View Source
const CloudVolumeCreatedForClaimNameTag = ""

CloudVolumeCreatedForClaimNameTag is a name of a tag attached to a real volume in cloud (e.g. AWS EBS or GCE PD) with name of a persistent volume claim used to create this volume.

View Source
const CloudVolumeCreatedForClaimNamespaceTag = ""

CloudVolumeCreatedForClaimNamespaceTag is a name of a tag attached to a real volume in cloud (e.g. AWS EBS or GCE PD) with namespace of a persistent volume claim used to create this volume.

View Source
const CloudVolumeCreatedForVolumeNameTag = ""

CloudVolumeCreatedForVolumeNameTag is a name of a tag attached to a real volume in cloud (e.g. AWS EBS or GCE PD) with name of appropriate Kubernetes persistent volume .

View Source
const VolumeSchedulerSubsystem = "scheduler_volume"

VolumeSchedulerSubsystem - subsystem name used by scheduler


View Source
var (
	VolumeBindingRequestSchedulerBinderCache = prometheus.NewCounterVec(
			Subsystem: VolumeSchedulerSubsystem,
			Name:      "binder_cache_requests_total",
			Help:      "Total number for request volume binding cache",
	VolumeSchedulingStageLatency = prometheus.NewHistogramVec(
			Subsystem: VolumeSchedulerSubsystem,
			Name:      "scheduling_duration_seconds",
			Help:      "Volume scheduling stage latency",
			Buckets:   prometheus.ExponentialBuckets(1000, 2, 15),
	VolumeSchedulingStageFailed = prometheus.NewCounterVec(
			Subsystem: VolumeSchedulerSubsystem,
			Name:      "scheduling_stage_error_total",
			Help:      "Volume scheduling stage error count",


func GetBindVolumeToClaim

func GetBindVolumeToClaim(volume *v1.PersistentVolume, claim *v1.PersistentVolumeClaim) (*v1.PersistentVolume, bool, error)

GetBindVolumeToClaim returns a new volume which is bound to given claim. In addition, it returns a bool which indicates whether we made modification on original volume.

func IsDelayBindingMode

func IsDelayBindingMode(claim *v1.PersistentVolumeClaim, classLister storagelisters.StorageClassLister) (bool, error)

IsDelayBindingMode checks if claim is in delay binding mode.

func IsVolumeBoundToClaim

func IsVolumeBoundToClaim(volume *v1.PersistentVolume, claim *v1.PersistentVolumeClaim) bool

IsVolumeBoundToClaim returns true, if given volume is pre-bound or bound to specific claim. Both claim.Name and claim.Namespace must be equal. If claim.UID is present in volume.Spec.ClaimRef, it must be equal too.

func NewAssumeCache

func NewAssumeCache(informer cache.SharedIndexInformer, description, indexName string, indexFunc cache.IndexFunc) *assumeCache

func RegisterVolumeSchedulingMetrics

func RegisterVolumeSchedulingMetrics()

RegisterVolumeSchedulingMetrics is used for scheduler, because the volume binding cache is a library used by scheduler process.


type AssumeCache

type AssumeCache interface {
	// Assume updates the object in-memory only
	Assume(obj interface{}) error

	// Restore the informer cache's version of the object
	Restore(objName string)

	// Get the object by name
	Get(objName string) (interface{}, error)

	// Get the API object by name
	GetAPIObj(objName string) (interface{}, error)

	// List all the objects in the cache
	List(indexObj interface{}) []interface{}

AssumeCache is a cache on top of the informer that allows for updating objects outside of informer events and also restoring the informer cache's version of the object. Objects are assumed to be Kubernetes API objects that implement meta.Interface

type ControllerParameters

type ControllerParameters struct {
	KubeClient                clientset.Interface
	SyncPeriod                time.Duration
	VolumePlugins             []vol.VolumePlugin
	Cloud                     cloudprovider.Interface
	ClusterName               string
	VolumeInformer            coreinformers.PersistentVolumeInformer
	ClaimInformer             coreinformers.PersistentVolumeClaimInformer
	ClassInformer             storageinformers.StorageClassInformer
	PodInformer               coreinformers.PodInformer
	NodeInformer              coreinformers.NodeInformer
	EventRecorder             record.EventRecorder
	EnableDynamicProvisioning bool

ControllerParameters contains arguments for creation of a new PersistentVolume controller.

type FakeVolumeBinder

type FakeVolumeBinder struct {
	AssumeCalled bool
	BindCalled   bool
	// contains filtered or unexported fields

func NewFakeVolumeBinder

func NewFakeVolumeBinder(config *FakeVolumeBinderConfig) *FakeVolumeBinder

NewVolumeBinder sets up all the caches needed for the scheduler to make topology-aware volume binding decisions.

func (*FakeVolumeBinder) AssumePodVolumes

func (b *FakeVolumeBinder) AssumePodVolumes(assumedPod *v1.Pod, nodeName string) (bool, error)

func (*FakeVolumeBinder) BindPodVolumes

func (b *FakeVolumeBinder) BindPodVolumes(assumedPod *v1.Pod) error

func (*FakeVolumeBinder) FindPodVolumes

func (b *FakeVolumeBinder) FindPodVolumes(pod *v1.Pod, node *v1.Node) (unboundVolumesSatisfied, boundVolumesSatsified bool, err error)

func (*FakeVolumeBinder) GetBindingsCache

func (b *FakeVolumeBinder) GetBindingsCache() PodBindingCache

type FakeVolumeBinderConfig

type FakeVolumeBinderConfig struct {
	AllBound             bool
	FindUnboundSatsified bool
	FindBoundSatsified   bool
	FindErr              error
	AssumeErr            error
	BindErr              error

type PVAssumeCache

type PVAssumeCache interface {

	GetPV(pvName string) (*v1.PersistentVolume, error)
	GetAPIPV(pvName string) (*v1.PersistentVolume, error)
	ListPVs(storageClassName string) []*v1.PersistentVolume

PVAssumeCache is a AssumeCache for PersistentVolume objects

func NewPVAssumeCache

func NewPVAssumeCache(informer cache.SharedIndexInformer) PVAssumeCache

type PVCAssumeCache

type PVCAssumeCache interface {

	// GetPVC returns the PVC from the cache with given pvcKey.
	// pvcKey is the result of MetaNamespaceKeyFunc on PVC obj
	GetPVC(pvcKey string) (*v1.PersistentVolumeClaim, error)
	GetAPIPVC(pvcKey string) (*v1.PersistentVolumeClaim, error)

PVCAssumeCache is a AssumeCache for PersistentVolumeClaim objects

func NewPVCAssumeCache

func NewPVCAssumeCache(informer cache.SharedIndexInformer) PVCAssumeCache

type PersistentVolumeController

type PersistentVolumeController struct {
	NodeLister       corelisters.NodeLister
	NodeListerSynced cache.InformerSynced
	// contains filtered or unexported fields

PersistentVolumeController is a controller that synchronizes PersistentVolumeClaims and PersistentVolumes. It starts two cache.Controllers that watch PersistentVolume and PersistentVolumeClaim changes.

func NewController

NewController creates a new PersistentVolume controller

func (*PersistentVolumeController) DeleteServiceAccountTokenFunc

func (ctrl *PersistentVolumeController) DeleteServiceAccountTokenFunc() func(types.UID)

func (*PersistentVolumeController) GetCloudProvider

func (ctrl *PersistentVolumeController) GetCloudProvider() cloudprovider.Interface

func (*PersistentVolumeController) GetConfigMapFunc

func (ctrl *PersistentVolumeController) GetConfigMapFunc() func(namespace, name string) (*v1.ConfigMap, error)

func (*PersistentVolumeController) GetEventRecorder

func (ctrl *PersistentVolumeController) GetEventRecorder() record.EventRecorder

func (*PersistentVolumeController) GetExec

func (adc *PersistentVolumeController) GetExec(pluginName string) mount.Exec

func (*PersistentVolumeController) GetHostIP

func (ctrl *PersistentVolumeController) GetHostIP() (net.IP, error)

func (*PersistentVolumeController) GetHostName

func (ctrl *PersistentVolumeController) GetHostName() string

func (*PersistentVolumeController) GetKubeClient

func (ctrl *PersistentVolumeController) GetKubeClient() clientset.Interface

func (*PersistentVolumeController) GetMounter

func (ctrl *PersistentVolumeController) GetMounter(pluginName string) mount.Interface

func (*PersistentVolumeController) GetNodeAllocatable

func (ctrl *PersistentVolumeController) GetNodeAllocatable() (v1.ResourceList, error)

func (*PersistentVolumeController) GetNodeLabels

func (ctrl *PersistentVolumeController) GetNodeLabels() (map[string]string, error)

func (*PersistentVolumeController) GetNodeName

func (ctrl *PersistentVolumeController) GetNodeName() types.NodeName

func (*PersistentVolumeController) GetPluginDir

func (ctrl *PersistentVolumeController) GetPluginDir(pluginName string) string

func (*PersistentVolumeController) GetPodPluginDir

func (ctrl *PersistentVolumeController) GetPodPluginDir(podUID types.UID, pluginName string) string

func (*PersistentVolumeController) GetPodVolumeDeviceDir

func (ctrl *PersistentVolumeController) GetPodVolumeDeviceDir(ppodUID types.UID, pluginName string) string

func (*PersistentVolumeController) GetPodVolumeDir

func (ctrl *PersistentVolumeController) GetPodVolumeDir(podUID types.UID, pluginName string, volumeName string) string

func (*PersistentVolumeController) GetPodsDir

func (ctrl *PersistentVolumeController) GetPodsDir() string

func (*PersistentVolumeController) GetSecretFunc

func (ctrl *PersistentVolumeController) GetSecretFunc() func(namespace, name string) (*v1.Secret, error)

func (*PersistentVolumeController) GetServiceAccountTokenFunc

func (ctrl *PersistentVolumeController) GetServiceAccountTokenFunc() func(_, _ string, _ *authenticationv1.TokenRequest) (*authenticationv1.TokenRequest, error)

func (*PersistentVolumeController) GetSubpather

func (ctrl *PersistentVolumeController) GetSubpather() subpath.Interface

func (*PersistentVolumeController) GetVolumeDevicePluginDir

func (ctrl *PersistentVolumeController) GetVolumeDevicePluginDir(pluginName string) string

func (*PersistentVolumeController) NewWrapperMounter

func (ctrl *PersistentVolumeController) NewWrapperMounter(volName string, spec vol.Spec, pod *v1.Pod, opts vol.VolumeOptions) (vol.Mounter, error)

func (*PersistentVolumeController) NewWrapperUnmounter

func (ctrl *PersistentVolumeController) NewWrapperUnmounter(volName string, spec vol.Spec, podUID types.UID) (vol.Unmounter, error)

func (*PersistentVolumeController) Run

func (ctrl *PersistentVolumeController) Run(stopCh <-chan struct{})

Run starts all of this controller's control loops

type PodBindingCache

type PodBindingCache interface {
	// UpdateBindings will update the cache with the given bindings for the
	// pod and node.
	UpdateBindings(pod *v1.Pod, node string, bindings []*bindingInfo, provisionings []*v1.PersistentVolumeClaim)

	// ClearBindings will clear the cached bindings for the given pod and node.
	ClearBindings(pod *v1.Pod, node string)

	// GetBindings will return the cached bindings for the given pod and node.
	// A nil return value means that the entry was not found. An empty slice
	// means that no binding operations are needed.
	GetBindings(pod *v1.Pod, node string) []*bindingInfo

	// A nil return value means that the entry was not found. An empty slice
	// means that no provisioning operations are needed.
	GetProvisionedPVCs(pod *v1.Pod, node string) []*v1.PersistentVolumeClaim

	// GetDecisions will return all cached decisions for the given pod.
	GetDecisions(pod *v1.Pod) nodeDecisions

	// DeleteBindings will remove all cached bindings and provisionings for the given pod.
	// TODO: separate the func if it is needed to delete bindings/provisionings individually
	DeleteBindings(pod *v1.Pod)

podBindingCache stores PV binding decisions per pod per node. Pod entries are removed when the Pod is deleted or updated to no longer be schedulable.

func NewPodBindingCache

func NewPodBindingCache() PodBindingCache

type SchedulerVolumeBinder

type SchedulerVolumeBinder interface {
	// FindPodVolumes checks if all of a Pod's PVCs can be satisfied by the node.
	// If a PVC is bound, it checks if the PV's NodeAffinity matches the Node.
	// Otherwise, it tries to find an available PV to bind to the PVC.
	// It returns true if all of the Pod's PVCs have matching PVs or can be dynamic provisioned,
	// and returns true if bound volumes satisfy the PV NodeAffinity.
	// This function is called by the volume binding scheduler predicate and can be called in parallel
	FindPodVolumes(pod *v1.Pod, node *v1.Node) (unboundVolumesSatisified, boundVolumesSatisfied bool, err error)

	// AssumePodVolumes will:
	// 1. Take the PV matches for unbound PVCs and update the PV cache assuming
	// that the PV is prebound to the PVC.
	// 2. Take the PVCs that need provisioning and update the PVC cache with related
	// annotations set.
	// It returns true if all volumes are fully bound
	// This function will modify assumedPod with the node name.
	// This function is called serially.
	AssumePodVolumes(assumedPod *v1.Pod, nodeName string) (allFullyBound bool, err error)

	// BindPodVolumes will:
	// 1. Initiate the volume binding by making the API call to prebind the PV
	// to its matching PVC.
	// 2. Trigger the volume provisioning by making the API call to set related
	// annotations on the PVC
	// 3. Wait for PVCs to be completely bound by the PV controller
	// This function can be called in parallel.
	BindPodVolumes(assumedPod *v1.Pod) error

	// GetBindingsCache returns the cache used (if any) to store volume binding decisions.
	GetBindingsCache() PodBindingCache

SchedulerVolumeBinder is used by the scheduler to handle PVC/PV binding and dynamic provisioning. The binding decisions are integrated into the pod scheduling workflow so that the PV NodeAffinity is also considered along with the pod's other scheduling requirements.

This integrates into the existing default scheduler workflow as follows: 1. The scheduler takes a Pod off the scheduler queue and processes it serially:

a. Invokes all predicate functions, parallelized across nodes.  FindPodVolumes() is invoked here.
b. Invokes all priority functions.  Future/TBD
c. Selects the best node for the Pod.
d. Cache the node selection for the Pod. AssumePodVolumes() is invoked here.
   i.  If PVC binding is required, cache in-memory only:
       * For manual binding: update PV objects for prebinding to the corresponding PVCs.
       * For dynamic provisioning: update PVC object with a selected node from c)
       * For the pod, which PVCs and PVs need API updates.
   ii. Afterwards, the main scheduler caches the Pod->Node binding in the scheduler's pod cache,
       This is handled in the scheduler and not here.
e. Asynchronously bind volumes and pod in a separate goroutine
    i.  BindPodVolumes() is called first. It makes all the necessary API updates and waits for
        PV controller to fully bind and provision the PVCs. If binding fails, the Pod is sent
        back through the scheduler.
    ii. After BindPodVolumes() is complete, then the scheduler does the final Pod->Node binding.

2. Once all the assume operations are done in d), the scheduler processes the next Pod in the scheduler queue

while the actual binding operation occurs in the background.

func NewVolumeBinder

NewVolumeBinder sets up all the caches needed for the scheduler to make volume binding decisions.


Path Synopsis