We have 5 nodes in the cluster with IPvSC enabled and kube-proxy configured in IPvSC mode: least connections. There are two applications: Application A with 2 pods and Application B with 5 pods. When a user makes a request to Application A, it creates a reverse proxy connection to the pods of Application B using the service URL. Application A is exposed through a load balancer. However, when we send 8,000 requests to the ALB, the requests are only distributed among 4 of the Application B pods (and sometimes just 2 pods). In contrast, when we increase or decrease the number of pods to a non-multiple of 5 (e.g., 10 pods result in distribution across 8), the traffic is spread across all pods. Is there a specific reason for this behavior?
Update : if we change the number of nodes to 6 ,issues occurs with multiples of 6.
