WebRTC Performance issue

[Jonesy]

Hi - I recently setup in AWS webcallserver v5.2.267,

In testing WebRTC streaming , from a RTSP input stream, I am only able to get about 450 WebRTC streams out from the server until the stream stops responding and i have to restart the webcallserver service.

I have made corrections to java heap but still see the issue. What else can i do to fix this?

On older version 5.1.3611 I was able to get about 1000 streams output before the server started to show issues with the stream.

Thanks for any help you can provide.

 

[Max]

Good day.
Perhaps there is not enough hardware (for example, a processor or memory) for the server to work and your AWS instance needs to be upgraded.
Please give us statistics during the maximum load hours from the page:

https://{Your WCS server}:8888/?action=stat

And besides please send SSH access to support@flashphoner.com
We will check.

 

[Jonesy]

Thank you Max, I sent you the SSH access via email address above.
I have also matched the AWS instance type now to the machines that we have in production which can handle ~1000 concurrent streams. (m4.2xlarge)
We will be load testing again this afternoon and will update once I have those stats

 

[Max]

Good day.
We have analyzed your WCS server, but today it is unavailable and we offer the following steps (by previously collected data from server):
1. Recommended to allocate at least 1/2 of server physical memory for Java memory heap. You server RAM is 32 Gb, then it is recommended to allocate 16 Gb with the following settings in wcs-core.properties file:

-Xmx16g
-Xms16g

2. Check your network load during streaming. Probably not enough bandwidth (1 Gb/s). For further advice, please specify bitrate you use in stream (to each subscriber, approximate).
3. What incoming streams do you use (RTMP, RTSP, SIP) and how many.

If it doesn’t help (increased Java memory heap), at the time the problem occurred please collect a report as described here (you can collect logs using script) and send to support@flashphoner.com, we will check.

 

[Max]

Good day.
We’ve connected to your server to verify the configuration and we see that it has changed:

1. You have reduced the number of media ports. Up to 4 media ports are used per WebRTC stream. In the setting for using media_port, change to large values. For example:

media_port_from = 31001
media_port_to = 51000

2. In the logs we see a large number of messages:

"info": "Failed by ICE timeout",

Your server is located behind NAT and you may need to configure port forwarding on the Border Router.
You can check the availability of ports (as indicated in the documentation).

3. If in browser settings WebRTC bitrate are not defined, server settings are applied. Try lowering bitrate of WebRTC streams to check the network load:

webrtc_cc_min_bitrate = 300000
webrtc_cc_max_bitrate = 500000

3. Indicate time when testing will be carried out, we will SSH-connect and check.