Some RTMP feeds not reconnecting after restart

Hi All,

I’m running three dedicated Wowza origin servers which our CDN pulls from. Our first 2 servers work extremely well. We just added our third, using the exact same configuration, and the inputs being the exact same Teradek encoders we use with the original 2 servers.

Whenever I restart the live application in wowza (or restart the wowza process) on our third server, a couple of RTMP input feeds simply drop. It’s different feeds that drop each time I restart. All the encoders are configured to auto-reconnect.

There is no one geo-location that it consistently occurs from either (many of those locations are feeding our original 2 servers without any issues as well). This does not occur on our original 2 servers.

Any ideas?

Hi,

There’s really not much to go on here. Any entries in the encoder logs at the time of drop? corresponding entries in the access logs?

You can post anything you find.

Server software, and Java versions are the same?

Daren

Hi,

That’s possible, but but the backup could encounter the same problem if, for instance, the problem is a connection issue between encoders and the particular server.

You’ll really want to compare everything that is different between the working and problematic server setup to understand where the problem might be.

Daren

Hi Daren,

I suppose I was hoping there was something obvious. I will be checking the Java versions today.

I checked the access logs, and I don’t see any reference to the feeds that drop after the restart. They just disappear. Of course, I see the connection/transcoding of the feeds that reconnect just fine. I don’t have access to the encoder logs.

Would it be possible to put the backup encoder rtmp publish to be the same as the primary, so worse case if it doesn’t immediately reconnect and it falls over, it begins publishing to the “backup”, which would actually be our primary?

Thanks!