Allow automatic recreation of AWS instances that are about to retire

1. Proposed title of this feature request

Allow automatic recreation of AWS instances that are about to retire

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

AWS from time to time retires instances, which means they will be terminated at a specified date. OCP customers often want to specify when this kind of maintenance happens as their workloads may not be designed to cope with the associated rescheduling.
Hence, they need to make sure the instance is recreated before the retirement date so they can control when it should happen.

On managed OpenShift (OSD/ROSA), customers in that case reach out to Red Hat support, requiring SREs to manually perform the rotation at the specified time window.
This is not scalable and hence we need a way to automate it.

See AWS docs on instance retirement.

An instance is scheduled to be retired when AWS detects irreparable failure of the underlying hardware that hosts the instance.

When an instance is scheduled for retirement, this is reflected in the instance status.

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

• Customers want to specify a time window when nodes can be replaced
• Managed OpenShift customers (OSD/ROSA) can't do the replacement themselves and need manual action by the SRE team.
• To automate the instance retirement handling for managed OpenShift as well as other OCP on AWS customers, a solution in OCP is necessary.

4. List any affected packages or components.

machine-api-operator
cluster-version-operator?