An impact statement for the OCPBUGS-62073 bug.

Which 4.y.z to 4.y'.z' updates increase vulnerability?

• Clusters that were born before 4.12 (whose control plane nodes have the legacy master label, see OCPBUGS-58820), upgraded to any release between 4.15.0 to 4.15.56 and then subsequently attempted to upgrade to 4.19.12 or 4.19.13.

Which types of clusters?

• All platforms will be affected by this. 

What is the impact? Is it serious enough to warrant removing update recommendations?

• This will block the later stage of the upgrade to 4.19.12 or 4.19.13, while the CVO is rolling out the new MCO.

How involved is remediation?

• Admin can run a single oc command to resume the upgrade:

oc label node <control-plane-node-name> node-role.kubernetes.io/control-plane=

• The cronjob should schedule successfully after this and delete the old, problematic CRD (see also kcs#7131832). 

Is this a regression?

• Yes, it is a regression for only 4.19.12 and 4.19.13, as the MachineConfigNodes feature was GAed in 4.19.12. In 4.19.14, the CronJob was patched to use the legacy master label in its NodeSelector, so this failure should not be an issue going forward.