Fix taint removal retry for non-swallowed errors

pkg/driver/node.go

@@ -103,7 +103,7 @@ func newNodeService(driverOptions *DriverOptions) nodeService {
 
 	// Remove taint from node to indicate driver startup success
 	// This is done at the last possible moment to prevent race conditions or false positive removals
-	go removeTaintInBackground(cloud.DefaultKubernetesAPIClient)
+	go removeTaintInBackground(cloud.DefaultKubernetesAPIClient, removeNotReadyTaint)
 
 	return nodeService{
 		metadata:         metadata,
@@ -854,12 +854,12 @@ type JSONPatch struct {
 }
 
 // removeTaintInBackground is a goroutine that retries removeNotReadyTaint with exponential backoff
-func removeTaintInBackground(k8sClient cloud.KubernetesAPIClient) {
+func removeTaintInBackground(k8sClient cloud.KubernetesAPIClient, removalFunc func(cloud.KubernetesAPIClient) error) {
 	backoffErr := wait.ExponentialBackoff(taintRemovalBackoff, func() (bool, error) {
-		err := removeNotReadyTaint(k8sClient)
+		err := removalFunc(k8sClient)
 		if err != nil {
 			klog.ErrorS(err, "Unexpected failure when attempting to remove node taint(s)")
-			return false, err
+			return false, nil
 		}
 		return true, nil
 	})

pkg/driver/node_test.go

@@ -34,6 +34,7 @@ import (
 	"github.com/golang/mock/gomock"
 	"github.com/kubernetes-sigs/aws-ebs-csi-driver/pkg/cloud"
 	"github.com/kubernetes-sigs/aws-ebs-csi-driver/pkg/driver/internal"
+	"github.com/stretchr/testify/assert"
 	"google.golang.org/grpc/codes"
 	"google.golang.org/grpc/status"
 	corev1 "k8s.io/api/core/v1"
@@ -2405,6 +2406,21 @@ func TestRemoveNotReadyTaint(t *testing.T) {
 	}
 }
 
+func TestRemoveTaintInBackground(t *testing.T) {
+	mockRemovalCount := 0
+	mockRemovalFunc := func(_ cloud.KubernetesAPIClient) error {
+		mockRemovalCount += 1
+		if mockRemovalCount == 3 {
+			return nil
+		} else {
+			return fmt.Errorf("Taint removal failed!")
+		}
+	}
+
+	removeTaintInBackground(nil, mockRemovalFunc)
+	assert.Equal(t, mockRemovalCount, 3)
+}
+
 func getNodeMock(mockCtl *gomock.Controller, nodeName string, returnNode *corev1.Node, returnError error) (kubernetes.Interface, *MockNodeInterface) {
 	mockClient := NewMockKubernetesClient(mockCtl)
 	mockCoreV1 := NewMockCoreV1Interface(mockCtl)