AWS: Update copy-paste of GCE PD code to latest version

We are (sadly) using a copy-and-paste of the GCE PD code for AWS EBS.
This code hasn't been updated in a while, and it seems that the GCE code
has some code to make volume mounting more robust that we should copy.

pkg/cloudprovider/providers/aws/aws.go

@@ -158,11 +158,13 @@ type Volumes interface {
 	AttachDisk(instanceName string, volumeName string, readOnly bool) (string, error)
 	// Detach the disk from the specified instance
 	// instanceName can be empty to mean "the instance on which we are running"
-	DetachDisk(instanceName string, volumeName string) error
+	// Returns the device where the volume was attached
+	DetachDisk(instanceName string, volumeName string) (string, error)
 
 	// Create a volume with the specified options
-	CreateVolume(volumeOptions *VolumeOptions) (volumeName string, err error)
+	CreateDisk(volumeOptions *VolumeOptions) (volumeName string, err error)
-	DeleteVolume(volumeName string) error
+	// Delete the specified volume
+	DeleteDisk(volumeName string) error
 
 	// Get labels to apply to volume on creation
 	GetVolumeLabels(volumeName string) (map[string]string, error)
@@ -201,6 +203,8 @@ type AWSCloud struct {
 	mutex sync.Mutex
 }
 
+var _ Volumes = &AWSCloud{}
+
 type AWSCloudConfig struct {
 	Global struct {
 		// TODO: Is there any use for this?  We can get it from the instance metadata service
@@ -901,7 +905,7 @@ func (self *awsInstance) getInfo() (*ec2.Instance, error) {
 // Gets the mountDevice already assigned to the volume, or assigns an unused mountDevice.
 // If the volume is already assigned, this will return the existing mountDevice with alreadyAttached=true.
 // Otherwise the mountDevice is assigned by finding the first available mountDevice, and it is returned with alreadyAttached=false.
-func (self *awsInstance) getMountDevice(volumeID string) (assigned mountDevice, alreadyAttached bool, err error) {
+func (self *awsInstance) getMountDevice(volumeID string, assign bool) (assigned mountDevice, alreadyAttached bool, err error) {
 	instanceType := self.getInstanceType()
 	if instanceType == nil {
 		return "", false, fmt.Errorf("could not get instance type for instance: %s", self.awsID)
@@ -939,11 +943,17 @@ func (self *awsInstance) getMountDevice(volumeID string) (assigned mountDevice,
 	// Check to see if this volume is already assigned a device on this machine
 	for mountDevice, mappingVolumeID := range self.deviceMappings {
 		if volumeID == mappingVolumeID {
+			if assign {
 				glog.Warningf("Got assignment call for already-assigned volume: %s@%s", mountDevice, mappingVolumeID)
+			}
 			return mountDevice, true, nil
 		}
 	}
 
+	if !assign {
+		return mountDevice(""), false, nil
+	}
+
 	// Check all the valid mountpoints to see if any of them are free
 	valid := instanceType.getEBSMountDevices()
 	chosen := mountDevice("")
@@ -1172,7 +1182,7 @@ func (c *AWSCloud) AttachDisk(instanceName string, diskName string, readOnly boo
 		return "", errors.New("AWS volumes cannot be mounted read-only")
 	}
 
-	mountDevice, alreadyAttached, err := awsInstance.getMountDevice(disk.awsID)
+	mountDevice, alreadyAttached, err := awsInstance.getMountDevice(disk.awsID, true)
 	if err != nil {
 		return "", err
 	}
@@ -1220,15 +1230,25 @@ func (c *AWSCloud) AttachDisk(instanceName string, diskName string, readOnly boo
 }
 
 // Implements Volumes.DetachDisk
-func (aws *AWSCloud) DetachDisk(instanceName string, diskName string) error {
+func (aws *AWSCloud) DetachDisk(instanceName string, diskName string) (string, error) {
 	disk, err := newAWSDisk(aws, diskName)
 	if err != nil {
-		return err
+		return "", err
 	}
 
 	awsInstance, err := aws.getAwsInstance(instanceName)
 	if err != nil {
-		return err
+		return "", err
+	}
+
+	mountDevice, alreadyAttached, err := awsInstance.getMountDevice(disk.awsID, false)
+	if err != nil {
+		return "", err
+	}
+
+	if !alreadyAttached {
+		glog.Warning("DetachDisk called on non-attached disk: ", diskName)
+		// TODO: Continue?  Tolerate non-attached error in DetachVolume?
 	}
 
 	request := ec2.DetachVolumeInput{
@@ -1238,12 +1258,16 @@ func (aws *AWSCloud) DetachDisk(instanceName string, diskName string) error {
 
 	response, err := aws.ec2.DetachVolume(&request)
 	if err != nil {
-		return fmt.Errorf("error detaching EBS volume: %v", err)
+		return "", fmt.Errorf("error detaching EBS volume: %v", err)
 	}
 	if response == nil {
-		return errors.New("no response from DetachVolume")
+		return "", errors.New("no response from DetachVolume")
 	}
 
+	// TODO: Fix this - just remove the cache?
+	// If we don't have a cache; we don't have to wait any more (the driver does it for us)
+	// Also, maybe we could get the locally connected drivers from the AWS metadata service?
+
 	// At this point we are waiting for the volume being detached. This
 	// releases the volume and invalidates the cache even when there is a timeout.
 	//
@@ -1253,6 +1277,7 @@ func (aws *AWSCloud) DetachDisk(instanceName string, diskName string) error {
 	// works though. An option would be to completely flush the cache upon timeouts.
 	//
 	defer func() {
+		// TODO: Not thread safe?
 		for mountDevice, existingVolumeID := range awsInstance.deviceMappings {
 			if existingVolumeID == disk.awsID {
 				awsInstance.releaseMountDevice(disk.awsID, mountDevice)
@@ -1263,14 +1288,15 @@ func (aws *AWSCloud) DetachDisk(instanceName string, diskName string) error {
 
 	err = disk.waitForAttachmentStatus("detached")
 	if err != nil {
-		return err
+		return "", err
 	}
 
-	return err
+	hostDevicePath := "/dev/xvd" + string(mountDevice)
+	return hostDevicePath, err
 }
 
 // Implements Volumes.CreateVolume
-func (s *AWSCloud) CreateVolume(volumeOptions *VolumeOptions) (string, error) {
+func (s *AWSCloud) CreateDisk(volumeOptions *VolumeOptions) (string, error) {
 	// TODO: Should we tag this with the cluster id (so it gets deleted when the cluster does?)
 
 	request := &ec2.CreateVolumeInput{}
@@ -1302,7 +1328,7 @@ func (s *AWSCloud) CreateVolume(volumeOptions *VolumeOptions) (string, error) {
 		tagRequest.Tags = tags
 		if _, err := s.createTags(tagRequest); err != nil {
 			// delete the volume and hope it succeeds
-			delerr := s.DeleteVolume(volumeName)
+			delerr := s.DeleteDisk(volumeName)
 			if delerr != nil {
 				// delete did not succeed, we have a stray volume!
 				return "", fmt.Errorf("error tagging volume %s, could not delete the volume: %v", volumeName, delerr)
@@ -1313,8 +1339,8 @@ func (s *AWSCloud) CreateVolume(volumeOptions *VolumeOptions) (string, error) {
 	return volumeName, nil
 }
 
-// Implements Volumes.DeleteVolume
+// Implements Volumes.DeleteDisk
-func (aws *AWSCloud) DeleteVolume(volumeName string) error {
+func (aws *AWSCloud) DeleteDisk(volumeName string) error {
 	awsDisk, err := newAWSDisk(aws, volumeName)
 	if err != nil {
 		return err

pkg/volume/aws_ebs/aws_ebs.go

@@ -27,7 +27,6 @@ import (
 	"github.com/golang/glog"
 	"k8s.io/kubernetes/pkg/api"
 	"k8s.io/kubernetes/pkg/api/resource"
-	awscloud "k8s.io/kubernetes/pkg/cloudprovider/providers/aws"
 	"k8s.io/kubernetes/pkg/types"
 	"k8s.io/kubernetes/pkg/util/exec"
 	"k8s.io/kubernetes/pkg/util/mount"
@@ -103,12 +102,12 @@ func (plugin *awsElasticBlockStorePlugin) newBuilderInternal(spec *volume.Spec,
 			podUID:    podUID,
 			volName:   spec.Name(),
 			volumeID:  volumeID,
+			partition: partition,
 			manager:   manager,
 			mounter:   mounter,
 			plugin:    plugin,
 		},
 		fsType:      fsType,
-		partition:   partition,
 		readOnly:    readOnly,
 		diskMounter: &mount.SafeFormatAndMount{plugin.host.GetMounter(), exec.New()}}, nil
 }
@@ -181,6 +180,8 @@ type awsElasticBlockStore struct {
 	podUID  types.UID
 	// Unique id of the PD, used to find the disk resource in the provider.
 	volumeID string
+	// Specifies the partition to mount
+	partition string
 	// Utility interface that provides API calls to the provider to attach/detach disks.
 	manager ebsManager
 	// Mounter interface that provides system calls to mount the global path to the pod local path.
@@ -196,22 +197,10 @@ func detachDiskLogError(ebs *awsElasticBlockStore) {
 	}
 }
 
-// getVolumeProvider returns the AWS Volumes interface
-func (ebs *awsElasticBlockStore) getVolumeProvider() (awscloud.Volumes, error) {
-	cloud := ebs.plugin.host.GetCloudProvider()
-	volumes, ok := cloud.(awscloud.Volumes)
-	if !ok {
-		return nil, fmt.Errorf("Cloud provider does not support volumes")
-	}
-	return volumes, nil
-}
-
 type awsElasticBlockStoreBuilder struct {
 	*awsElasticBlockStore
 	// Filesystem type, optional.
 	fsType string
-	// Specifies the partition to mount
-	partition string
 	// Specifies whether the disk will be attached as read-only.
 	readOnly bool
 	// diskMounter provides the interface that is used to mount the actual block device.
@@ -304,6 +293,7 @@ func makeGlobalPDPath(host volume.VolumeHost, volumeID string) string {
 	return path.Join(host.GetPluginDir(awsElasticBlockStorePluginName), "mounts", name)
 }
 
+// Reverses the mapping done in makeGlobalPDPath
 func getVolumeIDFromGlobalMount(host volume.VolumeHost, globalPath string) (string, error) {
 	basePath := path.Join(host.GetPluginDir(awsElasticBlockStorePluginName), "mounts")
 	rel, err := filepath.Rel(basePath, globalPath)

pkg/volume/aws_ebs/aws_ebs_test.go

@@ -68,11 +68,12 @@ func TestGetAccessModes(t *testing.T) {
 	if err != nil {
 		t.Errorf("Can't find the plugin by name")
 	}
+
 	if !contains(plug.GetAccessModes(), api.ReadWriteOnce) {
-		t.Errorf("Expected to find AccessMode:  %s", api.ReadWriteOnce)
+		t.Errorf("Expected to support AccessModeTypes:  %s", api.ReadWriteOnce)
 	}
-	if len(plug.GetAccessModes()) != 1 {
+	if contains(plug.GetAccessModes(), api.ReadOnlyMany) {
-		t.Errorf("Expected to find exactly one AccessMode")
+		t.Errorf("Expected not to support AccessModeTypes:  %s", api.ReadOnlyMany)
 	}
 }
 
@@ -85,7 +86,10 @@ func contains(modes []api.PersistentVolumeAccessMode, mode api.PersistentVolumeA
 	return false
 }
 
-type fakePDManager struct{}
+type fakePDManager struct {
+	attachCalled bool
+	detachCalled bool
+}
 
 // TODO(jonesdl) To fully test this, we could create a loopback device
 // and mount that instead.
@@ -95,6 +99,10 @@ func (fake *fakePDManager) AttachAndMountDisk(b *awsElasticBlockStoreBuilder, gl
 	if err != nil {
 		return err
 	}
+	fake.attachCalled = true
+	// Simulate the global mount so that the fakeMounter returns the
+	// expected number of mounts for the attached disk.
+	b.mounter.Mount(globalPath, globalPath, b.fsType, nil)
 	return nil
 }
 
@@ -104,6 +112,7 @@ func (fake *fakePDManager) DetachDisk(c *awsElasticBlockStoreCleaner) error {
 	if err != nil {
 		return err
 	}
+	fake.detachCalled = true
 	return nil
 }
 
@@ -121,7 +130,7 @@ func (fake *fakePDManager) DeleteVolume(cd *awsElasticBlockStoreDeleter) error {
 func TestPlugin(t *testing.T) {
 	tmpDir, err := utiltesting.MkTmpdir("awsebsTest")
 	if err != nil {
-		t.Fatalf("can't make a temp dir: %v")
+		t.Fatalf("can't make a temp dir: %v", err)
 	}
 	defer os.RemoveAll(tmpDir)
 	plugMgr := volume.VolumePluginMgr{}
@@ -140,13 +149,16 @@ func TestPlugin(t *testing.T) {
 			},
 		},
 	}
-	builder, err := plug.(*awsElasticBlockStorePlugin).newBuilderInternal(volume.NewSpecFromVolume(spec), types.UID("poduid"), &fakePDManager{}, &mount.FakeMounter{})
+	fakeManager := &fakePDManager{}
+	fakeMounter := &mount.FakeMounter{}
+	builder, err := plug.(*awsElasticBlockStorePlugin).newBuilderInternal(volume.NewSpecFromVolume(spec), types.UID("poduid"), fakeManager, fakeMounter)
 	if err != nil {
 		t.Errorf("Failed to make a new Builder: %v", err)
 	}
 	if builder == nil {
 		t.Errorf("Got a nil Builder")
 	}
+
 	volPath := path.Join(tmpDir, "pods/poduid/volumes/kubernetes.io~aws-ebs/vol1")
 	path := builder.GetPath()
 	if path != volPath {
@@ -170,8 +182,12 @@ func TestPlugin(t *testing.T) {
 			t.Errorf("SetUp() failed: %v", err)
 		}
 	}
+	if !fakeManager.attachCalled {
+		t.Errorf("Attach watch not called")
+	}
 
-	cleaner, err := plug.(*awsElasticBlockStorePlugin).newCleanerInternal("vol1", types.UID("poduid"), &fakePDManager{}, &mount.FakeMounter{})
+	fakeManager = &fakePDManager{}
+	cleaner, err := plug.(*awsElasticBlockStorePlugin).newCleanerInternal("vol1", types.UID("poduid"), fakeManager, fakeMounter)
 	if err != nil {
 		t.Errorf("Failed to make a new Cleaner: %v", err)
 	}
@@ -187,9 +203,12 @@ func TestPlugin(t *testing.T) {
 	} else if !os.IsNotExist(err) {
 		t.Errorf("SetUp() failed: %v", err)
 	}
+	if !fakeManager.detachCalled {
+		t.Errorf("Detach watch not called")
+	}
 
 	// Test Provisioner
-	cap := resource.MustParse("100Gi")
+	cap := resource.MustParse("100Mi")
 	options := volume.VolumeOptions{
 		Capacity: cap,
 		AccessModes: []api.PersistentVolumeAccessMode{

pkg/volume/aws_ebs/aws_util.go

@@ -17,47 +17,58 @@ limitations under the License.
 package aws_ebs
 
 import (
-	"errors"
+	"fmt"
 	"os"
+	"path/filepath"
+	"strings"
 	"time"
 
 	"github.com/golang/glog"
-	aws_cloud "k8s.io/kubernetes/pkg/cloudprovider/providers/aws"
+	"k8s.io/kubernetes/pkg/cloudprovider"
+	"k8s.io/kubernetes/pkg/cloudprovider/providers/aws"
+	"k8s.io/kubernetes/pkg/util"
+	"k8s.io/kubernetes/pkg/util/exec"
+	"k8s.io/kubernetes/pkg/util/keymutex"
+	"k8s.io/kubernetes/pkg/util/sets"
 	"k8s.io/kubernetes/pkg/volume"
 )
 
+const (
+	diskPartitionSuffix = ""
+	diskXVDPath         = "/dev/xvd"
+	diskXVDPattern      = "/dev/xvd*"
+	maxChecks           = 60
+	maxRetries          = 10
+	checkSleepDuration  = time.Second
+	errorSleepDuration  = 5 * time.Second
+)
+
+// Singleton key mutex for keeping attach/detach operations for the same PD atomic
+var attachDetachMutex = keymutex.NewKeyMutex()
+
 type AWSDiskUtil struct{}
 
-// Attaches a disk specified by a volume.AWSElasticBlockStore to the current kubelet.
+// Attaches a disk to the current kubelet.
 // Mounts the disk to it's global path.
-func (util *AWSDiskUtil) AttachAndMountDisk(b *awsElasticBlockStoreBuilder, globalPDPath string) error {
+func (diskUtil *AWSDiskUtil) AttachAndMountDisk(b *awsElasticBlockStoreBuilder, globalPDPath string) error {
-	volumes, err := b.getVolumeProvider()
+	glog.V(5).Infof("AttachAndMountDisk(...) called for PD %q. Will block for existing operations, if any. (globalPDPath=%q)\r\n", b.volumeID, globalPDPath)
+
+	// Block execution until any pending detach operations for this PD have completed
+	attachDetachMutex.LockKey(b.volumeID)
+	defer attachDetachMutex.UnlockKey(b.volumeID)
+
+	glog.V(5).Infof("AttachAndMountDisk(...) called for PD %q. Awake and ready to execute. (globalPDPath=%q)\r\n", b.volumeID, globalPDPath)
+
+	xvdBefore, err := filepath.Glob(diskXVDPattern)
+	if err != nil {
+		glog.Errorf("Error filepath.Glob(\"%s\"): %v\r\n", diskXVDPattern, err)
+	}
+	xvdBeforeSet := sets.NewString(xvdBefore...)
+
+	devicePath, err := attachDiskAndVerify(b, xvdBeforeSet)
 	if err != nil {
 		return err
 	}
-	devicePath, err := volumes.AttachDisk("", b.volumeID, b.readOnly)
-	if err != nil {
-		return err
-	}
-	if b.partition != "" {
-		devicePath = devicePath + b.partition
-	}
-	//TODO(jonesdl) There should probably be better method than busy-waiting here.
-	numTries := 0
-	for {
-		_, err := os.Stat(devicePath)
-		if err == nil {
-			break
-		}
-		if err != nil && !os.IsNotExist(err) {
-			return err
-		}
-		numTries++
-		if numTries == 10 {
-			return errors.New("Could not attach disk: Timeout after 10s (" + devicePath + ")")
-		}
-		time.Sleep(time.Second)
-	}
 
 	// Only mount the PD globally once.
 	notMnt, err := b.mounter.IsLikelyNotMountPoint(globalPDPath)
@@ -87,64 +98,292 @@ func (util *AWSDiskUtil) AttachAndMountDisk(b *awsElasticBlockStoreBuilder, glob
 
 // Unmounts the device and detaches the disk from the kubelet's host machine.
 func (util *AWSDiskUtil) DetachDisk(c *awsElasticBlockStoreCleaner) error {
-	// Unmount the global PD mount, which should be the only one.
+	glog.V(5).Infof("DetachDisk(...) for PD %q\r\n", c.volumeID)
-	globalPDPath := makeGlobalPDPath(c.plugin.host, c.volumeID)
+
-	if err := c.mounter.Unmount(globalPDPath); err != nil {
+	if err := unmountPDAndRemoveGlobalPath(c); err != nil {
-		glog.V(2).Info("Error unmount dir ", globalPDPath, ": ", err)
+		glog.Errorf("Error unmounting PD %q: %v", c.volumeID, err)
-		return err
-	}
-	if err := os.Remove(globalPDPath); err != nil {
-		glog.V(2).Info("Error removing dir ", globalPDPath, ": ", err)
-		return err
-	}
-	// Detach the disk
-	volumes, err := c.getVolumeProvider()
-	if err != nil {
-		glog.V(2).Info("Error getting volume provider for volumeID ", c.volumeID, ": ", err)
-		return err
-	}
-	if err := volumes.DetachDisk("", c.volumeID); err != nil {
-		glog.V(2).Info("Error detaching disk ", c.volumeID, ": ", err)
-		return err
 	}
+
+	// Detach disk asynchronously so that the kubelet sync loop is not blocked.
+	go detachDiskAndVerify(c)
 	return nil
 }
 
 func (util *AWSDiskUtil) DeleteVolume(d *awsElasticBlockStoreDeleter) error {
-	volumes, err := d.getVolumeProvider()
+	cloud, err := getCloudProvider()
 	if err != nil {
-		glog.V(2).Info("Error getting volume provider: ", err)
 		return err
 	}
 
-	if err := volumes.DeleteVolume(d.volumeID); err != nil {
+	if err = cloud.DeleteDisk(d.volumeID); err != nil {
-		glog.V(2).Infof("Error deleting AWS EBS volume %s: %v", d.volumeID, err)
+		glog.V(2).Infof("Error deleting EBS Disk volume %s: %v", d.volumeID, err)
 		return err
 	}
-	glog.V(2).Infof("Successfully deleted AWS EBS volume %s", d.volumeID)
+	glog.V(2).Infof("Successfully deleted EBS Disk volume %s", d.volumeID)
 	return nil
 }
 
 func (util *AWSDiskUtil) CreateVolume(c *awsElasticBlockStoreProvisioner) (volumeID string, volumeSizeGB int, err error) {
-	volumes, err := c.getVolumeProvider()
+	cloud, err := getCloudProvider()
 	if err != nil {
-		glog.V(2).Info("Error getting volume provider: ", err)
 		return "", 0, err
 	}
 
 	requestBytes := c.options.Capacity.Value()
-	// AWS works with gigabytes, convert to GiB with rounding up
+	// The cloud provider works with gigabytes, convert to GiB with rounding up
-	requestGB := int(volume.RoundUpSize(requestBytes, 1024*1024*1024))
+	requestGB := volume.RoundUpSize(requestBytes, 1024*1024*1024)
-	volSpec := &aws_cloud.VolumeOptions{
-		CapacityGB: requestGB,
-		Tags:       c.options.CloudTags,
-	}
 
-	name, err := volumes.CreateVolume(volSpec)
+	volumeOptions := &aws.VolumeOptions{}
+	volumeOptions.CapacityGB = int(requestGB)
+
+	name, err := cloud.CreateDisk(volumeOptions)
 	if err != nil {
-		glog.V(2).Infof("Error creating AWS EBS volume: %v", err)
+		glog.V(2).Infof("Error creating EBS Disk volume: %v", err)
 		return "", 0, err
 	}
-	glog.V(2).Infof("Successfully created AWS EBS volume %s", name)
+	glog.V(2).Infof("Successfully created EBS Disk volume %s", name)
-	return name, requestGB, nil
+	return name, int(requestGB), nil
+}
+
+// Attaches the specified persistent disk device to node, verifies that it is attached, and retries if it fails.
+func attachDiskAndVerify(b *awsElasticBlockStoreBuilder, xvdBeforeSet sets.String) (string, error) {
+	var awsCloud *aws.AWSCloud
+	for numRetries := 0; numRetries < maxRetries; numRetries++ {
+		var err error
+		if awsCloud == nil {
+			awsCloud, err = getCloudProvider()
+			if err != nil || awsCloud == nil {
+				// Retry on error. See issue #11321
+				glog.Errorf("Error getting AWSCloudProvider while detaching PD %q: %v", b.volumeID, err)
+				time.Sleep(errorSleepDuration)
+				continue
+			}
+		}
+
+		if numRetries > 0 {
+			glog.Warningf("Retrying attach for EBS Disk %q (retry count=%v).", b.volumeID, numRetries)
+		}
+
+		devicePath, err := awsCloud.AttachDisk(b.volumeID, b.plugin.host.GetHostName(), b.readOnly)
+		if err != nil {
+			glog.Errorf("Error attaching PD %q: %v", b.volumeID, err)
+			time.Sleep(errorSleepDuration)
+			continue
+		}
+
+		devicePaths := getDiskByIdPaths(b.awsElasticBlockStore, devicePath)
+
+		for numChecks := 0; numChecks < maxChecks; numChecks++ {
+			path, err := verifyDevicePath(devicePaths, xvdBeforeSet)
+			if err != nil {
+				// Log error, if any, and continue checking periodically. See issue #11321
+				glog.Errorf("Error verifying EBS Disk (%q) is attached: %v", b.volumeID, err)
+			} else if path != "" {
+				// A device path has successfully been created for the PD
+				glog.Infof("Successfully attached EBS Disk %q.", b.volumeID)
+				return path, nil
+			}
+
+			// Sleep then check again
+			glog.V(3).Infof("Waiting for EBS Disk %q to attach.", b.volumeID)
+			time.Sleep(checkSleepDuration)
+		}
+	}
+
+	return "", fmt.Errorf("Could not attach EBS Disk %q. Timeout waiting for mount paths to be created.", b.volumeID)
+}
+
+// Returns the first path that exists, or empty string if none exist.
+func verifyDevicePath(devicePaths []string, xvdBeforeSet sets.String) (string, error) {
+	if err := udevadmChangeToNewDrives(xvdBeforeSet); err != nil {
+		// udevadm errors should not block disk detachment, log and continue
+		glog.Errorf("udevadmChangeToNewDrives failed with: %v", err)
+	}
+
+	for _, path := range devicePaths {
+		if pathExists, err := pathExists(path); err != nil {
+			return "", fmt.Errorf("Error checking if path exists: %v", err)
+		} else if pathExists {
+			return path, nil
+		}
+	}
+
+	return "", nil
+}
+
+// Detaches the specified persistent disk device from node, verifies that it is detached, and retries if it fails.
+// This function is intended to be called asynchronously as a go routine.
+func detachDiskAndVerify(c *awsElasticBlockStoreCleaner) {
+	glog.V(5).Infof("detachDiskAndVerify(...) for pd %q. Will block for pending operations", c.volumeID)
+	defer util.HandleCrash()
+
+	// Block execution until any pending attach/detach operations for this PD have completed
+	attachDetachMutex.LockKey(c.volumeID)
+	defer attachDetachMutex.UnlockKey(c.volumeID)
+
+	glog.V(5).Infof("detachDiskAndVerify(...) for pd %q. Awake and ready to execute.", c.volumeID)
+
+	var awsCloud *aws.AWSCloud
+	for numRetries := 0; numRetries < maxRetries; numRetries++ {
+		var err error
+		if awsCloud == nil {
+			awsCloud, err = getCloudProvider()
+			if err != nil || awsCloud == nil {
+				// Retry on error. See issue #11321
+				glog.Errorf("Error getting AWSCloudProvider while detaching PD %q: %v", c.volumeID, err)
+				time.Sleep(errorSleepDuration)
+				continue
+			}
+		}
+
+		if numRetries > 0 {
+			glog.Warningf("Retrying detach for EBS Disk %q (retry count=%v).", c.volumeID, numRetries)
+		}
+
+		devicePath, err := awsCloud.DetachDisk(c.volumeID, c.plugin.host.GetHostName())
+		if err != nil {
+			glog.Errorf("Error detaching PD %q: %v", c.volumeID, err)
+			time.Sleep(errorSleepDuration)
+			continue
+		}
+
+		devicePaths := getDiskByIdPaths(c.awsElasticBlockStore, devicePath)
+
+		for numChecks := 0; numChecks < maxChecks; numChecks++ {
+			allPathsRemoved, err := verifyAllPathsRemoved(devicePaths)
+			if err != nil {
+				// Log error, if any, and continue checking periodically.
+				glog.Errorf("Error verifying EBS Disk (%q) is detached: %v", c.volumeID, err)
+			} else if allPathsRemoved {
+				// All paths to the PD have been succefully removed
+				unmountPDAndRemoveGlobalPath(c)
+				glog.Infof("Successfully detached EBS Disk %q.", c.volumeID)
+				return
+			}
+
+			// Sleep then check again
+			glog.V(3).Infof("Waiting for EBS Disk %q to detach.", c.volumeID)
+			time.Sleep(checkSleepDuration)
+		}
+
+	}
+
+	glog.Errorf("Failed to detach EBS Disk %q. One or more mount paths was not removed.", c.volumeID)
+}
+
+// Unmount the global PD mount, which should be the only one, and delete it.
+func unmountPDAndRemoveGlobalPath(c *awsElasticBlockStoreCleaner) error {
+	globalPDPath := makeGlobalPDPath(c.plugin.host, c.volumeID)
+
+	err := c.mounter.Unmount(globalPDPath)
+	os.Remove(globalPDPath)
+	return err
+}
+
+// Returns the first path that exists, or empty string if none exist.
+func verifyAllPathsRemoved(devicePaths []string) (bool, error) {
+	allPathsRemoved := true
+	for _, path := range devicePaths {
+		if err := udevadmChangeToDrive(path); err != nil {
+			// udevadm errors should not block disk detachment, log and continue
+			glog.Errorf("%v", err)
+		}
+		if exists, err := pathExists(path); err != nil {
+			return false, fmt.Errorf("Error checking if path exists: %v", err)
+		} else {
+			allPathsRemoved = allPathsRemoved && !exists
+		}
+	}
+
+	return allPathsRemoved, nil
+}
+
+// Returns list of all paths for given EBS mount
+// This is more interesting on GCE (where we are able to identify volumes under /dev/disk-by-id)
+// Here it is mostly about applying the partition path
+func getDiskByIdPaths(d *awsElasticBlockStore, devicePath string) []string {
+	devicePaths := []string{}
+	if devicePath != "" {
+		devicePaths = append(devicePaths, devicePath)
+	}
+
+	if d.partition != "" {
+		for i, path := range devicePaths {
+			devicePaths[i] = path + diskPartitionSuffix + d.partition
+		}
+	}
+
+	return devicePaths
+}
+
+// Checks if the specified path exists
+func pathExists(path string) (bool, error) {
+	_, err := os.Stat(path)
+	if err == nil {
+		return true, nil
+	} else if os.IsNotExist(err) {
+		return false, nil
+	} else {
+		return false, err
+	}
+}
+
+// Return cloud provider
+func getCloudProvider() (*aws.AWSCloud, error) {
+	awsCloudProvider, err := cloudprovider.GetCloudProvider("aws", nil)
+	if err != nil || awsCloudProvider == nil {
+		return nil, err
+	}
+
+	// The conversion must be safe otherwise bug in GetCloudProvider()
+	return awsCloudProvider.(*aws.AWSCloud), nil
+}
+
+// TODO: This udev code is copy-and-paste from the gce_pd provider; refactor
+
+// Calls "udevadm trigger --action=change" for newly created "/dev/xvd*" drives (exist only in after set).
+// This is workaround for Issue #7972. Once the underlying issue has been resolved, this may be removed.
+func udevadmChangeToNewDrives(xvdBeforeSet sets.String) error {
+	xvdAfter, err := filepath.Glob(diskXVDPattern)
+	if err != nil {
+		return fmt.Errorf("Error filepath.Glob(\"%s\"): %v\r\n", diskXVDPattern, err)
+	}
+
+	for _, xvd := range xvdAfter {
+		if !xvdBeforeSet.Has(xvd) {
+			return udevadmChangeToDrive(xvd)
+		}
+	}
+
+	return nil
+}
+
+// Calls "udevadm trigger --action=change" on the specified drive.
+// drivePath must be the the block device path to trigger on, in the format "/dev/sd*", or a symlink to it.
+// This is workaround for Issue #7972. Once the underlying issue has been resolved, this may be removed.
+func udevadmChangeToDrive(drivePath string) error {
+	glog.V(5).Infof("udevadmChangeToDrive: drive=%q", drivePath)
+
+	// Evaluate symlink, if any
+	drive, err := filepath.EvalSymlinks(drivePath)
+	if err != nil {
+		return fmt.Errorf("udevadmChangeToDrive: filepath.EvalSymlinks(%q) failed with %v.", drivePath, err)
+	}
+	glog.V(5).Infof("udevadmChangeToDrive: symlink path is %q", drive)
+
+	// Check to make sure input is "/dev/xvd*"
+	if !strings.Contains(drive, diskXVDPath) {
+		return fmt.Errorf("udevadmChangeToDrive: expected input in the form \"%s\" but drive is %q.", diskXVDPattern, drive)
+	}
+
+	// Call "udevadm trigger --action=change --property-match=DEVNAME=/dev/sd..."
+	_, err = exec.New().Command(
+		"udevadm",
+		"trigger",
+		"--action=change",
+		fmt.Sprintf("--property-match=DEVNAME=%s", drive)).CombinedOutput()
+	if err != nil {
+		return fmt.Errorf("udevadmChangeToDrive: udevadm trigger failed for drive %q with %v.", drive, err)
+	}
+	return nil
 }

plugin/pkg/admission/persistentvolume/label/admission_test.go

@@ -37,11 +37,11 @@ func (v *mockVolumes) AttachDisk(instanceName string, volumeName string, readOnl
 	return "", fmt.Errorf("not implemented")
 }
 
-func (v *mockVolumes) DetachDisk(instanceName string, volumeName string) error {
+func (v *mockVolumes) DetachDisk(instanceName string, volumeName string) (string, error) {
-	return fmt.Errorf("not implemented")
+	return "", fmt.Errorf("not implemented")
 }
 
-func (v *mockVolumes) CreateVolume(volumeOptions *aws.VolumeOptions) (volumeName string, err error) {
+func (v *mockVolumes) CreateDisk(volumeOptions *aws.VolumeOptions) (volumeName string, err error) {
 	return "", fmt.Errorf("not implemented")
 }
 

test/e2e/pd.go

@@ -326,7 +326,7 @@ func createPD() (string, error) {
 		}
 		volumeOptions := &awscloud.VolumeOptions{}
 		volumeOptions.CapacityGB = 10
-		return volumes.CreateVolume(volumeOptions)
+		return volumes.CreateDisk(volumeOptions)
 	}
 }
 
@@ -353,7 +353,7 @@ func deletePD(pdName string) error {
 		if !ok {
 			return fmt.Errorf("Provider does not support volumes")
 		}
-		return volumes.DeleteVolume(pdName)
+		return volumes.DeleteDisk(pdName)
 	}
 }
 
@@ -383,7 +383,8 @@ func detachPD(hostName, pdName string) error {
 		if !ok {
 			return fmt.Errorf("Provider does not support volumes")
 		}
-		return volumes.DetachDisk(hostName, pdName)
+		_, err := volumes.DetachDisk(hostName, pdName)
+		return err
 	}
 }