Q: Delegated Servers vs NS records used in queries?

Brad Knowles brad.knowles at skynet.be
Fri Sep 7 08:27:05 UTC 2001


At 12:37 PM +1000 9/7/01, David Connors wrote:

>  Consider the following:
>  1. Suppose you have a domain xyz.com.
>  2. xyz.com is delegated to ns1.xyz.com and ns2.xyz.com.
>  3. the xyz.com zone file contains THREE NS records for ns1.xyz.com,
>  ns2.xyz.com, and an additional server, ns3.xyz.com.

	In this scenario, I would call ns3.xyz.com an "orphan" 
nameserver.  This is not quite as bad as a lame delegation, but it's 
still an issue with differences in delegation data, and generally 
speaking *any* differences at all in delegation data is bad.

>  When a client on the other side of the world attempts to query xyz.com, will
>  there ever be a scenario where ns3.xyz.com will be queried?  In otherwords,
>  is it part of the DNS standard that ONLY the delegated servers are queried?
>  Or is it that after the NS records are retrieved, a client DNS server might
>  consider any NS record as fair game for a subsequent query?

	If they had previously queried for information about xyz.com and 
found out about ns3.xyz.com, then if they ask any additional 
questions in the future, there is a chance that they may direct those 
queries to ns3.xyz.com instead of one of the other two machines. 
Indeed, if this did not happen, there would be no point whatsoever in 
listing any additional machines as nameservers for a zone.

	But there is no definitive method I know of to use in determining 
when the remote nameserver may query a particular local nameserver 
(or vice-versa), because everything depends on the various RTTs that 
they have measured, how "fast" they think a particular nameserver is 
answering (and therefore how preferred it may or may not be), whether 
they've decided to go ahead and ask this question of that nameserver 
anyway, so as to collect additional RTT data, etc....

-- 
Brad Knowles, <brad.knowles at skynet.be>

H4sICIFgXzsCA2RtYS1zaWcAPVHLbsMwDDvXX0H0kkvbfxiwVw8FCmzAzqqj1F4dy7CdBfn7
Kc6wmyGRFEnvvxiWQoCvqI7RSWTcfGXQNqCUAnfIU+AT8OZ/GCNjRVlH0bKpguJkxiITZqes
MxwpSucyDJzXxQEUe/ihgXqJXUXwD9ajB6NHonLmNrUSK9nacHQnH097szO74xFXqtlbT3il
wMsBz5cnfCR5cEmci0Rj9u/jqBbPeES1I4PeFBXPUIT1XDSOuutFXylzrQvGyboWstCoQZyP
dxX4dLx0eauFe1x9puhoi0Ao1omEJo+BZ6XLVNaVpWiKekxN0VK2VMpmAy+Bk7ZV4SO+p1L/
uErNRS/qH2iFU+iNOtbcmVt9N16lfF7tLv9FXNj8AiyNcOi1AQAA


More information about the bind-users mailing list