I am doing load testing of my website hosted in nginx pod on a GKE k8s cluster. I am using jmeter to simulate traffic to the website. The traffic flows normally as i increase the load upto 40req/sec. If the load goes above that then the requests start to fail with connection timeout errors -

Once the requests start to fail, this goes on for about 5 mins. To determine if this was a load balancer issue or rate limiting in nginx, I port forwarded the nginx service locally and ran the tests. This time the tests went without any issues all the way upto 70req/sec. So this should be due to some rate limiting behavior imposed by the GKE load balancer.

So I have a few questions.

1. How can I investigate this further?
2. Where can I see the logs for rejected requests?
3. Is there a way to control the rate limiting?

Thanks