I have a Python Application in Pod Foo. This has 5 replicas foo_1, foo_2, foo_3, foo_4, foo_5.

Each Replica has been assigned resources (in Kubernetes)

resources:
  limits:
    cpu: 1500m
    memory: 1600Mi
  requests:
    cpu: 500m
    memory: 1200Mi

After the application is successfully deployed, I am using a Load Testing Library to simulate load, which again works fine BUT the load is not distributed equally.

kubectl top pods
foo-1   1273m         489Mi           
foo-2    279m         373Mi           
foo-3    283m         365Mi           
foo-4    268m         363Mi           
foo-5    280m         374Mi        

Is there any configuration to make sure CPU load is divided equally amoung all Pods-replicas?

How are the requests sent:

• Step 1. Ingress LoadBalancer (with external IP). Ingress if from https://github.com/kubernetes/ingress-nginx/releases/download/ingress-nginx-2.5.0/ingress-nginx-2.5.0.tgz
• Step 2. NGINX Server (djnago-uwsgi-nginx stack)
• Step 3. Djnago Server is hit where the processing is done