-
Bug
-
Resolution: Done-Errata
-
Critical
-
4.14.0
-
+
-
Important
-
Yes
-
Proposed
-
False
-
-
Release Note Not Required
-
In Progress
-
-
-
-
Description of problem:
After upgrade from 4.13.x to 4.14.10, the workload images that the customer stored inside the internal registry are lost, resulting the applications pods into error "Back-off pulling image". Even when manually pulling with podman, it fails then with "manifest unknown" because the image cannot be found in the registry anymore. - This behavior was found and reproduced 100% on ARO clusters, where the internal registry is by default backed up by the Storage Account created by the ARO RP service principal, which is the Containers blob service. - I do not know if in non-managed Azure clusters or any other architecture the same behavior is found.
Version-Release number of selected component (if applicable):
4.14.10
How reproducible:
100% with an ARO cluster (Managed cluster)
Steps to Reproduce: Attached.
The workaround found so far is to rebuild the apps or re-import the images. But those tasks are lengthy and costly specially if it is a production cluster.
- blocks
-
OCPBUGS-29525 Default Internal Registry cleans custom images stored on it from 4.13 to 4.14
- Closed
- is blocked by
-
IR-461 Impact assesment for OCPBUGS-29003: Default Internal Registry deletes custom images
- Closed
- is cloned by
-
OCPBUGS-29525 Default Internal Registry cleans custom images stored on it from 4.13 to 4.14
- Closed
- is related to
-
OCPBUGS-32328 Azure upgrades to 4.14.15+ fail with UPI storage account
- Closed
- links to
-
RHEA-2024:0041 OpenShift Container Platform 4.16.z bug fix update