[Asterisk-Users] 2 Sip phones behind un-natted Asterisk

[Bruce Komito]

Not to beat a dead horse, but I had the problem even with the two lines on
different ports.  The canreinvite=no thing solved the problem.

Bruce Komito
High Sierra Networks, Inc.
www.servers-r-us.com
(775) 284-5800 ext 115


On Mon, 24 May 2004, John Fraizer wrote:

> Bruce Komito wrote:
>
>  > In sip.conf, try setting canreinvite=no for both lines.
>  >
>  > Bruce Komito
>  > High Sierra Networks, Inc.
>  > www.servers-r-us.com
>  > (775) 284-5800 ext 115
>
> canreinvite=no will sometimes make a difference but, I believe that what
> most people are running into is what I described in my previous post.
> NAT devices track IP flows.  When you have two "flows" from the asterisk
> server that look like:
>
> ast.erisk.serv.er:5060<->nat.wan.ip.addr:5060<->NAT<->10.1.1.1:5060
> ast.erisk.serv.er:5060<->nat.wan.ip.addr:5060<->NAT<->10.1.1.2:5060
>
> The NAT device has no way of deciding which of the two or more SIP
> devices (hardphone or softphone) on the inside the flow is supposed to
> go to.  It looks through it's flow database and sees that there are two
> potential candidates on the inside that could be the owner of the flow.
>
> When you make sure that your SIP devices behind the NAT use unique
> source ports, you end up with this:
>
> ast.erisk.serv.er:5060<->nat.wan.ip.addr:5061<->NAT<->10.1.1.1:5061
> ast.erisk.serv.er:5060<->nat.wan.ip.addr:5062<->NAT<->10.1.1.2:5062
>
> I guess that many people don't really truly understand how the NAT works
> on the raw level and it's a plug and play device to them.  For that
> matter, I have been running large scale networks for nearly 16 hears and
> I got bit by the same thing because I assumed that the Cisco would do
> the "Right_Thing" (TM) and choose a random source port.  Once I
> discovered that it didn't do that, it was a simple matter of telling it
> to do so.
>
> I suppose that this is also somewhat the fault of the NAT device
> (linksys/cisco) as well since it should have noticed that there were two
> devices on the inside of the NAT trying to create identical flows.  It
> could have easilly done some mapping to change the source port on the
> WAN side of the connection of one of the flows.
>
> OK.  This is the * list and not routing 701 so, I'll stop now.  Suffice
> it to say that it is a good idea to have unique source ports if you have
> multiple devices on the inside contacting the same device:port on the
> outside.
>
> John Fraizer
> EnterZone, Inc
> http://www.enterzone.net/
>
>
>