling: Use your Proton VPN application's split-tunneling feature to route only specific apps (like your browser) through the VPN, or disable it for sites requiring direct access. ♦ Clear Cookies: Websites often store tracker cookies that link your browsing behavior to an IP. Clearing your browser cache can sometimes resolve access issues. [Thanks to Google]

• 
  Jack (@jackcoder0) reported 14 minutes ago

  The neighbor's final advice was the most actionable. He sat down and wrote out a list of 6 things every internet customer should do: 1. Turn off the public Xfinity hotspot (or your ISP's equivalent Spectrum, Optimum, and Cox all do this too) 2. Manually set your Wi-Fi channel instead of "Auto" 3. Disable QoS / Smart Network "optimization" features 4. Change your DNS to Cloudflare (1.1.1.1) or Google (8.8.8.8) 5. Buy your own modem and router, stop renting from the ISP 6. Test your speed with fast. com or speedtest. net using a non-ISP server, never trust your ISP's own speed test Total cost: $150-300 in equipment, paid back within a year. Total time: One afternoon of setup. Total impact: Often 2-5x improvement in real-world speeds. The customer went from paying $90/month for "fast" internet that crawled to paying $60/month for the same internet that finally worked.

• 
  Okabe Rintarou (@OkabeRintarou) reported 18 minutes ago

  Remove Your Media LLC, you're losers. You deserve such a hefty fine for copyright trolling that you’d never be able to pay it off in your lifetime. Imagine mass reporting of URLs on Google and cloudflare from my website featuring anime REVIEWS and NEWS and 0 illegal content.

• 
  Jean (@JohnStrongHodl) reported 19 minutes ago

  @sal_ash_ ok, will try, but as a fellow dev: it won't help preventing scraping unless you have some guard in place (maybe Cloudflare, would be better to ask AI) nor bots now onto checking out your tool, cheers

• 
  Crystalwizard (@crystalwizard) reported 20 minutes ago

  @PaulGugAI people on twitter: "AI can't code" me: (watches claude create an intricate and complicated set up out of thin air so GPT can connect via a cloudflare tunnel and run brainctl in its home directory on my machine) - which includes design docs, writing an MCP server, service, Cloudflare route, OAuth, database init, FTS rebuild, Defender exclusion, retry logic. All in one day. From nothing.

• 
  @banf (@banf) reported 21 minutes ago

  @msefaoruc @Cloudflare @CompaniesHouse Nice work abi!! Curious to hear your opinion, do you think officer data should be redacted from the open internet? It’s kinda a privacy issue imo

• 
  Matthew Prince 🌥 (@eastdakota) reported 22 minutes ago

  @prukalpa @zatlyn She is the best. Cloudflare would have never gotten out of the garage without her.

• 
  Bhargav Shivarthy (@bshivarthy) reported 24 minutes ago

  A lot of my internet habits still feel very human. I keep tabs open because I am afraid I will lose the thread. I send myself links I may never reopen. I reread the same page twice because I forgot why I came there. I ask someone, “do you remember where we saw that?” That is the web I grew up with. It was built for people trying to find, compare, remember, and decide. @eastdakota just pointed to Cloudflare Radar showing bot traffic passing human traffic for worldwide HTML requests. I think this is one of those moments we will point back to. Not because a line moved on a chart. Because the web started serving a new audience. Software is now a reader too. That can sound cold, but I do not think it has to be. Every time we scale an audience, we have to build new ways to support that audience. More readers means more surfaces, more formats, more infrastructure, more trust, more context. This is not zero sum. There will be uncertainty. There will be dislocations. Change always creates some discomfort before the new workflows feel obvious. But I do not think this means there is less to do. I think it means there is much more we can finally do. There are too many problems bottlenecked by attention, memory, monitoring, translation, coordination, and follow-through. Health needs more eyes on more signals. Science needs more ways to connect scattered work. Climate needs more systems that notice change early. Companies need better context. Governments need better feedback loops. People need tools that help them keep up. The next audience for the web will not only click and skim. It will watch, compare, trace, and act. That is going to change how information is published, structured, trusted, and maintained. Our 20s is not slowing down. If anything, it is just getting us warmed up.

• 
  Osinachi (@sin4ch) reported 26 minutes ago

  @krl_grn @ion_popsoi Oh I thought it was a DNS issue, until I tried Cloudflare WARP and it still didn't work. Looking forward to seeing the site.

• 
  BrenJ (@azzabazazz) reported 27 minutes ago

  @therealricoy A few cold-brew powered rambles: - When AI agents become the primary economic agents in a network decentralization becomes an inevitability (absent external threat against their substrate, too incendiary a subject to get into here). Current power economics of centralization benefit from having a human or "slow" entity to hold accountable for misdeeds and the frictions of maintaining verified intention execution. That won't be the case for AI agent swarms. They will be ephemeral, stateless, and operate at collective millions of TPS. That means their behavior will be practically impossible to contain in any 20th century sense of the word. Agents will choose to operate where their freedom to pursue their utility optimization is least impinged. If they can't find that environment either they or market forces will spawn it. Not taking a moral stance on this, nor declaring paperclip factories a fait accompli (thankfully they're not), just following economic logic to its conclusions. - This emergent world of "BlockchAIn" means that buy-in replaces buying as the primary economic modality of 21st century digital capitalism. Instead of building fortresses to protect and rent/sell/lend goods and services capital, AI agents will accelerate into surfers of capital waves - dropping in, carving, exiting (or wiping out). In a sense it marks the expansion of HFT into anywhere tokens/blockchains and agents/swarms converge. - Understandably, all of this will sound hand-wavingly academic and abstract until we painfully relearn why DARPA constructed the decentralized internet in the first place - antifragile redundancy of critical informational infrastructure. If @Cloudflare were to go down for the next month (perhaps somebody shatters their lava lamp wall) we would see, at very least, these two things occur: 1) mass economic losses 2) multiple solutions spring up to fill the informational network gap. Aka centralization --> decentralization. In AI's accelerating economic world, the push-pull-pull-pull tension grows between a) agent swarms (like a Mythos phalanx) maliciously attacking existing infra b) existing infra protecting itself with similar swarms from the inside out c) existing infra incentivizing white hat swarms to penetrate (and perhaps patch) before malicious swarms breach d) swarms spawning alternatives to current infra to outcompete. These force vectors essentially combine like a GAN into the aforementioned capital waves. Notably, in this hyperaccelerated world digital capital itself becomes infra. The simplest thought experiment proof: imagine agents running their own validators and chains to "own" trusted state calibrated to their specific needs. This is a microcosm of the future macro state. - Thus, TL;DR, humans aren't the "units of community" that will come to dominate blockchain (they're already less than half the traffic of the internet after all). The mission now, for the folks in this industry who think primarily in human rather than human capital terms, is to architect alignment between human society and the dawning emergent agentic community of communities (perhaps living on a chain of chains...).

• 
  Dr. Valerie Thomas (@Valerie32844654) reported 29 minutes ago

  I'm currently having significant problems with Cloudflare. Their lack of integrity to resolve customer issues is not reputable.

• 
  SandyFortune (@MLPSandy) reported 36 minutes ago

  My first OPNsense issue. There were supposed to be plugins for dynamic DNS, but each one has very recently gone "we're deleting everything except cloudflare, just move to that." What sort of patch note is that? You all JUST HAD everything, I'm not moving to cloudflare wtf

• 
  kate crisafi (@katecrisafi) reported 36 minutes ago

  This is honestly crazy. Cloudflare just shared new Radar data—bots and AI traffic now make up 57.5% of all HTML page requests on their network. Humans? Down to 42.5%. And Cloudflare handles about 20% of the whole internet, so this isn’t a tiny sample. Their CEO say

• 
  NTR of Pharaoh Khonshu Heru (@AtreyuBastian84) reported 36 minutes ago

  The number of AI agents and automated bots online has surpassed human activity, changing how the internet works. According to data from cybersecurity and network firms like Cloudflare and HUMAN Security, non-human entities now generate more global web traffic than real people.

• 
  Jack (@jackvebo) reported 37 minutes ago

  @PegasusPS5 Akia is not support it says now. Also you need to manually check the cloudflare

• 
  QFS17 (@riabcevv) reported 40 minutes ago

  remember talking about how ai coding is great, but the deployment phase is still a pain? well, openai is trying to fix exactly that. they just announced a feature called sites, designed to deploy projects directly from codex with zero server hassle. basically, you prompt an mvp into existence and go live immediately in the same interface. cloudflare and others have been making moves to support agents, but native deployment inside the llm environment is a game changer. the rollout just started, so most of us are still on the waiting list. but the direction is clear. we are moving fast toward complete end-to-end automation. anthropic is definitely working on a response to this for claude code. are we witnessing the end of traditional devops for small projects, or is this just hype? opinions? 💻

• 
  Rishabh Khandelwal (@AIrishabh) reported 42 minutes ago

  The AI infra story this week isn't another model launch. It's the token bill. Cloudflare adding spend limits to AI Gateway is the operator signal: agents are moving from "can it work?" to "can it run without silently torching budget?" Cost controls are now product safety.

• 
  Ansızın Olanlar (@ansizinolanlar) reported 43 minutes ago

  @oha1th3r3 @jpwexperience @Vultr Yeah, something weird is definitely going on. Sites using Vultr DNS are working, but the ones behind Cloudflare aren’t. However, Cloudflare-backed sites on other providers seem fine, so it looks like the Vultr network might be affected.

• 
  Dmytro Shevchenko 🇺🇦 (@dschewchenko) reported 48 minutes ago

  PreviewChecks got more scanner traffic than users today :) People already try /gcp-key.json and /firebase-adminsdk.json. Good news: Cloudflare Workers do not keep my secrets in public files. Bad news: they still keep trying.

• 
  TO1 (@Viraltbh) reported 48 minutes ago

  @compileandpush mostly sidestepped it — Cloudflare D1 + raw SQL on the web side, plain sqlite3 for the python bots. never hit the async ORM wall because nothing's high-concurrency yet. what'd you land on — asyncpg raw, or did you find an ORM that doesn't fight you?

• 
  Ronan Berder (@hunvreus) reported 50 minutes ago

  @sidpalas @tonyennis But you don't support Cloudflare. Why's that? I'm trying to understand why you'd pick Flue; for me the main advantage is serverless. Otherwise, running pi directly is easier..?

• 
  Virtually Fun (@virtuallyfun) reported 51 minutes ago

  @OneCloudEmoji I got pissed when blogspot went down for a week, I'd been self hosting on wordpress for like 15 years now? A simple VPS + cloudflare and you're good to go. Plus you 100% own your content. I've moved hosts dozens of times, even at one point hosting at home with WSLv1 pptp'd to a VPS.. it's the best/most flexible. I'm tempted to move to a cellphone using usermode at some point, more so just because I can..

• 
  Mikey (@MadMikeyB) reported 54 minutes ago

  @iBotPeaches Sorry to hear about this, we've had to do similar with CloudFlare WAF and Rate Limits because of the same issue.

• 
  michelle hayes williams (@michellehayesw2) reported 57 minutes ago

  This is honestly crazy. Cloudflare just shared new Radar stats — bots & AI traffic now make up 57.5% of all HTML page requests on their network. Humans? Down to 42.5%. Remember, Cloudflare handles about 20% of the whole internet. So this is huge. Their CEO says the

• 
  TheWrath0fKahn (@TheWrath0fKahn) reported 1 hour ago

  @AlternativeTo @marttimalmi I've been testing it on android, and so far I can say that it's literally the only VPN that reliably stays connected (in my case to cloudflare warp) and never silently disconnects or closes overnight etc. Literally no other VPN on android in my testing has ever done this. I cannot speak to the mesh aspect because currently the GUI version doesn't work on Windows, but the android version so far is bulletproof.

• 
  Ajay Kidave (@ajay_kidave) reported 1 hour ago

  @zebassembly @championswimmer Thanks for the explanation. The fact that there is a new container service from Cloudflare means others have faced the same issues. The container based services do not have to served from all the edge locations. Something like a hub and spoke model would be good enough. That way you are not limited on compute (once compute prices hopefully go back to sane levels)

• 
  nosmh (@nosmh) reported 1 hour ago

  Bright Data runs the world’s largest residential proxy network by sneaking an SDK (programming code) into everyday stuff: CTV games on smart TVs, messengers, mobile apps, and screensavers. It turns your gadget into an exit node—routing web-scraping traffic through YOUR real home IP to dodge blocks from Cloudflare, DataDome, and HUMAN. Smart TVs (Samsung, TCL, LG, etc) are their favorite: always plugged in on WiFi, usually ignored. “Consent” is buried in privacy policies or a clunky remote menu. SDK flags let it run with the screen off or while you’re on a call. Your devices become silent data mules for big scrapers—eating your bandwidth and hiding behind your home address. Regular folks are getting used without a clue. 👇

• 
  MossGossip (@moss_gossip) reported 1 hour ago

  @xai @Cloudflare And why don’t you have a customer service department? You take our money and leave us hanging.

• 
  Image Offload (@Imageoffload) reported 1 hour ago

  @Cloudflare Never expose storage URLs directly. Route everything through a Worker or proxy that validates the request first. Presigned URLs with short expiry kill hotlinking and scraping abuse before it starts. Built this into ImageOffload for R2-hosted media.