Improve how do we determine the control plane machine to be remediated

[fabriziopandini]

#3830 introduces KCP remediation, and according to the proposal, in case of more than one unhealthy machine we are picking up the oldest one for remediation.

As commented in reconcileUnhealthyMachines, the current solution is considered acceptable for the most frequent use case (only one unhealthy machine), however, in the future, this could potentially be improved for the scenario where more than one unhealthy machine exists by considering which machine has a lower impact on etcd quorum.

This effort will provide better support for the following use case:

• two or more CP nodes are marked unhealthy
• at least another (healthy) machine has a failing etcd member

/kind feature

[fabriziopandini]

/milestone v0.4.0

[fabriziopandini]

/area control-plane
/priority backlog

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle stale

[vincepri]

/lifecycle frozen

[vincepri]

/milestone v1.0

[fabriziopandini]

/close
given that we don't have evidence of problems due to the current rule

[k8s-ci-robot]

@fabriziopandini: Closing this issue.

In response to this:

/close
given that we don't have evidence of problems due to the current rule

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.