Uncordon nodes that failed draining during update

[thegreenbear]

What would you like to be added:
When running the upgrade-cluster.yml playbook, I would like nodes that were cordoned but failed to be drained to be uncordoned.

Why is this needed:
During the upgrade of the cluster, Kubespray cordons and then drain nodes before actually updating them.
However, if a node fails to be drained (for example because of no poddisruption budget), Kubespray will leave it cordoned.
This means repeated failing runs (CI/CD) would cause more and more nodes to be left cordoned creating capacity issues on the running cluster.

The enhanced behavior I propose is the following: if the node fails to drain, no upgrade task was actually performed yet.
So it is safe to uncordon the node. However the upgrade should STILL fail as the upgrade did not work.

[floryut]

/close

(merged)

[k8s-ci-robot]

@floryut: Closing this issue.

In response to this:

/close

(merged)

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.