-
Bug
-
Resolution: Done
-
Normal
-
None
-
4.18.0
-
None
-
None
-
False
-
Description of problem:
When deploying a Nodepool on an OpenStack cloud which doesn't have OpenStack Swift (object storage), the Image Registry operator will fall back on using OpenStack Cinder (block storage) to store the Image registry content. When it happens, it currently fails on this error: E1025 18:54:03.596862 1 swift.go:73] error connecting to Swift: container endpoint not found in the OpenStack catalog: No suitable endpoint could be found in the service catalog. I1025 18:54:03.598634 1 controller.go:294] object changed: *v1.Config, Name=cluster (metadata=false, spec=true): added:spec.storage.pvc.claim="image-registry-storage", added:status.conditions.0.lastTransitionTime="2024-10-25T18:54:03Z", added:status.conditions.0.message="persistentvolumeclaims \"image-registry-storage\" not found", added:status.conditions.0.reason="PVC Issues Found", added:status.conditions.0.status="False", added:status.conditions.0.type="StorageExists", added:status.conditions.1.lastTransitionTime="2024-10-25T18:54:03Z", added:status.conditions.1.message="The deployment does not exist", added:status.conditions.1.reason="DeploymentNotFound", added:status.conditions.1.status="False", added:status.conditions.1.type="Available", added:status.conditions.2.lastTransitionTime="2024-10-25T18:54:03Z", added:status.conditions.2.message="Unable to apply resources: unable to sync storage configuration: persistentvolumeclaims \"image-registry-storage\" not found", added:status.conditions.2.reason="Error", added:status.conditions.2.status="True", added:status.conditions.2.type="Progressing", added:status.conditions.3.lastTransitionTime="2024-10-25T18:54:03Z", added:status.conditions.3.status="False", added:status.conditions.3.type="Degraded", added:status.conditions.4.lastTransitionTime="2024-10-25T18:54:03Z", added:status.conditions.4.status="False", added:status.conditions.4.type="Removed"
Version-Release number of selected component (if applicable):
4.18.0
How reproducible:
Deploy a HostedCluster on any platform and then Nodepools on OpenStack where Swift isn't available.
Actual results:
The Nodepool isn't healthy, the image registry operator is in error state
Expected results:
A PVC should be created and the image registry operator should be healthy.
Additional info:
This was observed in our CI where the management cluster runs on AWS and the Nodepools runs on an OpenStack cloud used for CI.
- relates to
-
OSASINFRA-3632 Configure cluster-storage-operator to support Hypershift for Cinder CSI
- In Progress