"Doug Crabtree" <(E-Mail Removed)> wrote in message
news:eZ6dnbkHwuZb8DnfRVn-(E-Mail Removed)...
> No, that is not what I meant. Sorry.
>
> I am not asking from the client side, I am asking from the server side.
>
> 1) The user has in his/her VPN settings to connect to ip xxx.xxx.xxx.xxx
> (this static side of the server not defined with a default gateway)
> 2) The user authenticates and establishes the connection.
> 3) I go into RRAS and check out the IP routing table (or route print).
> That connection is shown to have a route out via the dynamic IP
> yyy.yyy.yyy.yyy (defined with a default gateway) on the server.
>
> My question was, why is W2K3 establishing the VPN connection on the NIC
> (yyy.yyy.yyy.yyy) that was not originally tried by the client
> (xxx.xxx.xxx.xxx)?
I'm not sure of all the gory details, but you can really have only one
"true" external Nic and it will be the one with the Default Gateway. Any
outbound traffic will alway use that Nic unless the destination is a
specific "known" route that is already established to be associated with one
of the other Nics. Since the "caller" is not part of a pre-established
route which would be associated with a different Nic,..they become
associated with the Nic that is tied to their "unknown" route which is the
one with the Default Gateway. This would be the case whether you were using
VPN or not.
As an experiment,..if the "caller" comes from only a certain number of
addresses that are part of a single subnet (their ISP's subnet) then create
a Static Route on the Server that tells it to use the xxx.xxx.xxx.xxx Nic to
get to that subnet destination. Now it should show the Nic you were
expecting instead of the one with the Default Gateway because now the
xxx.xxx.xxx.xxx Nic is associated with that particular destination subnet.
--
Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com