Incorrect request response mapping for a few requests in a Kubernetes environment

I have installed packetbeat OSS (v6.8.3) using this YAML reference on a K8s cluster (v 1.10.13) with Docker (v 17.9.0) running on Debian GNU/Linux 9 (stretch) (4.9.0-7-amd64).

The traffic ingress to our applications works this way.
AWS ALB -> A K8s node port -> Ambassador API Gateway pod (running inside the cluster) -> K8s service -> Application pod

However, I have observed that the requests were being mapped to responses from other services running in the same cluster in some rare cases. Is there a way to fix this? Or Is this a bug in Packetbeat?
Also, it would be great if someone can help me understand the request-response mapping criteria in an environment like Kubernetes.
Also, Kubernetes metadata is not being captured despite using the above configuration.