Transport on two interfaces are "melting"

[cehes]

Hi,

we have 2 interfaces, one DATA and one VoIP, as usual.
We’d like to have phone/softphones use VoIP interface and SIP trunk use DATA

Then we configures two transports :

[transport-udp-phone]
bind=ip-of-voip-interface:5060

[transport-udp-trunk]
bind=ip-of-data-interface:5060

and when configuring a phone user of trunk we put the right transport

but when calling between two phones, it starts ok into sngrep, using ip of voip and then there is a SDP message telling “ok use data interface ip!” and then it goes wrong.
I read thinks about rtp-symmetric, we tried, but that wasn’t better

Any idea of why this SDP message tell to change interface ?

Thanks a lot

[jcolp]

You’d need to show the actual configuration in use including endpoints, as well as specify what version of Asterisk is in use.

[david551]

RTP will generally be sent on the interface that the OS says is the right interface for the destination IP address. The transports only sets the signalling interface.

[cehes]

Hi,
yes, this is what we thought and we did a routing per source IP to be sure that a packet comming for voip will go out via the same interface.
But the sdp packet seem to say “use now data ip” then all goes wrong.

[cehes]

Were are using asterisk 18.8.
We uses realtime then all is in database, except very few thinks like transport into pjsip.conf

cat /etc/asterisk/pjsip.conf

;---------Transports-----------
[transport-udp]
type=transport
protocol=udp
bind=10.1.200.2:5060

;[transport-trunk-udp]
;type=transport
;protocol=udp
;bind=192.168.240.41:5060

Our endpoints are quite normal. We are using softphone for tests that are in another network, but registering on voip interface of asterisk. We do not see any reason why SDP should suddenly say to change network.