We've gotten rate-limited out of the blue on clustered development servers in the past 3 years now, this last one was on servers we setup 830 days ago, before we knew that getting rate-limited/banned on DNS servers where even possible. The worst thing about the last incident was that we entered a death spiral, DNS resolution failing started a logging job, that failed (due to DNS resolution failing to call log server) that then started a job about the failing DNS resolution.. You get the gist..<p>Of course, this is an issue of engineering and code, not only a rate-limiting issue.<p>However, many developers rely and depend on upstream DNS resolution to "Just Work" when you add it to a server, which has been the case with Googles DNS servers for the past 15+ years that I've been a sysop. I'm just hoping that this time, this will get SOME attention, because either you want dev-ops to use Cloudflare DNS on servers or you don't - and if you don't - there should be an official warning that this WILL happen, you WILL get rate-limited eventually.
> However, many developers rely and depend on root DNS resolution to "Just Work" when you add it to a server<p>As a sysops you're probably aware that neither Google nor CloudFlare are DNS root servers.<p>Using actual root servers through your own resolver would have avoided this issue. Bind doesn't even need any config for that use case.
What kind of volume was this? I have a server that does some rather specific DNS monitoring resulting in millions of <i>unique</i> lookups with 1.1.1.1 a day.