Understanding rndc referral statistics

[Chris Buxton]

On Mar 29, 2013, at 8:13 AM, M. Meadows wrote:
> Thinking about this ... perhaps this is more to do with the behavior of BIND 9.3 versus BIND 9.7. Did the referral mechanism change? Here are my thoughts on the subject:
> 
> Nameserver A is the authority for zone1.com and it is the authority for sub.zone1.com. Sub.zone1.com is delegated from zone1.com. If a query comes to nameserver A from a resolver asking for info about host.sub.zone1.com and the namserver looks in zone1.com and sees the delegation of sub.zone1.com an inefficient method of handling the query would be to pass back a referral to sub.zone1.com (which just points back at itself). But that would work and would result in a referral. In a more efficient application ... the nameserver would recognize that the delegated authority for sub.zone1.com is ... itself. It would complete the query of host.sub.zone1.com and return an answer instead of a referral. Am I on the right track with this or just wasting my time with wild and inaccurate speculation?

Yeah, that's not it.

I can't answer the original question, but the responses to queries in the scenario outlined are the same. BIND always returns the best answer it can. If it worked the way you were speculating (for 9.3), the BIND name server would not know that the second query was a follow up to the referral, and would process it in the same way. Responses in this scenario would be the same (at least as far as the answer section of the response is concerned) coming from BIND 9.9, 9.3, 9.1, 8.2, or 4.9. (I can't speak for 4.8.)

Chris Buxton
BlueCat Networks