Dear all,
I'm working to separate on two different servers registrar and mediaproxy.
Mediaproxy server (with both mediaproxy and ser proxy process) will be the front-end and all registers and calls from Intenet will be directed to this server and then routed to internal registrar.
Now, I'm working with natted clients. I've started with only one server (with only one ser) and it works fine. When I've tried to split the architecture on two server I can't fine a good configuration. In particular, when a user with a public IP calls a natted user, the RTP doesn't use mediaproxy Anyone can help me? do you have any working configurations?
thanks,
Andrea
Hi,
I am not sure if you want to keep only openser+mediaproxy module+mediaproxy in front of registrar (openser+registrar module+usrloc module). If yes, this is rather impossible right now, mediaproxy module has to be deployed together with usrloc module in order to send NAT pings to NATed sip phones (the usrloc module stores the needed information about phones' contact addresses).
Cheers, Daniel
On 10/04/05 13:49, andrea wrote:
Dear all,
I'm working to separate on two different servers registrar and mediaproxy.
Mediaproxy server (with both mediaproxy and ser proxy process) will be the front-end and all registers and calls from Intenet will be directed to this server and then routed to internal registrar.
Now, I'm working with natted clients. I've started with only one server (with only one ser) and it works fine. When I've tried to split the architecture on two server I can't fine a good configuration. In particular, when a user with a public IP calls a natted user, the RTP doesn't use mediaproxy Anyone can help me? do you have any working configurations?
thanks,
Andrea
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Dear Daniel,
I'd implement this architecture because I need to separate front-end proxy from core network (registrar, voice gateway and other apps). Can you suggest me another possible architecture?
A possible approach can be: - openser+mediaporxy module+usrloc module receive register, applies fix_nat_contact and then sends it to the internal register - on INVITE openser+mediaproxy module+usrloc can lookup directly into registar DB and then decides howto handle the sip message any other suggestions? thanks
Andrea
Daniel-Constantin Mierla wrote:
Hi,
I am not sure if you want to keep only openser+mediaproxy module+mediaproxy in front of registrar (openser+registrar module+usrloc module). If yes, this is rather impossible right now, mediaproxy module has to be deployed together with usrloc module in order to send NAT pings to NATed sip phones (the usrloc module stores the needed information about phones' contact addresses).
Cheers, Daniel
On 10/04/05 13:49, andrea wrote:
Dear all,
I'm working to separate on two different servers registrar and mediaproxy.
Mediaproxy server (with both mediaproxy and ser proxy process) will be the front-end and all registers and calls from Intenet will be directed to this server and then routed to internal registrar.
Now, I'm working with natted clients. I've started with only one server (with only one ser) and it works fine. When I've tried to split the architecture on two server I can't fine a good configuration. In particular, when a user with a public IP calls a natted user, the RTP doesn't use mediaproxy Anyone can help me? do you have any working configurations?
thanks,
Andrea
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Hi Andrea,
On 10/04/05 16:55, andrea wrote:
Dear Daniel,
I'd implement this architecture because I need to separate front-end proxy from core network (registrar, voice gateway and other apps). Can you suggest me another possible architecture?
A possible approach can be:
- openser+mediaporxy module+usrloc module receive register, applies
fix_nat_contact and then sends it to the internal register
- on INVITE openser+mediaproxy module+usrloc can lookup directly into
registar DB and then decides howto handle the sip message any other suggestions?
You must have in mind that in most of the cases, the phone behind the nat can be reachable only from the system where it had sent the register request (pinhole binding for symmetric NATs takes in consideration the source address/port and destination address/port). So, even if you duplicate the registrations and have a nat traversal box + registrar box, the registrar has to route the messages through nat traversal box when it needs to contact the phone. In this case, you have to store the address of the nat traversal box along the contact address.
The second option needs a cacheless usrloc module and same logic as I said above.
There are some devices dedicated for SIP nat traversal, but there none free I am aware. If you do not want to develop one, you have to go for one of them.
Cheers, Daniel
thanks
Andrea
Daniel-Constantin Mierla wrote:
Hi,
I am not sure if you want to keep only openser+mediaproxy module+mediaproxy in front of registrar (openser+registrar module+usrloc module). If yes, this is rather impossible right now, mediaproxy module has to be deployed together with usrloc module in order to send NAT pings to NATed sip phones (the usrloc module stores the needed information about phones' contact addresses).
Cheers, Daniel
On 10/04/05 13:49, andrea wrote:
Dear all,
I'm working to separate on two different servers registrar and mediaproxy.
Mediaproxy server (with both mediaproxy and ser proxy process) will be the front-end and all registers and calls from Intenet will be directed to this server and then routed to internal registrar.
Now, I'm working with natted clients. I've started with only one server (with only one ser) and it works fine. When I've tried to split the architecture on two server I can't fine a good configuration. In particular, when a user with a public IP calls a natted user, the RTP doesn't use mediaproxy Anyone can help me? do you have any working configurations?
thanks,
Andrea
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Daniel,
if I understand well, I can implement this architecture: - one or more mediaproxy (openser+mediaproxy module+usrloc module) as front-end - one registrar which must store also infos about the mediaproxy sending message (only if I've more then one mediaproxy; if I've only one mediaproxy I don't need this info) - when a user call a natted user the registrar have to send the message to the wright mediaproxy (no problem if I've only one...)
A second question: there is a better (easiest?) approch? ser/openser as "nat box" is only designed as "all in one" system? thanks,
Andrea
Daniel-Constantin Mierla wrote:
Hi Andrea,
On 10/04/05 16:55, andrea wrote:
Dear Daniel,
I'd implement this architecture because I need to separate front-end proxy from core network (registrar, voice gateway and other apps). Can you suggest me another possible architecture?
A possible approach can be:
- openser+mediaporxy module+usrloc module receive register, applies
fix_nat_contact and then sends it to the internal register
- on INVITE openser+mediaproxy module+usrloc can lookup directly into
registar DB and then decides howto handle the sip message any other suggestions?
You must have in mind that in most of the cases, the phone behind the nat can be reachable only from the system where it had sent the register request (pinhole binding for symmetric NATs takes in consideration the source address/port and destination address/port). So, even if you duplicate the registrations and have a nat traversal box + registrar box, the registrar has to route the messages through nat traversal box when it needs to contact the phone. In this case, you have to store the address of the nat traversal box along the contact address.
The second option needs a cacheless usrloc module and same logic as I said above.
There are some devices dedicated for SIP nat traversal, but there none free I am aware. If you do not want to develop one, you have to go for one of them.
Cheers, Daniel
thanks
Andrea
Daniel-Constantin Mierla wrote:
Hi,
I am not sure if you want to keep only openser+mediaproxy module+mediaproxy in front of registrar (openser+registrar module+usrloc module). If yes, this is rather impossible right now, mediaproxy module has to be deployed together with usrloc module in order to send NAT pings to NATed sip phones (the usrloc module stores the needed information about phones' contact addresses).
Cheers, Daniel
On 10/04/05 13:49, andrea wrote:
Dear all,
I'm working to separate on two different servers registrar and mediaproxy.
Mediaproxy server (with both mediaproxy and ser proxy process) will be the front-end and all registers and calls from Intenet will be directed to this server and then routed to internal registrar.
Now, I'm working with natted clients. I've started with only one server (with only one ser) and it works fine. When I've tried to split the architecture on two server I can't fine a good configuration. In particular, when a user with a public IP calls a natted user, the RTP doesn't use mediaproxy Anyone can help me? do you have any working configurations?
thanks,
Andrea
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
-
andrea -
Daniel-Constantin Mierla