How to optimize apprtc gae Web Services

[chenwandong]

Browsers and versions affected

Description

I use gae SDK to run apprtc on Linux platform, use JMeter test tool to access apprtc concurrently 200 times per second, and modify app.yaml files: automatic_ scaling max_ concurrent_ requests =10 max_ instances =2
； I have observed that gae has gradually started more than 300 Python processes to process these HTTP services; My question is how to limit gae to create so many Python processes. I modify max_ concurrent_ The requests value also has no effect. Gae will still gradually start hundreds of Python to process HTTP services; Because if I don't limit the number of started Python instances, the coarse ram resources in the server will be exhausted, so I must limit the running of started Python instances

Steps to reproduce

Expected results

Actual results