Cloudflare Outage: How to Check Status & What to Do
When multiple sites go down with 500 or 502 errors showing Cloudflare's logo, it's likely a Cloudflare outage. Here's how to check status and what you can do.
👁 122 reads
When several major websites suddenly stop working and you see a 500 or 502 error with the Cloudflare logo, it’s most likely a Cloudflare outage, not your internet connection. You can check this in a minute at cloudflarestatus.com, and as a user, you can only wait - changing DNS only helps when the 1.1.1.1 resolver itself is down. Below we’ve gathered how to recognize an outage, what to do, and the history of major outages with dates. For more about the company, see our Cloudflare guide.
How to check in one minute, as of October 5, 2026
- Official status: cloudflarestatus.com - list of incidents and status of individual locations.
- User reports: downdetector.com, search for “Cloudflare”.
- Network traffic: radar.cloudflare.com - Outage Center.
- Your location: the address
https://1.1.1.1/cdn-cgi/traceshows in thecololine which data center you’re connecting through to Cloudflare.
How to recognize a Cloudflare outage
Cloudflare intermediates traffic to about one quarter of websites on the internet (W3Techs: 26.5% of all sites in October 2026), so its outage looks like “half the internet is down”. You’ll recognize it by several signs:
- Errors on many unrelated sites at once - for example, ChatGPT, X, Discord and an online store simultaneously.
- Cloudflare error page with three icons - Browser, Cloudflare, Host. If the middle one (Cloudflare) is red, the problem is on Cloudflare’s side. If the Host icon is red, that individual site’s server is down.
- 500 and 502 codes with “Cloudflare” written at the bottom of the page and an “Internal server error” message.
- Verification screens that don’t complete, because the Turnstile service is also part of Cloudflare’s network. A single verification loop is usually caused by something else - we described this in our article about Cloudflare verification.
If the error only affects one site and has a 520-526 code, it’s usually that site’s server problem, and Cloudflare is just reporting it.
What you can do as a user
- Check cloudflarestatus.com. If there’s an incident, further tinkering with your router won’t help.
- Don’t reset your router or call your ISP. Your connection is working - the intermediary between you and the site isn’t.
- Change DNS only during a 1.1.1.1 outage. If you have 1.1.1.1 in your settings and no sites open (even those without Cloudflare), temporarily enter 8.8.8.8 or your ISP’s DNS. This happened on July 14, 2025, when the 1.1.1.1 resolver didn’t work worldwide for 62 minutes.
- Use apps instead of websites. Some mobile apps connect to servers bypassing Cloudflare and continue working.
- Wait. Most outages in recent years lasted from 25 minutes to a few hours.
What a site owner can do
- Don’t panic-change your domain nameservers. DNS propagation takes up to 48 hours, so the outage will end before the change takes effect, and you’ll be left with a messed-up configuration.
- Disabling proxy (gray cloud) may help, but it exposes your server’s real IP address and removes protection against attacks. Besides, during a major outage the Cloudflare panel may also not work - this happened on November 18, 2025.
- Have a status page outside Cloudflare so you can inform customers.
- For large services consider multi-CDN or DNS with two providers. For a blog this is overkill - the cost exceeds the loss from a few hours of downtime once a year.
Major Cloudflare outages - dates and duration
All data comes from post-mortem reports that Cloudflare publishes on its blog. Times are given in UTC.
| Date | What happened | Duration | Scope |
|---|---|---|---|
| July 2, 2019 | poorly written regular expression in firewall rule consumed 100% of CPUs | 27 min | entire network |
| July 17, 2020 | router configuration error in Atlanta directed all backbone traffic there | 27 min | some locations, traffic drop of ~50% |
| June 21, 2022 | network configuration change in 19 busiest data centers | ~75 min | 19 data centers |
| June 12, 2025 | Workers KV storage failure, dependent on external cloud | 2 h 28 min | Workers KV, WARP, Access, Turnstile, Workers AI and others |
| July 14, 2025, from 23:52 | configuration error in legacy IP address announcement system | 62 min | 1.1.1.1 resolver worldwide |
| November 18, 2025, from 12:20 | permission change in ClickHouse database doubled Bot Management function file, servers crashed | main traffic returned after ~3 h 10 min, full restoration after 5 h 46 min | entire network, worst outage since 2019 |
| December 5, 2025, from 9:47 | change in request body parsing while patching React Server Components vulnerability | ~25 min | ~28% of HTTP traffic |
| January 22, 2026 | BGP route leak from router in Miami | 25 min | IPv6 traffic through Miami |
| February 20, 2026, from 18:48 | change in BYOIP address handling withdrew ~1100 customer prefixes | 6 h 7 min | customers with own IP addresses, 1.1.1.1 site |
| May 5, 2026 | incorrect DNSSEC signatures in .de zone (fault of DENIC registry, not Cloudflare) | for 1.1.1.1 ~2 h 47 min (until 0:17 UTC) | .de domains in validating resolvers, including 1.1.1.1 |
Time to full service restoration (minutes, selected outages)
In addition there are smaller incidents. According to industry service compilations on the status page, between August 7 and 14, 2026, there were 13 incidents (R2, Durable Objects, Workers KV, Workers AI, email), of which 12 were marked as minor. We haven’t verified this in the status archive, so treat this as secondhand information.
November 18 and December 5, 2025 outages - what happened
These two dates are most frequently searched on Google (“cloudflare outage december 5”, “cloudflare outage 2025”). On November 18 from around 12:20, ChatGPT, X, Canva and many online stores and services with Cloudflare panel stopped working. Cloudflare CEO Matthew Prince wrote that it was the company’s worst outage since 2019. The cause wasn’t an attack, but a permission change in the database that caused the bot detection system configuration file to grow twofold and exceed the software limit on servers.
On December 5 morning the outage was shorter (about 25 minutes), but affected 28% of all HTTP traffic handled by Cloudflare. This time the culprit was a security patch introduced hastily after disclosure of a vulnerability in React Server Components. Cloudflare admitted directly that they “failed the internet again” after the November 18 incident.
Is Cloudflare less reliable than before?
In our view, it’s not so much less reliable as more visible. In 2019 Cloudflare handled a much smaller portion of the web, and today its share of websites exceeds 26%, so each outage disables more things you use. The common denominator of recent outages isn’t attacks, but configuration changes rolled out to the entire network at once. After the late 2025 outages, the company announced changes in how it rolls out configurations, so errors don’t hit all servers at once. For the average user the conclusion is simple: once or twice a year, a few hours without part of the internet is now a realistic scenario, regardless of whether you use Cloudflare knowingly.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →Sources
- Cloudflare - status page
- Cloudflare - July 2, 2019 outage
- Cloudflare - July 17, 2020 outage
- Cloudflare - June 21, 2022 outage
- Cloudflare - June 12, 2025 outage
- Cloudflare - 1.1.1.1 outage, July 14, 2025
- Cloudflare - November 18, 2025 outage
- Cloudflare - December 5, 2025 outage
- Cloudflare - route leak January 22, 2026
- Cloudflare - February 20, 2026 outage
- Cloudflare - .de domain outage, May 5, 2026
- CoderCops - 13 Cloudflare incidents in August 2026
More on this from QUANTSLANT
ai tools·What Is Cloudflare? Free Plan, Pricing, AI Tools, and Bot ControlsA complete guide to Cloudflare in 2026: what the Free tier includes, official USD pricing, Workers AI, AI bot crawler controls, 1.1.1.1 DNS, and outages.