Hacker News new | ask | show | jobs
by lopkeny12ko 995 days ago
What's the point of having a secondary endpoint 1.0.0.1 if an outage breaks both that and 1.1.1.1? Are these two servers not running in physically isolated regions with independent code deploys?
3 comments

DNS servers are run anycast from each cloudflare POP. Why are the two addresses' fault domains not separated otherwise is the proper question.
Will DNS clients actually try the other if one of them returns SERVFAIL?
Yes.
routing redundancy