We run an m3.large instance on EC2, and are seeing varying performance metrics. In EC2/CloudWatch, average CPU hovers around 23%. In the Wowza Streaming Engine portal, the CPU usage hovers around 100%. See screenshots.
Adding more streams causes those streams to cut in and out intermittently. It seems that the application is at peak load, despite what EC2 says is available. (I acknowledge that RAM may be a factor, too). My theory so far is that the application is only using one of the available threads/cores on the instance.
How am I able to verify, and ultimately, change this?
Thanks for the response. I did look there, and saw that it was reporting 2 cores. This doesn’t really explain the big discrepancy between the two methods of reporting CPU usage though.
I understand. I once unknowingly had a hidden application burning up so much of my CPU, my Engine kept crashing and I was frustrated as well. Did you already tune the server and your java for optimal performance? Did you do a Heap dump to clear some memory? These things can be tricky to diagnose in forums without full access to your files to test as you know.
Engineers suggest looking at this and if still an issue they need a ticket to see what is going on.
Ignore the title of the article @Ricky_D_Amelio and I know your server is not crashing, but it’s the same set of instructions to do all the things the engineers suggested from above.
Did you send a ticket yet? They would like to further investigate and test.
I agree it’s very odd with the two different reporting scenarios. Engineers are curious to have you share in your ticket how you were able to replicate that and what the full workflow was. Thanks.
