[Live-devel] Bad performance with 4 or more clients at proxy server

[Frank van Eijkelenburg]

Hi Ross (and others),

I increased the buffer size to 16 MB and there are still performance 
issues. However, sometimes it seems to be working for a few seconds and 
than the performance is getting worse. So increasing the socket buffer 
size in the "RTPSink" object has a little effect, but does not solve the 
problem. I verified the requested buffer size.

Best regards,

Frank van Eijkelenburg

On 02-06-15 08:02, Ross Finlayson wrote:
>> Good guess, I am indeed running the vlc players at the same computer. 
>> However, if I use multiple proxy servers and have one player per 
>> proxy server it is running fine. In that case there are still 
>> multiple copies of vlc on the same computer. What could be the 
>> explanation for this behavior?
>
> I’m not sure.  However, earlier, you wrote:
>
>> I already tried to increase the buffer size in the RTSPServer 
>> (increaseSendBufferTo) to 2 MB, but it does not result in a better 
>> performance. By the way, I also increased the buffer size at the OS 
>> (Linux) with sysctl net.core.rmem_max.
>
> Make sure that you’re increasing the OS buffer size for the socket 
> that matters: *Not* the RTSP server’s main socket (because that’s used 
> only to handle connections from new clients), nor the 
> “RTSPClientConnection” socket (because that’s used only to handle 
> incoming RTSP commands, unless you’re streaming everything 
> RTP-over-TCP), but instead the socket that’s used by the “RTPSink” 
> objects (to transmit RTP/UDP packets).
>
> Unfortunately right now I don’t know of a way to access this socket 
> without modifying the “ProxyServerMediaSession.cpp” code, but you can 
> access this socket in the 
> “ProxyServerMediaSubsession:createNewRTPSink()” implementation 
> (starting at line 554 of “ProxyServerMediaSession.cpp”).  The socket 
> in question is
> rtpGroupsock->socketNum()
>
> Note also that the “increaseSendBufferTo()” function returns the 
> resulting size of the buffer, so you can look at the result of this 
> call, to see what the resulting buffer size actually becomes.
>
> I’d be interested to see if increasing the RTP socket’s OS buffer size 
> alleviates your problem at all.
>
> Ross Finlayson
> Live Networks, Inc.
> http://www.live555.com/
>
>
>
> _______________________________________________
> live-devel mailing list
> live-devel at lists.live555.com
> http://lists.live555.com/mailman/listinfo/live-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.live555.com/pipermail/live-devel/attachments/20150602/b0bda91d/attachment.html>