This change supports robust kubelet volume cleanup

Currently kubelet volume management works on the concept of desired
and actual world of states. The volume manager periodically compares the
two worlds and perform volume mount/unmount and/or attach/detach
operations. When kubelet restarts, the cache of those two worlds are
gone. Although desired world can be recovered through apiserver, actual
world can not be recovered which may cause some volumes cannot be cleaned
up if their information is deleted by apiserver. This change adds the
reconstruction of the actual world by reading the pod directories from
disk. The reconstructed volume information is added to both desired
world and actual world if it cannot be found in either world. The rest
logic would be as same as before, desired world populator may clean up
the volume entry if it is no longer in apiserver, and then volume
manager should invoke unmount to clean it up.

pkg/volume/plugins.go

@@ -98,6 +98,12 @@ type VolumePlugin interface {
 	// - name: The volume name, as per the api.Volume spec.
 	// - podUID: The UID of the enclosing pod
 	NewUnmounter(name string, podUID types.UID) (Unmounter, error)
+
+	// ConstructVolumeSpec constructs a volume spec based on the given volume name
+	// and mountPath. The spec may have incomplete information due to limited
+	// information from input. This function is used by volume manager to reconstruct
+	// volume spec by reading the volume directories from disk
+	ConstructVolumeSpec(volumeName, mountPath string) (*Spec, error)
 }
 
 // PersistentVolumePlugin is an extended interface of VolumePlugin and is used
@@ -151,6 +157,7 @@ type AttachableVolumePlugin interface {
 	VolumePlugin
 	NewAttacher() (Attacher, error)
 	NewDetacher() (Detacher, error)
+	GetDeviceMountRefs(deviceMountPath string) ([]string, error)
 }
 
 // VolumeHost is an interface that plugins can use to access the kubelet.