Option to not force drain a node from kubernetes

[gregsidelinger]

aws-asg-roller/kubernetes.go

Lines 72 to 77 in 0b31468

	err = drain.Drain(k.clientset, []*corev1.Node{node}, &drain.DrainOptions{
	IgnoreDaemonsets: k.ignoreDaemonSets,
	GracePeriodSeconds: -1,
	Force: true,
	DeleteLocalData: k.deleteLocalData,
	})

Would it be possible to create an option to not force delete standalone pods? We have a bunch of short term running pods that are not managed by a deployment or job that should be given a chance to finish running. Since there is no backing deployment I can not use a Pod Disruption Budget to instruct kube to not delete them so setting the force=false and trying again later is the only option I can think of.

[deitch]

Certainly. The hardest part about this usually is the user interface, with everything in env vars, but we don't have too many, so it shouldn't be too bad. Maybe ROLLER_PREVENT_DRAIN or similar?

Go for a PR.