Issue with metric container_network_receive_bytes_total

[DeenDe]

What happened?
I''ve got issue with container_network_receive_bytes_total metric. I have significantly difference between bytes received and transmitted. I checked this metrics with iperf between two pods. Bytes transmitted are correct, but bytes received are multiple times smaller. I dont have this issue on newer version of Kubernetes, with the same version of hełm chart, so I believe three is a problem with the cluster.


Do you have any suggestions what is causing the issue?

Environment
Fedora CoreOS 33
CNI - cilium v1.10.5

Prometheus Operator version:
image: quay.io/prometheus-operator/prometheus-operator:v0.53.1
helm chart kube-prometheus-stack 30.2.0
Kubernetes version information:

Kubernetes 1.21.5 (images v1.21.5-rancher1)

Kubernetes cluster kind:

Cluster created with Opoenstack Magnum