Description of problem: A namespace contains 152 pods scheduled accross 4 nodes,
When the pods are restarted manually there is a sudden spike in terms of resource utilization, and CNI stops working on some nodes, 

Namespace: t023-cert-bck
Pod Creation Timestamp:  "2026-02-24T13:36:20Z"
Issue Time when Resource spike was there and CNI stopped working : Around 2026-02-24T15:40

I will be sharing the required logs in comments,

Please update what you need from the customer in order to troubleshoot this,