-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.20
Description of problem:
Since OCP 4.20, openshift-controller-manager pods are getting affected by constant restarts due to the following events (1 restart per minute eventually):
$ oc logs -n <ns> openshift-controller-manager-xxx -p | tail -n 3 2026-01-22T14:04:36.637044417Z I0122 14:04:36.637003 1 observer_polling.go:120] Observed file "/etc/kubernetes/config/config.yaml" has been modified (old="xxx16a", new="xxxe615") 2026-01-22T14:04:36.637044417Z W0122 14:04:36.637038 1 builder.go:160] Restart triggered because of file /etc/kubernetes/config/config.yaml was modified 2026-01-22T14:04:36.637072934Z W0122 14:04:36.637057 1 leaderelection.go:84] leader election lost
This seems to happen because control plane operator is updating the openshift controller manager configmap every minute:
- apiVersion: v1
fieldsType: FieldsV1
fieldsV1:
f:data:
.: {}
f:config.yaml: {}
f:metadata:
f:ownerReferences:
.: {}
k:{"uid":"xxx30f"}: {}
manager: hypershift-controlplane-manager
operation: Update
time: "2026-01-23T08:47:53Z" <---
- apiVersion: v1
fieldsType: FieldsV1
fieldsV1:
f:data:
.: {}
f:config.yaml: {}
f:metadata:
f:ownerReferences:
.: {}
k:{"uid":"xxx30f"}: {}
manager: hypershift-controlplane-manager
operation: Update
time: "2026-01-23T09:22:52Z" <---
No content from OCM configuration is modified, only the configmap is updated and resourceVersion is updated.
There are not clear logs in the control plane operator on why it is being created or reconciled.
The restart trigger does not appear to be the cause, because the library-go/observer_polling.go identifies any changes in the configmap due to a hash difference, it decides to restart the pods which it is an expected behavior.
Version-Release number of selected component (if applicable):
4.20
How reproducible:
Steps to Reproduce:
1.
2.
3.
Actual results:
Expected results:
Additional info:
- blocks
-
-
- New
-
- is cloned by
-
-
- New
-
- links to