Configmaps and Secrets Left When Destroying the Shoot in Alicloud Landscape

[xiaofenhappy]

How to categorize this issue?

/area quality /kind bug /priority normal

What happened:

Testing with the terraformer image built with the master branch, it's found that when destroying the shoot, the two configmaps and one secret related with terraformer are not cleared, causing the shoot namespace can't be terminated.

shoot--core--u43kzeaqyg Terminating 21h

$ kubectl get configmaps -n shoot--core--u43kzeaqyg
NAME                         DATA   AGE
u43kzeaqyg.infra.tf-config   2      21h
u43kzeaqyg.infra.tf-state    1      21h

$ kubectl get secrets -n shoot--core--u43kzeaqyg
NAME                       TYPE     DATA   AGE
u43kzeaqyg.infra.tf-vars   Opaque   1      21h

  kind: ConfigMap
  metadata:
    creationTimestamp: "2021-12-28T08:56:57Z"
    deletionGracePeriodSeconds: 0
    deletionTimestamp: "2021-12-28T09:13:40Z"
    finalizers:
    - gardener.cloud/terraformer
    name: u43kzeaqyg.infra.tf-state
    namespace: shoot--core--u43kzeaqyg
    ownerReferences:
    - apiVersion: extensions.gardener.cloud/v1alpha1
      blockOwnerDeletion: true
      controller: true
      kind: Infrastructure
      name: u43kzeaqyg
      uid: 00d267a0-feba-4dd7-aa33-f5ac4effd448
    resourceVersion: "739556426"
    selfLink: /api/v1/namespaces/shoot--core--u43kzeaqyg/configmaps/u43kzeaqyg.infra.tf-state
    uid: ea371004-6aec-4c6b-80ff-d55f829b7d03

$ kubectl get infrastructure -n shoot--core--u43kzeaqyg 
No resources found in shoot--core--u43kzeaqyg namespace.

What you expected to happen:

All the resources are cleaned up in the namespace.

How to reproduce it (as minimally and precisely as possible):

1. build the terraformer image with the master branch
2. replace the image vector info of terraformer in the controller deployment provider-alicloud
3. create a shoot and delete the shoot

Anything else we need to know?:

Environment: Alicloud dev landscape

• Gardener version (if relevant): lssd 0.2332.0
• Terraformer version: master branch
• Kubernetes version (use kubectl version): v1.19.15
• Cloud provider or hardware configuration: alicloud
• Others:

[gardener-robot]

@xiaofenhappy Label priority/normal does not exist.

[KevinDW-Fluxys]

I'm having the same issue, it seems to be caused by the finalizers. When i remove the finalizers the resources get deleted. If the pod is adding these finalizers, shouldnt it also remove them? This behavior also makes it impossible to update the configmap by deleting and recreating them.