If we enable ImageMode in a MCP, and we remove the MOSC resource before the new osImage is fully applied, the MCP becomes degraded reporting this error:

  - lastTransitionTime: "2025-07-10T12:02:42Z"
    message: 'Node ip-10-0-11-240.us-east-2.compute.internal is reporting: "Node ip-10-0-11-240.us-east-2.compute.internal
      upgrade failure. failed to update OS to image-registry.openshift-image-registry.svc:5000/openshift-machine-config-operator/ocb-image@sha256:01971adebd6336f2da912840b92aac769a452a2830d34e420a58965cd77da5b8:
      error running rpm-ostree rebase --experimental ostree-unverified-registry:image-registry.openshift-image-registry.svc:5000/openshift-machine-config-operator/ocb-image@sha256:01971adebd6336f2da912840b92aac769a452a2830d34e420a58965cd77da5b8:
      error: Creating importer: failed to invoke method OpenImage: failed to invoke
      method OpenImage: reading manifest sha256:01971adebd6336f2da912840b92aac769a452a2830d34e420a58965cd77da5b8
      in image-registry.openshift-image-registry.svc:5000/openshift-machine-config-operator/ocb-image:
      manifest unknown\n: exit status 1", Node ip-10-0-11-240.us-east-2.compute.internal
      is reporting: "failed to update OS to image-registry.openshift-image-registry.svc:5000/openshift-machine-config-operator/ocb-image@sha256:01971adebd6336f2da912840b92aac769a452a2830d34e420a58965cd77da5b8:
      error running rpm-ostree rebase --experimental ostree-unverified-registry:image-registry.openshift-image-registry.svc:5000/openshift-machine-config-operator/ocb-image@sha256:01971adebd6336f2da912840b92aac769a452a2830d34e420a58965cd77da5b8:
      error: Creating importer: failed to invoke method OpenImage: failed to invoke
      method OpenImage: reading manifest sha256:01971adebd6336f2da912840b92aac769a452a2830d34e420a58965cd77da5b8
      in image-registry.openshift-image-registry.svc:5000/openshift-machine-config-operator/ocb-image:
      manifest unknown\n: exit status 1"'
    reason: 1 nodes are reporting degraded status on sync
    status: "True"
    type: NodeDegraded

4.20

Always

1. Create a MOSC resource to enable ImageMode in the worker pool
2. Wait until the MOSB finishes building the new osImage
3. Wait until the first node  starts updating
4. Immediately remove the MOSC resource

The MOSC resource will garbage collect the existing MOSBs and while doing that it will delete their osImages from the image registry.

Since the osImage is gabage collected, the MCP which is updating the image in the nodes cannot properly apply it and it becomes degraded.

I'm not sure about what should be the expected behaviour when we remove the MOSC resource while the images are being applied. Nevertheless, we should avoid a degradation.