[Serusers] rtpproxy and call hold

[Klaus Darilion]

Bogdan-Andrei IANCU wrote:

> Andrei Pelinescu-Onciul wrote:
> 
>> On Jul 26, 2004 at 03:28, Richard <mypop3mail at yahoo.com> wrote:
>>  
>>
>>> Hi,
>>>
>>> How does rtpproxy handle call hold? Normally when the
>>> call is put on hold, it doesn't have any rtp packets
>>> between two end UA. Would it cause rtpproxy to tear
>>> down the connection after timeout (default 60s)?
>>>   
>>
>>
>> Yes, but on the other hand when you exit the on hold state the UA sends
>> a re-INVITE. So if you properly handle re-INVITEs for natted calls, the
>> re-INVITE will be caught by nathelper, the sdp re-written and the
>> rtpproxy binding re-created (or updated if it already exists).
>>
>> Andrei
>>  
>>
> actually when you enter on hold state, a re-INVITE  with 0.0.0.0 addr in 
> SDP is sent. So if catch also this one, you can tear down the rtp 
> session without waiting for timeout. At the second re-Invite (when exit 
> from on hold) you have to create a new rtp session

AFAIK, using 0.0.0.0 in SDP for "hold" is depreciated. You should use 
a=inactive resp. a=sendonly or a=recvonly.

See RFC 3264, sec. 8.4 for details.

regards,
klaus