Add mutex to DeleteNodes in cluster-autoscaler CAPI provider

This change adds a mutex to the MachineController structure which is
used to gate access to the DeleteNodes function.

This is one in a series of PRs to mitigate kubernetes#3104

cluster-autoscaler/cloudprovider/clusterapi/clusterapi_controller.go

@@ -21,6 +21,7 @@ import (
 	"fmt"
 	"os"
 	"strings"
+	"sync"
 
 	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -65,6 +66,7 @@ type machineController struct {
 	machineSetResource        *schema.GroupVersionResource
 	machineResource           *schema.GroupVersionResource
 	machineDeploymentResource *schema.GroupVersionResource
+	accessLock                sync.Mutex
 }
 
 type machineSetFilterFunc func(machineSet *MachineSet) error

cluster-autoscaler/cloudprovider/clusterapi/clusterapi_nodegroup.go

@@ -82,6 +82,9 @@ func (ng *nodegroup) IncreaseSize(delta int) error {
 // group. This function should wait until node group size is updated.
 // Implementation required.
 func (ng *nodegroup) DeleteNodes(nodes []*corev1.Node) error {
+	ng.machineController.accessLock.Lock()
+	defer ng.machineController.accessLock.Unlock()
+
 	// Step 1: Verify all nodes belong to this node group.
 	for _, node := range nodes {
 		actualNodeGroup, err := ng.machineController.nodeGroupForNode(node)