Again: capsforid let unbound fail to resolv a domain

Unbound
1

Hello,

after unbound 1.5.2 was announced I was happy to read:
- Fix 0x20 capsforid fallback to omit gratuitous NS and additional section changes.

Yea!
Unfortunately that fix still make my unbound not resolve some problematic domains.

Current case:

dig @ns1.ansmann.de 111.105.6.217.IN-ADDR.ARPA. ptr +noall +answer +nocomments

; <<>> DiG 9.9.4-P2 <<>> @ns1.ansmann.de 111.105.6.217.IN-ADDR.ARPA. ptr +noall +answer +nocomments
; (1 server found)
;; global options: +cmd
111.105.6.217.in-addr.arpa. 3600 IN PTR mx1.ansmann.de.

-> notice the lowercase "in-addr.arpa" in the answer section

other domains I have workarounds for:
  * 224.182.213.in-addr.arpa
  * 23.216.91.in-addr.arpa
  * 14.177.82.in-addr.arpa
  * cn4e.com
  * gc-tinnen.internetcube.net

The last one is not related to capsforid. The Nameserver ns.internetcube.net[85.214.85.64]
serve the label with a ttl 0 seconds and ns2.internetcube.net don't answer at all.
At the end unbound to fail to resolve gc-tinnen.internetcube.net.

Andreas

2

Hi Andreas,

Hello,

after unbound 1.5.2 was announced I was happy to read: - Fix 0x20
capsforid fallback to omit gratuitous NS and additional section
changes.

Yea! Unfortunately that fix still make my unbound not resolve some
problematic domains.

The code repository contains fixes for capsforid fallbacks that makes
the cases below resolve. (This code just missed 1.5.3).

One domain 14.177.82.in-addr.arpa I do not know if it works now
(please try).

Most of the fixes hinge around accepting a smaller number of fallback
(but equal) replies, because no more can be fetched. The last resort
fallback for nameserver lookup was interfering with the 0x20 fallback,
and also maximum query sent count was interfering with the 0x20 fallback.

Best regards,
   Wouter