NAT'ing of one's own WAN IP address

Greetings,

We are developing an application that uses UPnP to open ports on our
customer's routers, and in doing so, have run into a problem with a
few models, while others seem to behave fine. We are trying to figure
out what happens exactly, so we can find a work-around.

Our application opens port 82 through UPnP to service HTTP requests.
Assuming the computer's externally visible address is 1.1.1.1 :

A) From any internet connected computer: 1.1.1.1:82/ -> Works,
requests are serviced
B) From the computer running the client: localhost:82/ -> Works,
requests are serviced
C) From the computer running the client: 1.1.1.1:82 -> Works with some
routers, timeouts with others.

What is preventing case C from working correctly? We are assuming that
when the faulty routers receives the incoming LAN packets destined for
themselves, the port forwarding is not applied properly, and the
packet is dropped. But this is at best a theory, and we'd be happy if
anyone could provide some insight into the causes of the problem.

For the records, the routers that gave us this problem are the
Gigafast EE 410-R and the NetGear WGR614v6.

Thanks in advance for any help,

For anyone stumbling on this post, here is what I found out:

This has to do with the router's support of Bi-directional NAT, and
NAT loopback. The routers who do not support this will always fail at
NAT'ing inbound traffic back correctly.

On Apr 19, 12:46 pm, Eric Fournier <ericfourni...@gmail.com> wrote:
> Greetings,
>
> We are developing an application that uses UPnP to open ports on our
> customer's routers, and in doing so, have run into a problem with a
> few models, while others seem to behave fine. We are trying to figure
> out what happens exactly, so we can find a work-around.
>
> Our application opens port 82 through UPnP to service HTTP requests.
> Assuming the computer's externally visible address is 1.1.1.1 :
>
> A) From any internet connected computer: 1.1.1.1:82/ -> Works,
> requests are serviced
> B) From the computer running the client: localhost:82/ -> Works,
> requests are serviced
> C) From the computer running the client: 1.1.1.1:82 -> Works with some
> routers, timeouts with others.
>
> What is preventing case C from working correctly? We are assuming that
> when the faulty routers receives the incoming LAN packets destined for
> themselves, the port forwarding is not applied properly, and the
> packet is dropped. But this is at best a theory, and we'd be happy if
> anyone could provide some insight into the causes of the problem.
>
> For the records, the routers that gave us this problem are the
> Gigafast EE 410-R and the NetGear WGR614v6.
>
> Thanks in advance for any help,