[OpenSIPS-Users] UDP to TCP issue with Re-INVITE

Mon Dec 16 09:51:26 EST 2019

please increase the debug level and paste the log. Also, check what is
saved as the location for the user.

On Mon, 16 Dec 2019 at 14:48, Mickael Hubert <mickael at winlux.fr> wrote:

> Maybe I founded:
> I tested a call through Internet connection and I can see (uac natted):
>
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:proto_tcp_send: *no open tcp connection found*, opening new one,
> async = 1
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:probe_max_sock_buff: getsockopt: snd is initially 16384
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> INFO:core:probe_max_sock_buff: using snd buffer of 416 kb
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> INFO:core:init_sock_keepalive: TCP keepalive enabled on socket 11
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:tcpconn_async_connect: Polling is overdue
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:tcpconn_async_connect: Create connection for async connect
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:print_ip: tcpconn_new: new tcp connection to: 192.168.10.203
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:tcpconn_new: on port 5060, proto 2
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:proto_tcp_send: Successfully connected from interface
> 192.168.10.203:5060 to 192.168.10.203:56899!
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:proto_tcp_send: Successfully started async connection
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:tm:insert_timer_unsafe: [0]: 0x7efe91196790 (99)
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:tm:t_relay_to: new transaction fwd'ed
> Dec 16 15:15:44 am-frontal1a-test /usr/local/sbin/opensips[31165]: retcode
> = 1
>
> The re-invite is never forwarded (no trace with sngrep, or classic
> tcpdump)
> I don't know why opensips wants use private IP and not natted IP... So...
>
> I tested through VPN connection (uac natted):
>
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:tcp_conn_get: *tcp connection found* (0x7efe9118cf48), acquiring
> fd
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:tcp_conn_get: c= 0x7efe9118cf48, n=16, Usock=168
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31187]:
> DBG:core:handle_worker: read response= 7efe9118cf48, 1, fd -1 from 35
> (31165)
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:tcp_conn_get: after receive_fd: c= 0x7efe9118cf48 n=8 fd=11
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:proto_tcp_send: sending via fd 11...
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:async_tsend_stream: Async successful write from first try on
> 0x7efe9118cf48
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:core:proto_tcp_send: after write: c= 0x7efe9118cf48 n/len=1014/1014
> fd=11
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:tm:insert_timer_unsafe: [0]: 0x7efe91196790 (1403)
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]:
> DBG:tm:t_relay_to: new transaction fwd'ed
> Dec 16 15:37:28 am-frontal1a-test /usr/local/sbin/opensips[31165]: retcode
> = 1
>
> The re-inivte is forwarded correctly !
>
> I pretty sure I use the keepalived mecasim ....
>
>
> Le lun. 16 déc. 2019 à 14:27, Mickael Hubert <mickael at winlux.fr> a écrit :
>
>> Hi David,
>> Yes I use it
>>
>> if (nat_uac_test("3") && ($Ri == $var(publicip) || $Ri == $var(vpnip)))
>> {
>> xlog("L_INFO","$avp(startlog) -- Nated EP Detected\n");
>> if (force_rport())
>> {
>> xlog("L_INFO","$avp(startlog) -- RPORT parameter forced\n");
>> }
>> if (fix_nated_contact())
>> {
>> xlog("L_INFO","$avp(startlog) -- Nated $rm's Contact Fixed !\n");
>> }
>> if (fix_nated_sdp("10"))
>> {
>> xlog("L_INFO","$avp(startlog) -- Nated SDP Fixed for $rm\n");
>> }
>> }
>>
>> Le lun. 16 déc. 2019 à 13:50, David Villasmil <
>> david.villasmil.work at gmail.com> a écrit :
>>
>>> Aré you using nathelper?
>>>
>>> On Mon, 16 Dec 2019 at 12:06, Mickael Hubert <mickael at winlux.fr> wrote:
>>>
>>>> Hi all,
>>>> I advanced in my LAB
>>>> I have this configuration:
>>>> SIP client (tcp:192.168.10.203) -- my NAT router --> (tcp:2.2.2.2:8060)
>>>> opensips (udp:10.1.15.126:5060) --> rest of infra (udp:
>>>> 10.1.15.0/24:5060)
>>>>
>>>> When I received the reinvite from "rest of infra" on private interface
>>>> (10.1.15.126), I could see this log:
>>>>
>>>> Dec 16 12:55:27 am-frontal1a-test /usr/local/sbin/opensips[26160]:
>>>> ERROR:tm:msg_send: send() to 192.168.10.203:5060 for proto tcp/2 failed
>>>> Dec 16 12:55:27 am-frontal1a-test /usr/local/sbin/opensips[26160]:
>>>> ERROR:tm:t_forward_nonack: sending request failed
>>>> Dec 16 12:55:27 am-frontal1a-test /usr/local/sbin/opensips[26160]:
>>>> retcode = -6
>>>>
>>>> I don't know why OpenSIPS tries to send the Re-invite to client private
>>>> IP instead client public port and IP (natted).
>>>>
>>>> Do you have an idea please ?
>>>>
>>>> thanks
>>>>
>>>> Le jeu. 12 déc. 2019 à 11:09, Mickael Hubert <mickael at winlux.fr> a
>>>> écrit :
>>>>
>>>>> Hi all,
>>>>> I have an issue, opensips doesn't want forward Re-INVITE during UDP to
>>>>> TCP mapping session.
>>>>> Customer (NATTED) -- TCP --> (public interface listen tcp:8060)
>>>>> OpenSIPS (private interface listen udp:5060) --> rest of infrastructure
>>>>> (udp:5060)
>>>>>
>>>>> I can send a call from customer to OpenSIPS (initial INVITE, 200OK,
>>>>> etc ...).
>>>>> But when I received Re-INVITE from UAS (sip session timer), OpenSIPS
>>>>> doesn't forward it to customer.
>>>>>
>>>>> You can see the call flow here:
>>>>> https://photos.app.goo.gl/eUSb5MvBhUfueaoM7
>>>>> You can see SIP messages and opensips's logs in txt file in attachment
>>>>>
>>>>> OpenSIPS handles on it's public interface:
>>>>> - Topology hiding
>>>>> - Nat detection
>>>>>
>>>>> Thanks a lot for you help !
>>>>>
>>>>> _______________________________________________
>>>> Users mailing list
>>>> Users at lists.opensips.org
>>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>>>
>>> --
>>> Regards,
>>>
>>> David Villasmil
>>> email: david.villasmil.work at gmail.com
>>> phone: +34669448337
>>> _______________________________________________
>>> Users mailing list
>>> Users at lists.opensips.org
>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>>
>> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
-- 
Regards,

David Villasmil
email: david.villasmil.work at gmail.com
phone: +34669448337
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20191216/9e19c5da/attachment-0001.html>