I have a set of IPs outside of the k8s cluster which i want to loadbalance natively. The way to do that is using a clusterIP service without a selector and defining a endpoint resource with the ips i want to forward. When I do that, the service discovers the ips, but does not forward requests to them.

While doing a cleanup of unused resources in our AWS environment, I discovered that an previous employee had setup an AWS EKS cluster which none of us were aware of. It looks like the cluster was never deployed or used in production. Is there any specific steps that I can take to confirm it is not being used, before deleting it? Unfortunately, AWS does not offer an option to stop it, only an option to delete it.

I am very new to this concept, so I was curious to know if we can create more than one controller node in EKS cluster or not. if yes how can we do so? Also, if not, how HA/fault tolerance is achieved in this kind of setup in the case of the control plane?

I have my software running in EKS cluster, where every node can fit up to 5 pods. The software uses multiple accounts on 3rd-party service which requires IP whitelisting. I have separate manifest files for every pod and all the pods should use unique IPs (so I can whitelist them). What options do I have to implement such system? Preferably I want to specify Elastic IP in YAML file, so a pod will use it for outbound traffic.