servfail response message question

[Barry Margolin]

In article <mailman.668.1372206152.20661.bind-users at lists.isc.org>,
 RYAN CHERVENKA <ryan_chervenka at yahoo.com> wrote:

> I currently have a domain example.com authoritative on my Ubuntu server and 
> it is delegating gslb.example.com to my load balancer.Ý

www.example.com is a 
> CNAME for www.gslb.example.comÝ
Gslb.example.com has an NS record pointing to 
> the LB

Client sends query for www.example.com to Ubuntu DNS server. The 
> Ubuntu DNS server sends a query to the load balancer for www.gslb.example.com 
> and the LB responds to the Ubuntu DNS server with the right A record in the 
> answer section. However, the Ubuntu server responds to the client with 
> servfail.Ý

When I look at the pcap from the Ubuntu server, the LB is 
> responding to it with the correct IP but the dig response from the Ubuntu 
> server to the client shows "no servers could be reached" when I dig against 
> the Ubuntu. I also see the same message in the dns response in the pcap 
> (obviously).

Ryans-MacBook-Pro:~ ryanc$ dig @10.10.1.50 www.example.com <-me 
> querying the Ubuntu for www.example.com

; <<>> DiG 9.8.3-P1 <<>> @10.10.1.50 
> www.example.com
; (1 server found)
;; global options: +cmd
;; connection 
> timed out; no servers could be reached


Do you have any ideas as to why this 
> is happening?

Ryan Chervenka

Why is the Ubuntu server sending the query to the LB at all? To emulate 
how a caching server works, you should be sending queries WITHOUT the RD 
flag, so the Ubuntu server should return a delegation to the LB, and 
then the caching server should query the LB.

Regarding the problem as you state it, is the LB responding 
authoritatively?

-- 
Barry Margolin
Arlington, MA