[Asterisk-Dev] Re: [Asterisk-Users] Asterisk does not handle call
from a Cisco IAD correctly
Tilghman Lesher
tilghman at mail.jeffandtilghman.com
Tue Dec 27 14:48:58 MST 2005
On Tuesday 27 December 2005 15:21, Steven Critchfield wrote:
> On Tue, 2005-12-27 at 15:07 -0600, Tilghman Lesher wrote:
> > On Tuesday 27 December 2005 14:15, James Sizemore wrote:
> > > I think I found what is munging up the peer lookup:
> > >
> > > This call from another Asterisk box starts:
> > >
> > > <-- SIP read from 192.168.69.254:5060:
> > >
> > > The peer lookup that fail reads:
> > >
> > > <-- SIP read from 192.168.7.250:52141:
> > >
> > > Asterisk seem to be thrown off by the fact that the return port
> > > is not 5060, and fails the peer lookup. This is a * bug then. I
> > > have documented it with both 1.0.9 and 1.2.1. Time to dig through
> > > the sip code.
> >
> > No, this is actually sane. This is necessary behavior in order to
> > support multiple SIP clients from behind the same NAT. Asterisk
> > needs to know the control port for your host, and it needs to stay
> > consistent between a REGISTER and an INVITE. If Asterisk sees a
> > different control port, it quite naturally assumes that that's a
> > different client.
>
> Your explination doesn't seem to satisfy James' observation. He is
> saying that on one host with the 5060, port the lookup succeeds, but
> one a DIFFERENT host, with a non 5060 port, it fails. While yes they
> both are on a private network address space, it doesn't seem that NAT
> has anything to do with this here, nor the client staying on the same
> port.
What I'm saying, though, is that the registration from that machine was
likely to be on port 5060, while it's attempting a call from a DIFFERENT
control port.
If he's not doing registrations of his Cisco machine to Asterisk, then
needs to specify the port number that machine is using for the control
port in sip.conf:
[machine]
type=user
port=65437
host=192.168.4.5
A registration is only good for an IP/control port combination, not for
an IP alone.
--
Tilghman
More information about the asterisk-dev
mailing list