One curiosity is that the reinvite does not appear to have the Via from the UAC that sent the initial invite. But certainly that would not cause this behaviour?
On Tue, Aug 14, 2018 at 12:52:50PM -0400, Alex Balashov wrote:
Another aspect of this mystery:
Here is the route set in the reinvite:
Route: sip:1.1.1.1:5061;transport=tls;r2=on;lr=on;ftag=gK00253981;vsf=AAAAACERAR0RHQYQJVJ3GAUdAx0EAAQfATEw;dlgcor=4411.f183;proxy_media=yes Route: sip:1.1.1.1;r2=on;lr=on;ftag=gK00253981;vsf=AAAAACERAR0RHQYQJVJ3GAUdAx0EAAQfATEw;dlgcor=4411.f183;proxy_media=yes Route: sip:Restricted@3.3.3.3:5060;lr
The expectation would be that Kamailio would strip off its two Routes and then relay this request to 3.3.3.3:5060, even if the RURI says to relay the request to itself.
But that's not what actually happens. Kamailio clearly forwards the request to itself, as per the RURI, because the next log message we see is:
[R-MAIN:...] Re-INVITE received from 1.1.1.1:5060 to RURI sip:Restricted@3.3.3.3:5060;lr
Um, what?
Here is the full logged reinvite, for reference:
INVITE sip:1.1.1.1:5061 SIP/2.0 CSeq: 2 INVITE To: sip:stricted@3.3.3.3;tag=gK00253981 From: sip:+17023880118@1.1.1.1;tag=8812465_1533349860 Call-ID: 992000768_14572846@3.3.3.3 Via: SIP/2.0/TLS 4.4.4.4:5061;rport;branch=z9hG4bK845565_1533349860 Route: sip:1.1.1.1:5061;transport=tls;r2=on;lr=on;ftag=gK00253981;vsf=AAAAACERAR0RHQYQJVJ3GAUdAx0EAAQfATEw;dlgcor=4411.f183;proxy_media=yes Route: sip:1.1.1.1;r2=on;lr=on;ftag=gK00253981;vsf=AAAAACERAR0RHQYQJVJ3GAUdAx0EAAQfATEw;dlgcor=4411.f183;proxy_media=yes Route: sip:Restricted@3.3.3.3:5060;lr User-Agent: ClownCar WhoKnows Max-Forwards: 32 ASupportedCodec: 0 8 Contact: sip:MyTrunk@4.4.4.4:5061;transport=tls Supported: em,timer,replaces,path,resource-priority Allow: REGISTER,OPTIONS,INVITE,ACK,CANCEL,BYE,NOTIFY,PRACK,REFER,INFO,SUBSCRIBE,UPDATE Content-Type: application/sdp Content-Length: 303
v=0 o=ClownCar 1534262664 1534262665 IN IP4 4.4.4.4 s=SIP Call c=IN IP4 4.4.4.4 t=0 0 m=image 50104 udptl t38 a=T38FaxVersion:0 a=T38FaxMaxBuffer:1024 a=T38FaxMaxDatagram:238 a=T38FaxRateManagement:transferredTCF a=T38FaxUdpEC:t38UDPRedundancy a=T38MaxBitRate:14400 a=sendrecv
-- Alex
On Tue, Aug 14, 2018 at 12:44:21PM -0400, Alex Balashov wrote:
Hi,
We have a scenario like this:
ITSP -----> Kamailio -----> Endpoint (UDP) (TLS)
So, TLS is only being used on the last hop, and the upstream interaction with the ITSP is plain-old UDP.
Kamailio has the following listeners:
listen=udp:1.1.1.1:5060 listen=udp:1.1.1.2:5060 listen=tcp:10.0.0.1:5060 listen=tls:1.1.1.1:5061
At some point, 'Endpoint' sends a reinvite which has the following RURI:
INVITE sip:1.1.1.1:5061 SIP/2.0
This is clearly improper, and caused a loop that led to the rtpengine SDP loop issue I previously reported in another thread.
So, in an effort to stop this, I added the following:
if(has_totag()) { if(loose_route()) { ...
if(is_method("INVITE")) { xlog("L_INFO", "[R-MAIN:$ci] Re-INVITE received from $si:$sp to RURI $ru\n"); xlog("L_INFO", "[R-MAIN:$ci] Reinvite body: $mb\n"); if(!is_method("ACK") && uri == myself) { sl_send_reply("400", "Bad Request"); exit; } }}
But it doesn't work. It appears that the '400 Bad Request' rejection never happens, presumably because the this domain doesn't match 'myself'.
Another perplexing mystery: the log message containing the reinvite's '$ru' does not show a RURI of 'sip:1.1.1.1:5061', but rather the remote target in the initial inbound INVITE, which we also logged:
Contact: "Anonymous" sip:Restricted@3.3.3.3:5060
The log message says:
[R-MAIN:...] Re-INVITE received from 4.4.4.4:5060 to RURI sip:Restricted@3.3.3.3:5060;lr.
Note a subtle detail here: the ';lr' parameter is present, which is an attribute of the Record-Route inserted by the sending ITSP (3.3.3.3). It's at the bottom of the Route set, of course, below Kamailio's two RRs (inserted for the ingress UDP interface and the egress TLS interface):
Route: sip:Restricted@3.3.3.3:5060;lr
This leads to two questions whose causes seem to be related:
- Why does Kamailio think the request URI of this re-invite is
something other than what $mb reveals it to be?
Is that, presumably, why it does not match 'myself'?
Why would Kamailio think it is actually set to the far-end
Record-Route URI?
This is version: kamailio 4.4.5 (x86_64/linux) d48094.
Thanks,
-- Alex
-- Alex Balashov | Principal | Evariste Systems LLC
Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
-- Alex Balashov | Principal | Evariste Systems LLC
Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users