I’m troubleshooting a fairly consistent crash when using the Wowza Streaming Engine 4.1.2 (it also happened with 4.1.1).
We’re running on EC2 c3.xlarge (built from your marketplace AMI).
We are using LIVE streaming from wirecast to wowza to JWPlayer.
I’m testing by opening several browser machines and pointing them to the live stream player.
The crash happens fairly consistently when I get to about 30-50 connections (usually within 15 minutes).
I’ve tried increasing to c3.8xlarge and the crash still happens.
The machine appears to have plenty mroe resources than wowza is using.
I’ve tried to simplify the wowza setup while troubleshooting.
I did a clean install of 4.1.1 upgraded to 4.1.2 from you AMI image.
I tried using a SMIL file with only the single source stream enabled.
I’ve turned off transcoder entirely.
I’ve tried skipping the SMIL file.
I’ve tried pointing to the stream playlist.m3u8 file directly.
I’ved tried some of the tips from the tuning guides.
I’ve tried changing the garbage collection settings.
I’ve tried using live, live origin, and live edge applications to compare, but wowza still crashes.
I cannot find a combination where it doesn’t crash.
In each crash case, I don’t see any significant warnings or errors in any of the wowza log files.
The CPU, memory and disk appear to be stable and consistent.
Once the crash occurs, then AWS console shows the instance as unhealthy. The server is totally unresponsive. No way to ssh in or access the wowza manager UI.
AWS cloudwatch metrics stop. Only recourse is to reboot the instance through the AWS console, after which it appears runs fine until the next crash.
Are there any known configuration issues or performance tuning settings that I should try?
Thanks,
Joe