Why do I need a dot at the end of my lookup?

Barry Margolin barmar at alum.mit.edu
Fri Jul 22 23:52:35 UTC 2005


In article <dbqo3v$2fl0$1 at sf1.isc.org>,
 "Thomas Kinghorn [MTNNS -Rosebank]" <thomask at mtnns.net> wrote:

> =20
> 
> >From: Eivind Olsen [mailto:eivind at aminor.no]=20
> >
> >--On 22. juli 2005 11:05 +0200 "Thomas Kinghorn [MTNNS -Rosebank]"=20
> ><thomask at mtnns.net> wrote:
> >> Query was c:\nslookup www.cisco.com server.domain.name
> >>
> >> A TCP dump on server shows:
> >>
> >> Host.domain.name.1959 > server.domain.name.domain:  2+ A?
> >> www.cisco.com.int.domain.name. (45) [tos 0x28]
> >
> >I'm not familiar with this exact output, but... Host.domain.name is the
> machine where you ran c:\nslookup ? And=20

You're not familiar with tcpdump output?  What kind of networking expert 
are you?

> >server.domain.name is your DNS-server where you did this TCP-dump? If
> so, it looks like it's not the SERVER which > attaches "int.domain.name"
> to the requests, since that is already mentioned in the DNS-request from
> Host.domain.name ?
> 
> Correct.
> 
> However, when I query other nameservers, it does not do this.

Regardless, it's clear that the suffix is being added by the client, not 
the server.  I don't know why it only does this when querying that 
server, but it can't be the server's fault since it's the recipient, not 
the sender.

You need to take this up with the vendor of your client resolver library 
(looks like Microsoft, I think).

> The server I query is a caching only server.
> This server give a 2 second timeout to any query.
> 
> This is causeing headaches for our clients.
> 
> If you add the dot at the end of the query, the 2 second timeout
> disappears.

Putting a dot at the end of a name indicates that it's fully qualified, 
so the resolver library will never try adding a default suffix.

-- 
Barry Margolin, barmar at alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***



More information about the bind-users mailing list