Conflicting glue records?

Milo Hyson milo at cyberlifelabs.com
Thu Jan 8 13:21:04 UTC 2009


We have a working hypothesis. It appears that glue-records are set  
only when one configures a self-referential domain (e.g. example.com -- 
 > ns.example.com), and they're only deleted if one explicitly asks  
the registrar to do so.

In our particular case, it seems when the domain in question was moved  
to the new ISP, not only did the name-server IPs get changed at the  
registrar but so did the names. Previously they were self-referential,  
but now they are pointing to servers in another domain (i.e.  
example.com --> ns.myisp.com). I'll bet anything that the person who  
made the change didn't ask the registrar to delete the old glue-records.

--
Milo Hyson
Chief Scientist
CyberLife Labs


On Jan 8, 2009, at 03:33, Milo Hyson wrote:

> I would think that any server-name should be permissible for any  
> domain. However, the IP should only be used in the case of a self- 
> referential entry (your first case).
>
> Is there any legitimate reason for someone to give an IP for a name- 
> server that differs from the authoritative source? That is, if the  
> authoritative name-server for myisp.com resolves ns1.myisp.com to  
> 1.2.3.4, is there any valid reason for someone to register  
> example.com with a name-server of ns1.myisp.com and an IP of 4.5.6.7?
>
> --
> Milo Hyson
> Chief Scientist
> CyberLife Labs
>
>
> On Jan 8, 2009, at 03:15, Matus UHLAR - fantomas wrote:
>
>> If I register example.com, it's OK to register  
>> _anything_.example.com as
>> glue. However, registering _anything_.example.net should not be  
>> accepted.
>>
>> I wonder how is it possible that anyone accepts that.
> _______________________________________________
> bind-users mailing list
> bind-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/bind-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/bind-users/attachments/20090108/2ed3cfd5/attachment.html>


More information about the bind-users mailing list