When the registry.conf changes, a draining is initiated. The nodes go to SchedulingDisabled, but they don't go down for reboot/NotReady as found in MCO-406. This spike aims to investigate if pod disruption and draining are required. 

There are some special cases:
https://github.com/openshift/machine-config-operator/blob/98447c22feec3ded06cbb11f799343746fc248f4/pkg/daemon/drain.go#L109

A reboot exception for policy.json and an exception for mirror-by-digest-only=true registries was added at the end of last year, so ensure that the customer that requested this was experiencing rebooting or draining. 
https://github.com/openshift/machine-config-operator/pull/2824

https://github.com/openshift/machine-config-operator/pull/3144