<div dir="auto">Yes, totally identical configs and software. And yes, the effect moves to the other party if I change their roles.<div dir="auto"><br></div><div dir="auto">Another hypothesis is that the calling party is sending bad RTP for some reason. I'll collect the statistics.</div><div dir="auto"><br></div><div dir="auto"><br></div><div dir="auto"><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On 20 Sep 2017 17:30, "Brandon Armstead" <<a href="mailto:brandon@cryy.com">brandon@cryy.com</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><div dir="auto">Same kernel same everything?</div><div dir="auto"><br></div><div dir="auto"> Are they "exact replicas" identical?</div><div dir="auto"><br></div><div dir="auto">Perhaps see if you can do this test again and receive the exact system calls A vs B are doing and then B to A should give you further insight to your answer.</div><br><div class="gmail_quote"><div>On Wed, Sep 20, 2017 at 6:46 AM Stanislav Sinyagin <<a href="mailto:ssinyagin@gmail.com" target="_blank">ssinyagin@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">I'm building an environment for voice quality testing, and found a<br>
strange behavior. Maybe someone has an explanation.<br>
<br>
Two identical physical computers (PC Engines APU2), connected back to<br>
back (A and B).<br>
<br>
A initiates the test calls to B's SIP address, and then both ends go<br>
through a small dialplan (less than 100 lines of XML, similar to<br>
<a href="https://github.com/voxserv/fsqa/blob/master/fsqa.xml" rel="noreferrer" target="_blank">https://github.com/voxserv/<wbr>fsqa/blob/master/fsqa.xml</a><br>
and start recording the session.<br>
<br>
The dialplan initiates session recording to a Raw file, and plays back<br>
a 20-second audio sample, and then inserts the filename into a Redis<br>
queue, then hangs up. Then the files get processed by Sevana AQuA<br>
software in lowest CPU priority, and deleted. The files are written<br>
into a tmpfs filesystem.<br>
<br>
With 2 calls per second, I get 48 simultaneous calls, and B indicates<br>
frame drops in the audio recordings, while A shows good results. The<br>
CPU average idle time is always above 50% on both computers.<br>
<br>
Why is the B side (the one receiving INVITEs) getting more frame drops<br>
than the A (the one originating the calls)?  I flipped the sides, and<br>
then the other machine started to get more drops.<br>
<br>
At first I started recording in WAV format, and similar effect started<br>
to appear at a call every 2 seconds, resulting in 12 simultaneous<br>
calls. With raw recording, the performance is much better, but still<br>
it's a mistery why there are dropped frames in the audio.<br>
<br>
The CPU is always running at its highest frequency, under<br>
"performance" governor.<br>
<br>
______________________________<wbr>______________________________<wbr>_____________<br>
Professional FreeSWITCH Consulting Services:<br>
<a href="mailto:consulting@freeswitch.org" target="_blank">consulting@freeswitch.org</a><br>
<a href="http://www.freeswitchsolutions.com" rel="noreferrer" target="_blank">http://www.<wbr>freeswitchsolutions.com</a><br>
<br>
Official FreeSWITCH Sites<br>
<a href="http://www.freeswitch.org" rel="noreferrer" target="_blank">http://www.freeswitch.org</a><br>
<a href="http://confluence.freeswitch.org" rel="noreferrer" target="_blank">http://confluence.freeswitch.<wbr>org</a><br>
<a href="http://www.cluecon.com" rel="noreferrer" target="_blank">http://www.cluecon.com</a><br>
<br>
FreeSWITCH-users mailing list<br>
<a href="mailto:FreeSWITCH-users@lists.freeswitch.org" target="_blank">FreeSWITCH-users@lists.<wbr>freeswitch.org</a><br>
<a href="http://lists.freeswitch.org/mailman/listinfo/freeswitch-users" rel="noreferrer" target="_blank">http://lists.freeswitch.org/<wbr>mailman/listinfo/freeswitch-<wbr>users</a><br>
UNSUBSCRIBE:<a href="http://lists.freeswitch.org/mailman/options/freeswitch-users" rel="noreferrer" target="_blank">http://lists.<wbr>freeswitch.org/mailman/<wbr>options/freeswitch-users</a><br>
<a href="http://www.freeswitch.org" rel="noreferrer" target="_blank">http://www.freeswitch.org</a></blockquote></div></div><div dir="ltr">-- <br></div><div class="m_-7866660110631317565gmail_signature" data-smartmail="gmail_signature">Sent from Gmail Mobile</div>
<br>______________________________<wbr>______________________________<wbr>_____________<br>
Professional FreeSWITCH Consulting Services:<br>
<a href="mailto:consulting@freeswitch.org">consulting@freeswitch.org</a><br>
<a href="http://www.freeswitchsolutions.com" rel="noreferrer" target="_blank">http://www.<wbr>freeswitchsolutions.com</a><br>
<br>
Official FreeSWITCH Sites<br>
<a href="http://www.freeswitch.org" rel="noreferrer" target="_blank">http://www.freeswitch.org</a><br>
<a href="http://confluence.freeswitch.org" rel="noreferrer" target="_blank">http://confluence.freeswitch.<wbr>org</a><br>
<a href="http://www.cluecon.com" rel="noreferrer" target="_blank">http://www.cluecon.com</a><br>
<br>
FreeSWITCH-users mailing list<br>
<a href="mailto:FreeSWITCH-users@lists.freeswitch.org">FreeSWITCH-users@lists.<wbr>freeswitch.org</a><br>
<a href="http://lists.freeswitch.org/mailman/listinfo/freeswitch-users" rel="noreferrer" target="_blank">http://lists.freeswitch.org/<wbr>mailman/listinfo/freeswitch-<wbr>users</a><br>
UNSUBSCRIBE:<a href="http://lists.freeswitch.org/mailman/options/freeswitch-users" rel="noreferrer" target="_blank">http://lists.<wbr>freeswitch.org/mailman/<wbr>options/freeswitch-users</a><br>
<a href="http://www.freeswitch.org" rel="noreferrer" target="_blank">http://www.freeswitch.org</a><br></blockquote></div></div>