[bind10-dev] some overflow? (NOERROR versus NXDOMAIN)

[Jeremy C. Reed]

Sometimes the server returns NXDOMAIN when doing a lookup for a record 
that doesn't exist but the label does exist so should return NOERROR. 

The label before the record I lookup appears to cause the problem.

I had a huge zone file that I slowly removed from until I could get a 
small test case.

-=-=-=-=-=-=-
Zone #1:

noerror.testzone.	600	IN SOA	dns foo 1 2 3 4 5
			86400	NS	dns
111111111111111111111111111111111111.noerror.testzone. 86400 IN CNAME whatever
22222222222222222222222.noerror.testzone. 86400 IN AAAA 2001:470:1F00:FFFF::1

When I do this lookup:

; <<>> DiG 9.4.1-P1 <<>> @127.0.0.1 -p 5300 
22222222222222222222222.noerror.testzone A
; (1 server found)
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 28570
;; flags: qr aa rd; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 0
(snipped)

You see I get NOERROR with 0 answers. This is what I expect.

-=-=-=-=-=-=
Zone #2 has one character longer in the CNAME label above:

noerror.testzone.	600	IN SOA	dns foo 1 2 3 4 5
			86400	NS	dns
1111111111111111111111111111111111111.noerror.testzone. 86400 IN CNAME whatever
22222222222222222222222.noerror.testzone. 86400 IN AAAA 2001:470:1F00:FFFF::1

But when I do the exact same query:

; <<>> DiG 9.4.1-P1 <<>> @127.0.0.1 -p 5300 
22222222222222222222222.noerror.testzone A
; (1 server found)
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 49284
;; flags: qr aa rd; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 0

I get NXDOMAIN instead of NOERROR.

I need another set of eyes because I have tried over 30 zone files 
tracking this down.

(I didn't see anything interesting when looking at the records in 
sqlite3.)