We've been having an odd behaviour with Wowza 3.6.3 and I'd appreciate your help.
We have 5 servers at AWS, one configured as Origin and 4 as Edge servers. All of them are in the same datacenter/availability zone. Everything runs just fine, Edge servers are able to pull the stream from the origin and re-distribute to the end users. So far, so good.
However, for some very odd reason, sometimes, in a very random way, the Edge servers simply "interrupt" the stream and the errors are logged as "LiveMediaStreamReceiver.doWatchdog: streamTimeout". Consequently, users start to experience streaming loss. If I point all users to the origin server, everything runs just smoothly.
I've tried pointing the Edge servers both the internal and to the public IP of the Origin server, but it doesn't make any difference. Also, while monitoring all servers, they appear perfectly "healthy", with plenty of CPU and RAM to use.
So, our doubt now is "what could possibly be causing those random 'streamTimeout' events"? Anyone else has ever experienced that?
Thanks in advance,