[pulseaudio-discuss] Make discoverable PA network sound devices available locally results in huge network traffic

[Brian J. Murrell]

Maarten Bosmans <mkbosmans <at> gmail.com> writes: 
> 
> This alone should not result in a significant amount of packets, only
> a couple of mDNS requests.

Indeed, that was my expectation also.
 
> I assume you also loaded module-zeroconf-publish on the other machine.

No, in fact it appears to be commented out in the /etc/pulse/default.pa of the 
workstation (where I hear the audio from the laptop):

#load-module module-zeroconf-publish

> Looking at the traffic with wireshark reveals a bit of mDNS chat
> between the machines and a short burst of packets every 10 seconds. By
> no means a constant spew. The burst of packets every 10 seconds is
> probably due to module-tunnel requesting the latency of each sink.
> It's not a lot of traffic though.

No, I'd be quite happy with that level of traffic.

> Hmm, this really sound more like you have activated rtp-send.

I don't think I have.  How can I verify?

> Well, other than what I described above, I can't reproduce it here, so
> it would probably not be expected. Could you look with wireshark to
> see a bit more clear what the packets are?

Well, of course since there isn't a decode module for the pulse protocol I can't 
really tell what they all are, but in terms of "statistics" a capture of just 
port 4713 between the two machines over a period of 205 seconds resulted in 
234130 packets which was an avg. pps of 1141, with an avg. packet size of 568 
bytes yielding an avg. bandwidth use of 5Mb/s!
 
> BTW, a wireshark protocol analyzer for the native pulse tcp protocol
> would be awesome.

Way awesome.  :-)

b.