1. Proposed title of this feature request

HyperShift NodePools to support update-aware scheduling

2. What is the nature and description of the request?

When the NodePool is rolling out an update, any Nodes running the outgoing configuration should receive an UpdateInProgress PreferNoSchedule taint.

3. Why does the customer need this? (List the business requirements here)

Some customer workloads do not handle drains and reschedules well. Reducing the number of reschedules during an update will make the update less disruptive for those workloads, which makes it easier for customers to apply NodePool updates to bring in new bugfixes and features.

For standalone clusters, the machine-config operator added this functionality in mco#2686, which landed in OCP 4.10 via WRKLDS-279 and this enhancement text (although I haven't been able to turn up official documentation for this behavior). This enhancement would bring that functionality to HyperShift's NodePools.

4. List any affected packages or components.

HyperShift.