I have a question about the kind of traffic that the clients send/receive to/from the videobridge.
I disabled the cam and the mic of all clients, but I see that there are 70~100kbps from/to the videobridge of traffic.
do you know which kind of traffic is this?
Thank you very much for information.
This should be audio and data channel with stats.
another question about this scenario I see with tcpdump that a high part of the traffic is RTCP (receiver report).
do you know where I can find in the code of videobridge when this packet is sent? I need to understand the frequency.
RTCP is send by the clients every few seconds, not sure about the interval. Most of the traffic should be RTP.
Let me explain better my problem:
I am running an experiment where I have 20 clients connected to a jitsi VB. The clients have the webcam disable and all of them are muted except for one.
My expectation is to see around 1000kbps of traffic arrive at the VB and more or less the same traffic going out. This is because all the clients send an audio stream even when they are muted the VB drops the muted streams. Then I expect that the VB sends only audio from the unmuted client to all the other users. I see more or less the expected traffic going in, but the output traffic is much higher (a bit more than twice the input traffic). Most of the output traffic seems to composed by RCTP RR packets (payload type 201). I also noticed that each client receives the RTCP RR packets from all the other clients, while I was expecting that the VB was filtering them.
Did I miss configure something in my setting or this is the expected behavior? And do I also miss some other source of traffic or the overhead that I see is only due to RTCP packets?
@gpolitis Can you give more details for this ^? Thanks.
we don’t have rtcp termination for audio in the old bridge, not sure how this is handled in the new bridge /cc @Boris_Grozev
For what I can see from the capture done on a client it seems that the client send only one RR packet toward the VB and receives one RR packet from each user in the call. So I assume that the VB replicate this packets for the all the participants. Is this the way it works? This would explain the majority of the overhead that I see with respect to my initial expectation. And one last question: is there any other source of possible overhead that I didn’t found in addition the RTCP traffic?
Thank you very much
Do you have info?