Cloudflare explains outage that took down many websites

Sinbad said:
Lazy useless idiots who use generative AI to do their coding work for them.
Click to expand...

And then fight with people when they tell them their AI code is shyte and does stupid things that will never be allowed in production.
 
The other issue is that the entire internet has become very centralised. Cloudflare, Amazon and Azure.
If any of those three go down, a good portion of the internet gets cooked.
 
My_King said:
Never deploy to Prod on a Friday
Click to expand...
Some sites are great to deploy for on a Friday - low traffic. Like for example Sage. Like they did a few months back. Their problem is - they do not have support over weekends so their system was down for a whole weekend until some support agent finally decided to answer their phone sometime after 9am on the Monday. (I personally am still waiting for my support email to Sage to be answered which has been sent two years ago this month - I even still have the reference number I got back).

Friday updates are fine, just have real support over weekends.
 
Can someone explain in a non-technical way what Cloudflare does and why it affects so many websites?
 
Ďąhristopher said:
Can someone explain in a non-technical way what Cloudflare does and why it affects so many websites?
Click to expand...

Non-technical is a bit hard.

But in the simplest terms for the relevant products here (they have a huge variety of different product under one brand umbrella) they run local cache nodes (proxies) at your nearest point of presence (data centre) which means even if the real web-site is hosted in America you access most of it as if it was hosted locally.

Effectively they make the internet much faster than it really is from an end user perspective.

There is all sorts of other stuff that works inside those cache nodes but that’s the most basic part of it.

So they’ve removed the need for most websites to go do this themselves and host in multiple locations. They can get away with only a few distributed hosts they own and hand over the rest to Cloudflare which then self manages the traffic on their own network.

Say you connect to Walmart (US) in the old days you’d suffer 200+ms latency at a minimum and also put massive load on their own servers.

Now instead you are in Cape Town you’d connect to the local node in your nearest Data Centre for <30ms and it will balance this load accordingly as well. So if there is a need to fetch something from the real host in America that all happens on Cloudflare’s “internal” optimized and usually fastest network and not relying on your small ISP at the bottom of Africa to figure it out.

A version of this you can see in action for yourself is their own DNS server 1.1.1.1 which has the same IP to all users all across the world, but it’s not really real or direct and every person using it gets <25ms latency to it even though the actual IP lives in America, because everyone just connects to the locally cached version of it closest to them in routing proximity.

If we take Cloudflare away the internet immediately becomes a lot slower and more expensive for everyone.

Some moderately light reading - https://developers.cloudflare.com/fundamentals/concepts/how-cloudflare-works/
 
Last edited:
SauRoNZA said:
Non-technical is a bit hard.

But in the simplest terms for the relevant products here (they have a huge variety of different product under one brand umbrella) they run local cache nodes (proxies) at your nearest point of presence (data centre) which means even if the real web-site is hosted in America you access most of it as if it was hosted locally.

Effectively they make the internet much faster than it really is from an end user perspective.

There is all sorts of other stuff that works inside those cache nodes but that’s the most basic part of it.

So they’ve removed the need for most websites to go do this themselves and host in multiple locations. They can get away with only a few distributed hosts they own and hand over the rest to Cloudflare which then self manages the traffic on their own network.

Say you connect to Walmart (US) in the old days you’d suffer 200+ms latency at a minimum and also put massive load on their own servers.

Now instead you are in Cape Town you’d connect to the local node in your nearest Data Centre for <30ms and it will balance this load accordingly as well. So if there is a need to fetch something from the real host in America that all happens on Cloudflare’s “internal” optimized and usually fastest network and not relying on your small ISP at the bottom of Africa to figure it out.

A version of this you can see in action for yourself is their own DNS server 1.1.1.1 which has the same IP to all users all across the world, but it’s not really real or direct and every person using it gets <25ms latency to it even though the actual IP lives in America, because everyone just connects to the locally cached version of it closest to them in routing proximity.

If we take Cloudflare away the internet immediately becomes a lot slower and more expensive for everyone.

Some moderately light reading - https://developers.cloudflare.com/fundamentals/concepts/how-cloudflare-works/
Click to expand...
I think more importantly they allow you to remove your services from the "visible" internet and they proxy them for you. So you can't get DDoSd - and they also have managed WAF rules in place so that you're protected from many forms of attack.
 
Sinbad said:
I think more importantly they allow you to remove your services from the "visible" internet and they proxy them for you. So you can't get DDoSd - and they also have managed WAF rules in place so that you're protected from many forms of attack.
Click to expand...

Yeah I tried to keep it simple and less nerdy. 🤣
 
And again..

bad gateway.png


This went on for about 5 minutes, just came right now.
 
Thanks @cavedog

screenshot-2025-12-19-135236-png.1872331
 
Another problem session, at least 20 minutes , just ended. Couldn't reach the forum via any VPN server I tried worldwide.
 
Brian_G said:
Another problem session, at least 20 minutes , just ended. Couldn't reach the forum via any VPN server I tried worldwide.
Click to expand...

Yeah still getting errors but they say the outage is resolved :unsure:
 
BTW, this last time I was getting endless-loop Cloudflare human checks, every time.

But was able to reach the MyBB speedtest, always can.
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X