[OpenSIPS-Users] Call forking, branches, Record-routing

Fri Apr 1 15:13:14 UTC 2022

Yes, I originally assumed that I could not use two TLS sockets on one
interface. My last sentence in the last post showed that I understood
it now.
As you said, I now have port forwarding to another port and two
separate sockets for public and private phones.

socket = udp:10.0.2.3
socket = tls:10.0.2.3:5061
socket = tls:10.0.2.3:5062 as PUBLIC-IP:PUBLIC-PORT

It seems to work now. I now have to check the other cases.

Many thanks for the help and patience.

Have a nice weekend!

Best regards,

Karsten

Am Freitag, dem 01.04.2022 um 17:55 +0300 schrieb Bogdan-Andrei Iancu:
> Just to be sure, by sockets I mean OpenSIPS listeners. Even with one
> IP,
> you can have something like PUB:5060->NAT->PRIV:5070 and have
> OpenSIPS
> listening on PRIV:5070 (with advertise for external traffic) and
> PRIV:5060 for internal.
>
> Regards,
>
> Bogdan-Andrei Iancu
>
> OpenSIPS Founder and Developer
>    https://www.opensips-solutions.com
> OpenSIPS eBootcamp 23rd May - 3rd June 2022
>    https://opensips.org/training/OpenSIPS_eBootcamp_2022/
>
> On 4/1/22 5:38 PM, Karsten Wemheuer wrote:
> > Hi Bogdan-Andrei,
> >
> > unfortunately, only one interface is available for communication to
> > the
> > phones.
> >
> > For the single interface I have two sockets, one for UDP and one
> > for
> > TLS. TLS is used to access local phones and those over NAT.
> >
> > listen = udp:10.0.2.3
> > listen = tls:10.0.2.3:5061
> >
> > The way that doesn't work is a call coming from the PBX via UDP and
> > going to a phone via TLS behind a NAT. How can I influence the
> > CANCEL
> > request here, especially if it comes internally from OpenSIPs?
> >
> > Maybe I should use another port for the public side. Than it should
> > be
> > possible to use two sockets for tls.
> >
> > Thanks again.
> >
> > Best regards,
> > Karsten
> >
> > Am Freitag, dem 01.04.2022 um 17:04 +0300 schrieb Bogdan-Andrei
> > Iancu:
> > > Oh, so you have traffic both from public and private network,
> > > right ?
> > > If
> > > so, you have 2 options:
> > >
> > > 1) use a single socket, without advertise and use the script
> > > advertise
> > > function depending on the source of the call - see
> > > set_advertised_address() [1]
> > >
> > > 2) use 2 sockets, one for public traffic, with advertise and one
> > > for
> > > the
> > > internal traffic, without advertise. For each call, control the
> > > outbound
> > > interface via the $socket_out variable [2]
> > >
> > > My 2 cents, (1) may be too complicated as the
> > > set_advertised_address()
> > > function as a message level scope, so you need to take care and
> > > do
> > > it
> > > for each and every single request of the call. With (2) you need
> > > to
> > > do
> > > it only for the initial request (INVITE) as the sockets are
> > > remembered
> > > for the whole duration of the call by the Record Route mechanism.
> > >
> > > [1]
> > > https://www.opensips.org/Documentation/Script-CoreFunctions-3-2#set_advertised_address
> > > [2]
> > > https://opensips.org/Documentation/Script-CoreVar-3-2#socket_out
> > >
> > > Best Regards,
> > >
> > > Bogdan-Andrei Iancu
> > >
> > > OpenSIPS Founder and Developer
> > >     https://www.opensips-solutions.com
> > > OpenSIPS eBootcamp 23rd May - 3rd June 2022
> > >     https://opensips.org/training/OpenSIPS_eBootcamp_2022/
> > >
> > > On 3/31/22 7:27 PM, Karsten Wemheuer wrote:
> > > > Hi Bogdan-Andrei,
> > > >
> > > > in case of global advertising is active and set to the natted
> > > > address
> > > > the advertised address is used, but this leads to problems
> > > > using
> > > > phones
> > > > in the LAN.
> > > >
> > > > As written in my other post: Without setting the advertise
> > > > address
> > > > and
> > > > port, I have a problem with the phones behind NAT. Is it
> > > > possible
> > > > to
> > > > manipulate the route before in a branch or something like that?
> > > >
> > > > Regards,
> > > >
> > > > Karsten
> > > >
> > > > Am Donnerstag, dem 31.03.2022 um 18:53 +0300 schrieb Bogdan-
> > > > Andrei
> > > > Iancu:
> > > > > Hi Karsten,
> > > > >
> > > > > You say the record_route() does not take into consideration
> > > > > the
> > > > > global
> > > > > advertising ??
> > > > >
> > > > > Regards,
> > > > >
> > > > > Bogdan-Andrei Iancu
> > > > >
> > > > > OpenSIPS Founder and Developer
> > > > >      https://www.opensips-solutions.com
> > > > > OpenSIPS eBootcamp 23rd May - 3rd June 2022
> > > > >      https://opensips.org/training/OpenSIPS_eBootcamp_2022/
> > > > >
> > > > > On 3/31/22 6:22 PM, Karsten Wemheuer wrote:
> > > > > > Hi Bogdan-Andrei,
> > > > > >
> > > > > > many thanks for Your help.
> > > > > >
> > > > > > I tried with record_route. It doesn't work for me, as I set
> > > > > > "advertised_address" and "advertised_port" to the natted
> > > > > > address of
> > > > > > the
> > > > > > (only) interface. I wasn't able to avoid this. It seemed to
> > > > > > be
> > > > > > required
> > > > > > to be able to reflect the path "phone -> proxy -> pbx".
> > > > > >
> > > > > > I removed the "advertised"-stuff and checked again the call
> > > > > > with
> > > > > > record_route. Now this seems to work.
> > > > > >
> > > > > > I think, I have to fix the other call flow to avoid the
> > > > > > global
> > > > > > setting
> > > > > > of the advertised address and port.
> > > > > >
> > > > > > Best regards,
> > > > > >
> > > > > > Karsten
> > > > > >
> > > > > > Am Donnerstag, dem 31.03.2022 um 17:44 +0300 schrieb
> > > > > > Bogdan-
> > > > > > Andrei
> > > > > > Iancu:
> > > > > > > Hi Karsten,
> > > > > > >
> > > > > > > See my prev email, just to record_route() before the
> > > > > > > t_relay()
> > > > > > > for
> > > > > > > the
> > > > > > > initial INVITE. And the loose_route() stuff for whatever
> > > > > > > sequential/in-dialog requests.
> > > > > > >
> > > > > > > Best regards,
> > > > > > >
> > > > > > > Bogdan-Andrei Iancu
> > > > > > >
> > > > > > > OpenSIPS Founder and Developer
> > > > > > >       https://www.opensips-solutions.com
> > > > > > > OpenSIPS eBootcamp 23rd May - 3rd June 2022
> > > > > > >
> > > > > > > https://opensips.org/training/OpenSIPS_eBootcamp_2022/
> > > > > > >
> > > > > > > On 3/31/22 2:50 PM, Karsten Wemheuer wrote:
> > > > > > > > Hi*,
> > > > > > > >
> > > > > > > > I have a understanding problem regarding branches and
> > > > > > > > call
> > > > > > > > forking.
> > > > > > > > A call from a PBX is to be routed to phone(s) via
> > > > > > > > OpenSIPS.
> > > > > > > > The
> > > > > > > > phones
> > > > > > > > are registered to OpenSIPs.
> > > > > > > >
> > > > > > > > INVITE --> lookup ----> 1. Destination
> > > > > > > >                        |
> > > > > > > >                        \--> 2. Destination
> > > > > > > >
> > > > > > > > When the call is terminated by the caller, the BYE
> > > > > > > > request
> > > > > > > > shall
> > > > > > > > take
> > > > > > > > the same path. Currently, the BYE is sent from the PBX
> > > > > > > > directly
> > > > > > > > to
> > > > > > > > the
> > > > > > > > Contact URI (which is not reachable by the PBX).
> > > > > > > >
> > > > > > > > Is it possible to use record_route in the branch_route
> > > > > > > > so
> > > > > > > > that
> > > > > > > > different record route headers are used? Or is there
> > > > > > > > another
> > > > > > > > way?
> > > > > > > >
> > > > > > > > Thanks in advance,
> > > > > > > >
> > > > > > > > Karsten
> > > > > > > >
> > > > > > > >
> > > > > > > > _______________________________________________
> > > > > > > > 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
> > > > _______________________________________________
> > > > Users mailing list
> > > > Users at lists.opensips.org
> > > > http://lists.opensips.org/cgi-bin/mailman/listinfo/users