Alex Pietrow

After some investigation I can confirm it's a server-side behavior. Steps to reproduce: 1. Start reverse proxy to kube api-server: `kubectl proxy` 2. Send a GET request to get the...

@motey There is no surprise that python client assigns `None` to `kind`. After the client gets the response from the kube-apiserver, [it serializes it](https://github.com/kubernetes-client/python/blob/master/kubernetes/client/api_client.py#L192). The serializer see V1Deployment object inside...

We've run into a similar issue: `{"level":"ERROR","time":"2025-03-31T22:27:19.278Z","logger":"controller","message":"Reconciler error","commit":"058c665","controller":"nodeclass","controllerGroup":"karpenter.k8s.aws","controllerKind":"EC2NodeClass","EC2NodeClass":{"name":"default"},"namespace":"","name":"default","reconcileID":"2892bac9-03ee-408c-b355-dadb2a83962a","error":"failed to detect the cluster CIDR, operation error EKS: DescribeCluster, exceeded maximum number of attempts, 3, https response error StatusCode: 0, RequestID: ,...

Looks like I was able to reproduce the error: ### Steps to reproduce: 1. Check if Karpenter works as expected (no errors in logs, nodes are scaling up with no...

Should be resolved by #8408