Earlier quoted context omitted.
Sounds like automation deployed a configuration update to most of Facebook's peering routers simultaneously. Something similar brought down Google in 2019.
If so, then it would simply be a BGP issue - no FB servers reachable, as all routes are down. But media+claims a combo of BGP/DNS. Hard to believe world-wide border routers, only responsible for networks containing DNS servers, are misconfigured. I am rely curious about that post-mortem :)
I think it was only a BGP issue. The DNS servers apparently shared the same peering routers as the rest of Facebook's infrastructure. Everyone focused on DNS because that's the first sign of failure to an end user.