aws-observability / aws-sigv4-proxy-admission-controller Goto Github PK

Hi,

We faces the issue HPA stopped working because it can't read CPU utilization after injecting aws-sigv4-proxy as a sidecar by this admission controller.
I have investigated the issue, then found this github issue. According to the comment we have to define resources to all sidecars to work w/ HPA.

We are facing an issue where main container stops and aws-sigv4-proxy side car container is still hanging and not exited. Is there a process to handle graceful termination of side car container?

Internal error occurred: failed calling webhook “sigv4-aws-sigv4-proxy-admission-controller.k8s.aws”: failed to call webhook: Post “[https://sigv4-aws-sigv4-proxy-admission-controller-webhook-service.kube-system.svc:443/mutate?timeout=10s](https://sigv4-aws-sigv4-proxy-admission-controller-webhook-service.kube-system.svc/mutate?timeout=10s)”: no endpoints available for service “sigv4-aws-sigv4-proxy-admission-controller-webhook-service”

How to temporarily resolve if your cluster experiences this issue: aws/eks-charts#795 (comment)

Similar to aws/eks-charts#795 (comment)

The cluster spontaneously bricks itself (no pods come back into the cluster, due to being unable to) sometimes when this admission controller is used because when the webhook is unavailable it has no fallback.

It is something to do with the certificate generation and management system within the chart.