[asterisk-bugs] [JIRA] (ASTERISK-25408) One RTP stream is lost out of the NIC for approx 5 sec then returns

[Kevin Scott Adams (JIRA)]

Kevin Scott Adams created ASTERISK-25408:
--------------------------------------------

             Summary: One RTP stream is lost out of the NIC for approx 5 sec then returns
                 Key: ASTERISK-25408
                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-25408
             Project: Asterisk
          Issue Type: Bug
      Security Level: None
          Components: Bridges/bridge_native_rtp
    Affects Versions: 13.5.0
         Environment: CentOS 6. Hardware specs is a KVM with 8G memory and 4 processors. One NIC but the VM is by itself on it's own Dell R710 with 48Gb, 24 cores and 4 Ethernets.
            Reporter: Kevin Scott Adams


The RTP example is simple...

Endpoint A --Stream A--> Asterisk SBC --Stream C---> Endpoint B
Endpoint A <--Stream B-- Asterisk SBC <--Stream D-- Endpint B

Endpoint A is an Audiocodes Mediant 1000 and Endpoint B is a Shoretel 480 or 485 phone via a Shoretel PBX (I know I don't like it either).
For some reason (and this is where you might be able to lead me to give you more info) that Stream C disappears for approximately 5 seconds. There is no set condition that I can see on the server that leads to it except that it does seem to be when traffic is at a minimum.

Using NGREP to view traffic at the interface, I see the RTP streams between the devices transmitting both ways..but then just Call Path C disappears...A, B and D keep transmitting.

What I am not sure of is if the RTCP packets have anything to do with reestablishing the RTP stream or even a re-invite from the Shoretel device which I do see. Everything is uLaw (g711) so it's pretty simple.

Kamailio is in the mix between the Audiocodes and the Asterisk SBC for SIP LCR and Load Balancing devices.

Call duration does not seem to be a factor...I have listened to BBC across the phone for 4+ hours and there seems to be no time-of-day factor. It will happen a few times during that call but never drops the call...only the one RTP stream.

I have played around with the rtpstart and rtpend and along with strictrtp to yes and probation to 8. I thought it worked but I think some of the employee's just live with it.

Because it is a hard thing to follow, I thought maybe you'll could give me some more direction on getting more data for you.

I have not put it past Shoretel that they are telling Asterisk to reestablish the RTP stream...I just can't prove it.



--
This message was sent by Atlassian JIRA
(v6.2#6252)