[OpenSIPS-Users] Mediaproxy & "Zombie" Media Sessions.

Alan Frisch frisch.alan at gmail.com
Tue Jun 22 03:46:01 CEST 2010


Saul,

Thanks for looking into this.  The box that is suffering from this has
OpenSIPS (5060) fronting an Asterisk (5080), the MP relay and
dispatcher (default ports) on the same IP.  All RTP ranges are
separated from each other.

Doing a netstat -uap shows RTP ports being held open by the
media-relay/python process, though the corresponding Asterisk ports in
CDRTool's media session list have long been closed.

I noticed something else, that may or may not be related to what I
posted in an earlier message.  Take the following, a slightly
different example on a single server:

eth0 - 209.x.x.100
lo1: loopback IP 1 - 64.x.x.120  (MP Dispatcher)
lo:2 loopback IP 2 - 64.x.x.121  (MP Relay)

When the Relay's relay_ip is set to 64.x.x.121 and the Dispatcher has
listen and listen_management set to 64.x.x.120, once the relay
connects into the dispatcher, the dispatcher reports it coming from
the 64.x.x.120 rather than the 64.x.x.121 IP.

If you want to force OpenSIPs to use the media-relay on this machine,
you have to tell it to connect to the .120 relay rather than the .121
address.  All media streams however are sent from the 121 address, as
they should be.

Thanks again!

AF.


On Mon, Jun 21, 2010 at 5:30 PM, Saúl Ibarra Corretgé
<saul at ag-projects.com> wrote:
> Hi,
>
> On 21/06/10 19:03, Alan Frisch wrote:
>> Been running Mediaproxy in production for the last 12 hours with some
>> heavy loads and have noticed that there are many "zombie" media
>> sessions listed when using the CDRTool page to view current MP
>> sessions.
>>
>> These calls no longer exist in OpenSIPS dialog table nor the media
>> servers (Asterisk boxes) that originally served them.  I am using two
>> relays and it seems that all of these zombie sessions are occurring on
>> the box where I have the relay and dispatcher stuffed onto the same
>> IP.  Could the dispatcher and relay residing on the same IP cause this
>> issue?
>>
>> For instance, I have several calls that say over 5 hours in length,
>> but with 0 bytes listed for incoming/outgoing RTP.  My timeouts for
>> OpenSIPS dialog is set to 7200, stream_timeout=90 and
>> on_hold_timeout=7200 in Mediaproxy.
>>
>
> I also noticed this and will look into it asap.
>
>> Is there a way to manually remove these zombie media sessions?
>>
>
> You may stop the relay and start it again.
>
>
> Thanks for your report!
>
>
> Regards,
>
> --
> Saúl Ibarra Corretgé
> AG Projects
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>



More information about the Users mailing list