Upgrade failure: ClusterVersion stuck due to invalid capabilities after SCOS 4.15 upgrade attempt #2224
Description
Describe the bug
Bug Report
Description
When attempting to upgrade my OKD cluster from SCOS 4.14 to SCOS 4.15, the upgrade process fails partially. After rolling back to 4.14, the ClusterVersion resource remains stuck in a "Progressing" state.
Any attempt to patch or modify clusterversion/version results in validation errors due to invalid values in status.capabilities.enabledCapabilities, such as "Build", "DeploymentConfig", etc., which are not recognized by the current OKD version.
Steps to Reproduce
- Upgrade OKD cluster using SCOS 4.15 image:
harbor.tjylqwater.com:5443/okd/scos-content:4.15.0-0.okd-scos-2023-11-09-101502 - Upgrade partially completes then fails.
- Roll back to SCOS 4.14 image:
harbor.tjylqwater.com:5443/okd/scos-content:4.14.0-0.okd-scos-2023-05-23-224540 - ClusterVersion remains stuck and cannot be patched or updated.
Expected Behavior
- ClusterVersion should update cleanly without invalid capabilities.
- Upgrade or rollback process should complete without blocking on validation errors.
Actual Behavior
- Patch operations on ClusterVersion fail with errors like:
The ClusterVersion “version” is invalid:
• status.capabilities.enabledCapabilities[0]: Unsupported value: “Build”: supported values: “openshift-samples”, “baremetal”, … - ClusterVersion stuck in Progressing state indefinitely.
Environment
- OKD version before upgrade: 4.14.0-0.okd-scos-2023-05-23-224540
- Upgrade target image: 4.15.0-0.okd-scos-2023-11-09-101502
- Bare metal platform
Additional Information
- Tried patching spec.desiredUpdate and spec.channel, but validation errors persist.
- Issue blocks any further upgrades or rollbacks.
Thank you for your help!
Version
No response
Reproducibility
No response
Log Bundle
No response