AKS集群孤立资源自动删除失效问题求助

I’ve been hitting a frustrating issue with multiple AKS clusters after they’ve been running for some time: the cascading auto-deletion of orphaned resources stops working as expected, and this is breaking our business workflows that depend entirely on this functionality.

Observed Symptoms #

• When deleting a Deployment, its linked ReplicaSets or Pods aren’t removed at all
• When deleting a ReplicaSet, a new ReplicaSet gets created (as intended), but the old Pods linger in the cluster
• Some Deployments fail to update properly: triggering an update spawns new ReplicaSets and Pods, but the old resources never get cleaned up

This has already caused real business impact since our operations rely on Kubernetes handling resource cleanup automatically.

Troubleshooting Steps We’ve Tried (With No Success) #

• Re-creating the cluster does restore normal cascading deletion behavior, but we haven’t been able to pinpoint the exact operation that triggers the failure in the first place
• Upgrading both control plane nodes and agent nodes to newer Kubernetes versions didn’t fix the issue
• Restarting the kubelet service on agent nodes had no effect

Official Test Case Validation #

We confirmed the issue using the standard Kubernetes test deployment:

1. Create the deployment: kubectl apply -f https://k8s.io/examples/application/deployment.yaml
2. Delete the deployment: kubectl delete deployments.apps nginx-deployment

After running the delete command, the associated Pods still remain in the cluster

Additional Context #

This problem has popped up across 3 separate AKS clusters running different Kubernetes versions, so it doesn’t seem tied to a specific version bug.

Questions for the Community #

1. Can anyone help diagnose what’s causing this cascading deletion failure in AKS?
2. Could someone clarify which Kubernetes/AKS component is responsible for managing orphaned resource cascading deletion? Have other users run into similar issues with AKS clusters?

内容的提问来源于stack exchange，提问作者user1809666