What is proper fault-tolerant behavior?

[Dan McDaniel]

Here's the scenario. Last week one of the Fedora nameservers was not
responding so I was having trouble resolving fedoraproject.org.
Apparently my company's nameserver was sending out the request,
receiving the list of nameservers from the root and then querying the
failed fedora NS. Then it came back with a not found. My company's
networking group said it's the fault of the bad fedora NS.

My question is shouldn't our nameservers try another fedora NS in order
to resolve the name? If not what good is it for fedora to have multiple
nameservers? Or am I misunderstanding how this should work?


-- 
Dan