-
Feature Request
-
Resolution: Unresolved
-
Undefined
-
None
-
openshift-4.16
-
None
-
None
-
Product / Portfolio Work
-
None
-
False
-
-
None
-
None
-
None
-
-
None
-
None
-
None
-
None
-
None
1. Proposed title of this feature request
Warn if there are orphan objects that the garbage collector cannot delete over time.
2. What is the nature and description of the request?
The customer has an OpenShift Virtualization installation problem that went unnoticed for more than a year. The Kubevirt object is not owned by the HCO (ownerReference of the Kubevirt has a non-existent HCO), which made the garbage collector try to delete the Kubevirt numerous times over that year. It was blocked due to a Kubevirt webhook that rejects the deletion if there are VMs running.
For all that time, the Garbage collector was trying to delete the Kubevirt unsucessfully, and the issue was flying under the radar due to lack of proper alerts.
The customer wants us to implement some logic that warns them if there are any objects in the cluster that have been orphaned but for some reason are failing to delete, not just this specific Kubevirt problem, as this can happen in many other components and also user workloads.
This can help catch issues such as the one they had much earlier. Even some insights rule could be fine perhaps.
3. Why does the customer need this? (List the business requirements here)
Detect problems in the cluster earlier. Detect failures to cleanup by the Garbage Collector.
4. List any affected packages or components.