Hi! I work as a Support Engineer for a company which provides web hosting: Netlify. Our customers are reporting issues with sites hosted on our service when they are using the DNS resolver
18.104.22.168. To cut to the chase, I’m hoping to contact Cloudflare support so we can troubleshoot this issue affecting people using both Netlify and 22.214.171.124.
At Netlify, we also provide a CDN with our service. When a query is made for a Netlify site, our DNS servers will response with the IP address for a CDN node which is geographically closest to the IP address making the request. This is done using the Geographic Routing feature provided by the DNS service we use which is NS1.(
https://ns1.com/geographic-routing is the link to the description of Geographic Routing. I’m including it as a code block because my new forum user is only permitted to links per post.)
What normally happens in this. A user makes a DNS query for a site. If you are in Paris, we will return an IP address for a CDN node in Frankfurt. If you are in Melbourne, we will return an IP address for a CDN node in Syndey. The IP address returned is dependent on where the query comes from.
However, if a person uses
126.96.36.199 as their resolver, it always returns IP addresses in the U.S.A. The queries don’t send the person to their closest CDN node anymore. 188.8.131.52 breaks the geographic routing.
While this is less than ideal for people in the western U.S.A., it results in particularly terrible site performance for people on other continents. Being routed to New York instead Sydney when you are browsing from Australia results in slow pageloads (very high TTFB).
Here is an example of such a report from our own community site:
The topic above was a person in Spain being directed to a CDN node in North America instead of Europe. The resolver that was returning this IP address was 184.108.40.206.
This issue doesn’t occur when people use the DNS resolvers for their ISP or other “public use” resolvers like 220.127.116.11.
Would it be possible to create a support ticket so we can discuss why this is happening and what can be done to resolve this issue? I’m able to spin up EC2 instances or use VPN software to demonstrate the differences when 18.104.22.168 is used around the world compared to other resolvers.
It would seem ideal to take this conversation private so we can troubleshoot this. However, if you would prefer to troubleshoot here publicly, I’m happy to do this as well.
If there are any questions I can answer, please let me know.