[OpenSIPS-Users] NOTIFY and TLS issue

[Bogdan-Andrei Iancu]

Hi Damien,

IMHO, in most of the cases, the subscription is generated by the party 
interested to receive the notifications, so the contact in SUBSCRIBE 
will match the source address of the SUBSCRIBE requests.
To be honest, I never came across a kind of third-party subscription 
scenario.

Anyhow, your comment is fully correct, I do thank you for the reminder 
on looking into the connection reusage cases.

Regards,

Bogdan-Andrei Iancu
OpenSIPS Founder and Developer
http://www.opensips-solutions.com

On 31.08.2015 12:29, Damien Sandras wrote:
> Hi Bogdan,
>
> I'm not sure about that (see our previous discussions about connection 
> reuse).
>
> It will only reuse the active SUBSCRIBE TCP connection if the Contact 
> header of the SUBSCRIBE indicates the same IP/port than the one used 
> to create the outbound SUBSCRIBE TCP connection. That is rarely the case.
>
> Or am I missing something ?
>
>
>
> Le lundi 31 août 2015 à 12:17 +0300, Bogdan-Andrei Iancu a écrit :
>> Hi Bogdan,
>>
>> If the conn with B is still alive (the one created by SUBSCRIBE 
>> requests), it should be reused when OpenSIPS has to send the NOTIFY. 
>> Have you enabled the tcp aliases ?
>>
>> If still a problem, can you make a log (with debug 6) when the NOTIFY 
>> is to be send + a listing from list_tcp_conns ?
>>
>> Regards,
>> Bogdan-Andrei Iancu
>> OpenSIPS Founder and Developer
>> http://www.opensips-solutions.com
>> On 28.08.2015 20:53, Bogdan Chifor wrote:
>>> Hello,
>>>
>>> I have a question regarding the following scenario:
>>>
>>> 1. I have two devices connected to the server via two-way TLS(TCP).
>>>  1.1 Device A is behind a NAT
>>>  1.2 Device B is directly connected to the server
>>>
>>> 2. Device B subscribes to the presence of device A.
>>>
>>> 3. Device A gets offline and the server generates a NOTIFY message 
>>> to be sent to device B.
>>>
>>> 4. The server does not find an existing tcp connection (from the 
>>> logs), even though the socket is visible if the "opensipsctl fifo 
>>> list_tcp_conns" or "netstat" commands are used.
>>>
>>> 5. Because the server does not find an existing connection it 
>>> initiates one (TLS). After that the proto tls module logs the 
>>> following error: "NOTICE:proto_tls:verify_callback: verify 
>>> error:num=26:unsupported certificate purpose".
>>>
>>> 6. This error is normal because device B does not have a certificate 
>>> with server authentication extended key usage, it has only the 
>>> client authentication extended key usage (as normal).
>>>
>>> What is the reason behind the start of the new connection and how 
>>> should I handle this issue?
>>>
>>> This is my proto_tls config:
>>>
>>> *modparam("proto_tls", "verify_cert", "1")*
>>> *modparam("proto_tls", "require_cert", "1")*
>>> *modparam("proto_tls", "tls_method", "TLSv1")*
>>> *modparam("proto_tls", "certificate", "...")*
>>> *modparam("proto_tls", "private_key", "...")*
>>> *modparam("proto_tls", "ca_list", "...")*
>>> *modparam("proto_tls", "ca_dir", "...")*
>>>
>>>
>>> Any help is appreciated.
>>>
>>> Best regards,
>>>
>>> Bogdan.
>>>
>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users at lists.opensips.org  <mailto:Users at lists.opensips.org>
>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opensips.org  <mailto:Users at lists.opensips.org>
>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
> -- 
> Damien SANDRAS
>
> *Ekiga Project*
> http://www.ekiga.org
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20150902/645c5542/attachment.htm>