Description of problem:

After a successful spoke cluster deployment, cluster deprovisioning fails clusterinstance delete hangs. Finalizer remaining is clusterdeployments.agent-install.openshift.io/ai-deprovision.

Version-Release number of selected component (if applicable):

MCE v2.10.0     ACM v2.15.0

How reproducible:

100%

Steps to Reproduce:

1. Deploy a hub, RDS compliant configuration, with these day-2 operator versions: 
   • Platform: OCP-4.20.6
   • ACM-2.15.0
   • MCE-2.10
   • GITOPS-1.18.2
   • TALM-4.20.0
2. Deploy successfully a spoke cluster (SNO with OCP-4.20.6) with ZTP/GITOPS approach.
3. Detach the spoke cluster,

Actual results:

Managed cluster namespace CR's:

• ClusterInstance
  • Annotations:

    clusterinstance.siteconfig.open-cluster-management.io/paused: ""

  • Finalizers: 

    - clusterinstance.siteconfig.open-cluster-management.io/finalizer

  • Status: paused

          reason: 'deletion timeout exceeded for object (BareMetalHost:sno/sno-master-0.fede.ipv6.lab):
            Timed out waiting to delete object (BareMetalHost:sno/sno-master-0.fede.ipv6.lab)'
          timeSet: "2025-12-16T08:46:44Z"

• ClusterDeployment
  • Annotations:

    hive.openshift.io/minor-version-upgrade-unavailable: ""

  • Finalizers: 

    - clusterdeployments.agent-install.openshift.io/ai-deprovision
    - hive.openshift.io/deprovision

• BMH
  • Annotations:

    baremetalhost.metal3.io/detached: assisted-service-controller

  • Finalizers: 

    - foregroundDeletion

Expected results:

The managed cluster is successfully detached from the hub cluster.

Additional info:

Issue discussed on this thread, cchun@redhat.com commented is not the same issue as ACM-26371 because "there's an assisted service finalizer as well".