[Spce-user] sipwise 3.1 (CE) and websocket support

Fri Feb 14 16:59:28 EST 2014

Hi,
thanks for detailed info, the problem is that kamailio-lb fails to
remove its own Route headers so the ACK loops from proxy back to lb.

The request and specifically the Route headers look OK to my eye:
> Feb 14 20:41:58 spce lb[3773]: NOTICE: <script>: New request - M=ACK R=sip:ngcp-lb at 172.18.101.48:5060;ngcpct=7369703a3132372e302e302e313a35303830 F=sip:4252143385 at 172.18.101.48 T=sip:2067077495 at 172.18.101.48 IP=ws:172.18.101.38:62770 ID=58961608-d303-29b9-f2b3-827dc65b7179
> Feb 14 20:41:58 spce lb[3773]: INFO: <script>: Unmasked internal contact - R=sip:127.0.0.1:5080 ID=58961608-d303-29b9-f2b3-827dc65b7179
> Feb 14 20:41:58 spce lb[3773]: INFO: <script>: Reset loose-routing, du='sip:172.18.101.48;transport=ws;r2=on;lr=on;ftag=qKNJeBx56Duy6f26lMTq;nat=yes;ngcplb=yes;socket=ws:172.18.101.48:5060' - R=sip:127.0.0.1:5080 ID=58961608-d303-29b9-f2b3-827dc65b7179
> 
> ACK sip:127.0.0.1:5080 SIP/2.0
> Record-Route: <sip:127.0.0.1;r2=on;lr=on;ftag=qKNJeBx56Duy6f26lMTq;nat=yes;ngcplb=yes;socket=ws:172.18.101.48:5060>
> Record-Route: <sip:172.18.101.48;transport=ws;r2=on;lr=on;ftag=qKNJeBx56Duy6f26lMTq;nat=yes;ngcplb=yes;socket=ws:172.18.101.48:5060>
> Via: SIP/2.0/UDP 127.0.0.1;branch=z9hG4bK7e9f.7ae637ea9391145a1b3a2569182092f3.0
> Via: SIP/2.0/WS df7jal23ls0d.invalid;received=172.18.101.38;branch=z9hG4bK0LbF7QFyhgoyOk2gQVK1;rport=62770
> From: \"4252143385\"<sip:4252143385 at 172.18.101.48>;tag=qKNJeBx56Duy6f26lMTq
> To: <sip:2067077495 at 172.18.101.48>;tag=6111C524-52FE71860005D18C-321A4700
> Contact: \"4252143385\"<sip:4252143385 at df7jal23ls0d.invalid;alias=172.18.101.38~62770~5;rtcweb-breaker=yes;click2call=no;transport=ws>;+g.oma.sip-im;+sip.ice;language=\"en,fr\"
> Call-ID: 58961608-d303-29b9-f2b3-827dc65b7179
> CSeq: 19663 ACK
> Content-Length: 0
> Max-Forwards: 15
> Proxy-Authorization: Digest username=\"4252143385\",realm=\"172.18.101.48\",nonce=\"Uv5yslL+cYbPfssd+yGMtK3o3OXkGDEm\",uri=\"sip:ngcp-lb at 172.18.101.48:5060;ngcpct=7369703a3132372e302e302e313a35303830\",response=\"087b7a6b4c55c1b8be8734072be1628c\",algorithm=MD5
> Route: <sip:172.18.101.48;transport=ws;r2=on;lr=on;ftag=qKNJeBx56Duy6f26lMTq;nat=yes;ngcplb=yes;socket=ws:172.18.101.48:5060>
> Route: <sip:127.0.0.1;r2=on;lr=on;ftag=qKNJeBx56Duy6f26lMTq;nat=yes;ngcplb=yes;socket=ws:172.18.101.48:5060>
> Route: <sip:127.0.0.1:5062;lr=on;ftag=qKNJeBx56Duy6f26lMTq;did=1ab.20c1;mpd=ii;ice_caller=strip;savp_caller=force_srtp;avpf_caller=force_avpf;ice_callee=strip;savp_callee=force_rtp;avpf_callee=force_avp;rtpprx=yes;vsf=ampNeU9nS3R6XXJyQ2QgJVJTaiZjfFJ4S35Wf0E->
> User-Agent: IM-client/OMA1.0 sipML5-v1.2014.01.27
> Organization: Doubango Telecom
> P-NGCP-Src-Ip: 172.18.101.38
> P-NGCP-Src-Port: 62770
> P-NGCP-Src-Proto: ws
> P-NGCP-Src-Af: 4
> P-Sock-Info: ws:172.18.101.48:5060
> P-NGCP-Src-Nat: 1

I guess we need to debug kamailio-lb here. Could you please raise the
loglevel to 3:
ngcp-kamctl lb fifo debug 3
then redirect log to a file: tail -f /var/log/ngcp/kamailio-lb.log | tee
kamailio-lb-debug.log
make a test call and send me the kamailio-lb-debug.log file?
And the kamailio-proxy.log with debug for this call would be also nice.
>From my side I'll setup a testbed and try to test it with sipml5 soon.

Regards,
Andrew

On 14/02/14 21:35, Ashutosh Apte wrote:
> Hi Andrew,
>     Thanks for your reply. I made the recommended changes and now I can
> see that the INVITE request received by the legacy gateway does contain
> a RTP/AVP media profile. I've run into the next problem though.
> 
> The legacy gateway does not receive an ACK so the call never gets
> established. (I do see that media from the WebRTC client does make it
> all the way to the legacy gateway so that's a good news.)
> 
> After about 32 seconds (while the gateway is re-transmitting its 200 OK
> response), ngcp sends an ACK (it is internally generated) followed by a
> BYE request to end the call.
> 
> I see the following errors in the kamailio-proxy.log:
> 
> -----
> Feb 14 20:41:58 spce proxy[3785]: DEBUG: <script>: start of route
> ROUTE_OUTBOUND - 1014 ACK
> Feb 14 20:41:58 spce proxy[3785]: WARNING: <core>
> [msg_translator.c:2499]: via_builder(): *_TCP/TLS connection (id: 0) for
> WebSocket could not be found_*
> Feb 14 20:41:58 spce proxy[3785]: ERROR: <core> [msg_translator.c:1725]:
> build_req_buf_from_sip_req(): could not create Via header
> Feb 14 20:41:58 spce proxy[3785]: ERROR: <core> [forward.c:607]:
> forward_request(): ERROR: forward_request: building failed
> Feb 14 20:41:58 spce proxy[3785]: ERROR: sl [sl_funcs.c:371]:
> sl_reply_error(): ERROR: sl_reply_error used: I'm terribly sorry, server
> error occurred (1/SL)
> Feb 14 20:41:58 spce proxy[3785]: DEBUG: <script>: exit of route
> ROUTE_OUTBOUND - 1014 ACK
> -----
> 
> Note that I'm using *ws://*172.18.101.48:5060/ws
> <http://172.18.101.48:5060/ws> (and not http://) as the websocket server
> URL. Andreas had suggested earlier to use *http:*//ip:port/ws. The
> WebRTC clients that I'm using for testing (sipML5 and jsSIP) do not like
> anything other than ws:// for Websocket server URL.
> 
> Not sure if this is the real issue as the request fails to get sent to
> the legacy gateway, which is not using Websockets but the above
> mentioned error occurs when it is ready to send the ACK to the legacy
> gateway.
> 
> Also attached is a .zip file containing the proxy and lb logs.
> 
> Thanks again for all your help. I think I'm getting closer to get
> everything working. 
> 
> Ashutosh
> 
> 
> 
> On Fri, Feb 14, 2014 at 1:38 AM, Andrew Pogrebennyk
> <apogrebennyk at sipwise.com <mailto:apogrebennyk at sipwise.com>> wrote:
> 
>     On 14/02/14 01:33, Ashutosh Apte wrote:
>     > I've also created a SIP Peering Group to route calls to a legacy SIP
>     > gateway. On dialing, the call hits the legacy gateway but I notice
>     that
>     > the SDP body in the INVITE message uses RTP/SAVF. This gateway
>     does not
>     > support sRTP. Is there a way to force the mediaproxy to use RTP only?
>     >
>     > I've changed the parameter under Domains -> Preferences -> NAT and
>     Media
>     > Flow Control -> srtp_transcoding from "transparent" to "ForceRTP". The
>     > kamailio proxy debug logs do show that the parameter was indeed
>     changed
>     > but it did not have any effect on the INVITE message.
>     >
>     > What configuration parameter should I be changing to get this to work?
> 
>     Hi,
>     from my PoV you should set srtp_transcoding in the domain to "Prefer
>     SRTP" (AFAIK it doesn't work if you leave it as "transparent" and I want
>     to add automatic detection of endpoints that use RTP/SAVP and RTP/SAVPF)
>     and most important - srtp_transcoding in the peer preferences (for your
>     legacy gateway) to "Force RTP". This will enable transcoding. You may
>     also need to set rtcp_feedback to "Prefer AVPF" on the domain and "Force
>     AVP" on the legacy gateway.
> 
>     Andrew
> 
> 