Weird DNS behaviour resolution issues when more labels are present in a zone

[Mark Andrews]

> On 12 Dec 2020, at 20:36, Prasanna Mathivanan (pmathiva) via bind-users <bind-users at lists.isc.org> wrote:
> 
> Hi everyone,
>  
> I have an issue in resolving a domain, from logs I see its timing out.
> And from dig output we are getting SERV fail response.
> The bind version we are using 9.14.10, same domain resolves in bind version 9.11 and lower.
>  
> Example domain:-  a.b.c.eample.com
> When we took tcpdump and saw what’s happening when we do a dig, we see its querying the wrong domain“_.b.c.example.com” , and it’s not able to query the NS for this domain and we get timeout in logs.
> Adding to that we get SERVFAIL response when doing dig.

If you are getting a timeout then the server for _.b.c.example.com is broken or it is unreachable.
“_” is a legal label in the DNS.  If the server for b.c.example.com responds to b.c.example.com but
does not respond to _.b.c.example.com then the server is broken or there is a broken firewall in
front of it.

> We don’t see this behaviour for bind version 9.11 or lower and works with +trace as well.
>  
> If anyone can explain why this behaviour is happening, it will be very helpful in understanding the issue.
>  
> After looking into the problem, it appears that bind 9.14 ships with Query Name Minimisation feature as defined by RFC 7816 enabled by default.
> few have experienced this behaviour and solution was to disable QNAME minimization.
>  
> How does QNAME Minimisation alter this behaviour? To quote from RFC 7816:
> Instead of sending the full QNAME and the original QTYPE upstream, a resolver that implements QNAME minimisation and does not already have the answer in its cache sends a request to the name server authoritative for the closest known ancestor of the original QNAME. The request is done with:
> 	• the QTYPE NS

> 	• the QNAME that is the original QNAME, stripped to just one label more than the zone for which the server is authoritative
> A resolver using QNAME Minimisation implicitly assumes that each label in the query name corresponds to a zone cut. The resolver queries a parent zone server, using an abbreviated query name that is truncated after the name of the immediate child label and uses a query type of NS.
>  
>  
> Am adding the following links to justify this behaviour, but just wanted a suggestion if we are good with doing this.
> https://datatracker.ietf.org/doc/rfc7816/?include_text=1
> https://blog.apnic.net/2019/08/12/dns-query-privacy/
> https://labs.ripe.net/Members/wouter_de_vries/make-dns-a-bit-more-private-with-qname-minimisation
> https://github.com/iagox86/dnscat2/issues/144
>  
> Disabling QMIN does fix the issue, but I would like to understand why delegation breaks if there are more labels.
> And why the query goes to underscore domain even though it doesn’t exist.

Because people deploy non-RFC compliant nameservers.  If you find one complain to the operator of it.

_.<label-sequence> is chosen so that the resulting NXDOMAINs are not being cached for <label-sequence>.
The initial QMIN implementation used NS queries but that caused problems when servers returned NXDOMAIN
to NS queries but there were really records there.  Querying for A / AAAA at <label-sequence> also distorts
QTYPE statistics for <label-sequence>.  Prepending “_” prevents that by using a QNAME that almost never
exists.

> -- 
> Thanks
> Prasanna
> _______________________________________________
> Please visit https://lists.isc.org/mailman/listinfo/bind-users to unsubscribe from this list
> 
> ISC funds the development of this software with paid support subscriptions. Contact us at https://www.isc.org/contact/ for more information.
> 
> 
> bind-users mailing list
> bind-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/bind-users

-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742              INTERNET: marka at isc.org