3 ms·
The media coverage and lots of the comments don't make sense to me. FB would not be so stupid and put all of their crucial DNS servers into a single autonomous
by littlecranky67 5y ago
The media coverage and lots of the comments don't make sense to me. FB would not be so stupid and put all of their crucial DNS servers into a single autonomous system (which is now offline due to BGP issues). They operate literally dozens of datacenters around the world, and are surely not using a single AS for them - why not put secondary Nameservers there?
Can someone make a sense of this?
- mdavidn 5y agoSounds like automation deployed a configuration update to most of Facebook's peering routers simultaneously. Something similar brought down Google in 2019.
- littlecranky67 5y agoIf 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 :)
- mdavidn 5y agoI 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.