update machines scale test

this change adds an extra retry period for marking machines to scale
down. this change is being added to help improve API interactions when
there might be network disruptions within the cluster, or in cases where
the record has been updated by another controller.

Author: elmiko

test/extended/machines/scale.go

@@ -287,8 +287,9 @@ var _ = g.Describe("[sig-cluster-lifecycle][Feature:Machines][Serial] Managed cl
 		for _, machineSet := range machineSets {
 			newestMachine, err := getNewestMachineNameFromMachineSet(dc, machineName(machineSet))
 			o.Expect(err).NotTo(o.HaveOccurred())
-			err = markMachineForScaleDown(dc, newestMachine)
-			o.Expect(err).NotTo(o.HaveOccurred())
+			o.Eventually(func() error {
+				return markMachineForScaleDown(dc, newestMachine)
+			}, 3*time.Minute, 5*time.Second).Should(o.Succeed(), "Unable to mark Machine %q for scale down", newestMachine)
 		}
 
 		g.By("scaling the machinesets back to their original size")