small sibling glue records help

Alexander Ottl alexander.ottl at nci.de
Wed Dec 19 16:42:20 UTC 2007


Alan Clegg wrote, On 12/19/2007 03:20 PM:
> Anatoly Pugachev wrote:
>> Can someone explain me, what does mean:
>> 18-Dec-2007 13:34:10.856 zone dom1.ru/IN: avgust.dom1.ru/NS 'ns.sub1.dom1.ru' has no SIBLING GLUE address records (A or AAAA)
>>
>> Does that mean, that i should add A record for ns.sub1.dom1.ru into dom1.ru zone file?
>> I already have 
>>
>> $ORIGIN dom1.ru
>> sub1    NS ns1.zone.ru.
>>         NS ns2.zone.ru.
> 
> What this message means is that the nameservers (pointed to by the NS
> records) reside within the zone for which they are authoritative.  For
> the nameserver of the zone above (in this case dom1.ru) to be able to
> provide a response including the IP addresses of the nameservers for the
> subordinate zone, you need the IP addresses in the parent.
> 
> In other words, yes, you should add A records for the nameservers that
> live within the zone in question.  I'm thinking that your example is
> probably:
> 
>  $ORIGIN sub1.dom1.ru
>  sub1    NS ns1.sub1.dom1.ru.
>          NS ns2.sub1.dom1.ru.
> 
> In this case, you need to add:
> 
>  ns1.sub1.dom1.ru. IN A x.y.z.a
>  ns2.sub1.dom1.ru. IN A l.m.n.o
> 
> to the dom1.ru zone file.  (that's the "glue")
> 
> AlanC

Forgive me for correcting you, Alan, but that's not entirely true.
Actually - according to the error message above - there are *TWO* zone
cuts/delegations below dom1.ru (avgust.dom1.ru and sub1.dom1.ru) and
one's name servers lie below the other. That's why the error message
reads SIBLING GLUE instead of REQUIRED GLUE. In any case, the A record
for ns.sub1.dom1.ru is the glue that's missing.

If I understand correctly, not handing out this glue record invites a
resolving name server querying for any names below avgust.dom1.ru to
abandon its original query and start hunting for ns.sub1.dom1.ru, then
probably abandoning that query to hunt for the name servers of
sub1.dom1.ru (ns1.zone.ru,ns2.zone.ru). This would work with a BIND 9
resolver but used to be a problem with BIND 8.

Regards,
Alex



More information about the bind-users mailing list