Regular Disconnects with RoomAPI

Azhar Ali

Member
Hi,
We are experiencing regular disconnects sometimes every min while broadcasting using roomAPI.
There is only one publisher and about 300-400 listeners and it is audio-only.
We are not sure what the reason could be? If we start two rooms, the second one being the test one with one or two users in it. Only the one with more users gets disconnected.
i will emailed you the SSH access to the server in case you can have a look. This has happened several times today.
Room is called VoiceFeed.
We have also tried to stream on 4G in case there was a problem with the network but it still happens.

Thanks for looking.
 

Max

Administrator
Staff member
Good day.
The possible reason is RTP inactivity detection timer. You have audio only stream, but this timer is enabled by default and controls both audio and video packets. So you have to disable it with the following parameter in flashphoner.properties file
Code:
rtp_activity_detecting=false,60
WCS should be restarted to apply this setting
 

Azhar Ali

Member
Hi Max,

I will try that setting but I am not sure if that is the reason for it. This is because we only broadcast between 8 am ET-4pm ET and busiest time for the stream when there are most connections is around 8 am to 12 ET. During that time we get most disconnections. This is also the time when we have most connections too up to 500 on Friday. Once that time is passed, the stream gets stable and hardly any disconnections happen.
 

Max

Administrator
Staff member
Good day.
In your server logs we see a lot of messages "RTP ACTIVITY EVENT" followed by client disconnection during time period you mentioned. So it is most possible reason.
 

Azhar Ali

Member
Thanks Max.

We will give it a go, we also got a response back from our network team, do you think that could be a factor too?
I have performed some diagnostics, please see my findings below:
The 100 Mbps VPLS service VPL105950 (Tenant wired/wifi VPLS) has been heavily overutilized, ADVA has been dropping traffic to shape it to the desired 100M
This has been affecting routing protocol stability. I have made some minor changes to our core network to improve the routing protocol performance.
 
Top