-
Bug
-
Resolution: Done-Errata
-
Critical
-
4.13, 4.12, 4.14
-
Moderate
-
None
-
2
-
Sprint 235, Sprint 237, Sprint 236
-
3
-
Rejected
-
False
-
Description of problem:
If the HyperShift operator is installed onto a cluster, it creates VPC Endpoint Services fronting the hosted Kubernetes API Server for downstream HyperShift clusters to connect to. These VPC Endpoint Services are tagged such that the uninstaller would attempt to action them: "kubernetes.io/cluster/${ID}: owned" However they cannot be deleted until all active VPC Endpoint Connections are rejected - the uninstaller should be able to do this.
Version-Release number of selected component (if applicable):
4.12 (but shouldn't be version-specific)
How reproducible:
100%
Steps to Reproduce:
1. Create an NLB + VPC Endpoint Service in the same VPC as a cluster 2. Tag it accordingly and create a VPC Endpoint connection to it
Actual results:
The uninstaller will not be able to delete the VPC Endpoint Service + the NLB that the VPC Endpoint Service is fronting
Expected results:
The VPC Endpoint Service can be completely cleaned up, which would allow the NLB to be cleaned up
Additional info: