[Freeswitch-users] WebRTC connection delay (Auto Changing stun/rtp/dtls port from)

[Adam Ben-Ayoun]

Just to clarify:
54.187.44.182 <http://54.187.44.182:53379/> is our TURN server (I actually
forgot to replace it with X.X.X.X)
67.180.21.132 is the WiFi interface
166.170.42.141 is the 3G interface

On 22 February 2015 at 23:20, Adam Ben-Ayoun <adam.ben.ayoun1 at gmail.com>
wrote:

> Hi guys,
>
> We are using Freeswitch as an MCU for WebRTC (only for audio conferences),
> most of the times connecting from our WebRTC client is pretty quick,
> however, on some WiFi's (probably with a certain NAT/topology), the audio
> connection can take up to 10sec to initialize when using STUN and TURN, we
> also see there are differences on the Freeswitch logs when it happens:
>
> 2015-02-22 21:06:25.920612 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 166.170.42.141:40334
> 2015-02-22 21:06:25.960618 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:26.080616 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 54.187.44.182:53379
> 2015-02-22 21:06:26.680617 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 54.187.44.182:53379 to 166.170.42.141:40334
> 2015-02-22 21:06:27.040614 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:27.880612 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 166.170.42.141:40334
> 2015-02-22 21:06:28.020610 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:28.600614 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 166.170.42.141:40334
> 2015-02-22 21:06:28.980603 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:29.560612 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 166.170.42.141:40334
> 2015-02-22 21:06:29.940602 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:30.540561 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 166.170.42.141:40334
> 2015-02-22 21:06:30.940606 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:31.500595 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 166.170.42.141:40334
> 2015-02-22 21:06:31.900613 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:32.460602 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 67.180.21.132:59854 to 166.170.42.141:40334
> 2015-02-22 21:06:32.860614 [NOTICE] switch_rtp.c:1151 Auto Changing
> stun/rtp/dtls port from 166.170.42.141:40334 to 67.180.21.132:59854
> 2015-02-22 21:06:33.040612 [INFO] switch_rtp.c:2926 Changing audio DTLS
> state from HANDSHAKE to SETUP
> 2015-02-22 21:06:33.040612 [INFO] switch_rtp.c:2834 audio Fingerprint
> Verified.
> 2015-02-22 21:06:33.040612 [INFO] switch_rtp.c:3376 Activating Audio
> Secure RTP SEND
> 2015-02-22 21:06:33.040612 [INFO] switch_rtp.c:3354 Activating Audio
> Secure RTP RECV
> 2015-02-22 21:06:33.040612 [DEBUG] switch_core_sqldb.c:2599 Secure Type:
> srtp:dtls:AES_CM_128_HMAC_SHA1_80
> 2015-02-22 21:06:33.040612 [INFO] switch_rtp.c:2874 Changing audio DTLS
> state from SETUP to READY
> 2015-02-22 21:06:33.040612 [DEBUG] switch_core_sqldb.c:2599 Secure Type:
> srtp:dtls:AES_CM_128_HMAC_SHA1_80
> 2015-02-22 21:06:33.100606 [DEBUG] switch_rtp.c:1954 rtcp_stats_init:
> ssrc[-1706748029] base_seq[28893]
>
>
> X.X.X.X is the IP of our TURN server, now, it seems like although we have
> a good reflexive candidate, it takes some time for Freeswitch to pick it up
> and it actually tries to use the relay (TURN)... Any idea what is happening
> and what we can do to solve that?
>
> Here are the ICE candidates from the client SDP:
> a=candidate:1780409101 1 udp 2122063615 192.168.15.3 56494 typ host
> generation 0
> a=candidate:1780409101 2 udp 2122063615 192.168.15.3 56494 typ host
> generation 0
> a=candidate:4043381328 1 udp 2122129151 10.159.103.156 51485 typ host
> generation 0
> a=candidate:4043381328 2 udp 2122129151 10.159.103.156 51485 typ host
> generation 0
> a=candidate:2618202841 1 udp 1685855999 67.180.21.132 56494 typ srflx
> raddr 192.168.15.3 rport 56494 generation 0
> a=candidate:2618202841 2 udp 1685855999 67.180.21.132 56494 typ srflx
> raddr 192.168.15.3 rport 56494 generation 0
> a=candidate:614277117 1 tcp 1518083839 192.168.15.3 54312 typ host tcptype
> passive generation 0
> a=candidate:614277117 2 tcp 1518083839 192.168.15.3 54312 typ host tcptype
> passive generation 0
> a=candidate:3213026976 1 tcp 1518149375 10.159.103.156 54313 typ host
> tcptype passive generation 0
> a=candidate:3213026976 2 tcp 1518149375 10.159.103.156 54313 typ host
> tcptype passive generation 0
> a=candidate:3054806047 1 udp 24911615 X.X.X.X 65452 typ relay raddr
> 67.180.21.132 rport 54310 generation 0
> a=candidate:3054806047 2 udp 24911615 X.X.X.X 65452 typ relay raddr
> 67.180.21.132 rport 54310 generation 0
> a=candidate:1032594883 1 udp 1685921535 166.170.42.141 64296 typ srflx
> raddr 10.159.103.156 rport 51485 generation 0
> a=candidate:1032594883 2 udp 1685921535 166.170.42.141 64296 typ srflx
> raddr 10.159.103.156 rport 51485 generation 0
> a=candidate:3054806047 1 udp 24977151 X.X.X.X 49737 typ relay raddr
> 166.170.42.141 rport 56009 generation 0
> a=candidate:3054806047 2 udp 24977151 X.X.X.X 49737 typ relay raddr
> 166.170.42.141 rport 56009 generation 0
>
> Thanks,
> Adam
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20150222/898ca1dc/attachment.html