core

package
v1.2.0-alpha.5....-8a98da9 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Jan 29, 2018 License: Apache-2.0 Imports: 37 Imported by: 0

Documentation

Index

Constants

View Source
const (
	NoNodeAvailableMsg = "0/%v nodes are available"
	// NominatedNodeAnnotationKey is used to annotate a pod that has preempted other pods.
	// The scheduler uses the annotation to find that the pod shouldn't preempt more pods
	// when it gets to the head of scheduling queue again.
	// See podEligibleToPreemptOthers() for more information.
	NominatedNodeAnnotationKey = "scheduler.kubernetes.io/nominated-node-name"
)
View Source
const (
	DefaultExtenderTimeout = 5 * time.Second
)

Variables

View Source
var ErrNoNodesAvailable = fmt.Errorf("no nodes available to schedule pods")

Functions

func EqualPriorityMap

func EqualPriorityMap(_ *v1.Pod, _ interface{}, nodeInfo *schedulercache.NodeInfo) (schedulerapi.HostPriority, error)

EqualPriority is a prioritizer function that gives an equal weight of one to all nodes

func NewGenericScheduler

func NewGenericScheduler(
	cache schedulercache.Cache,
	eCache *EquivalenceCache,
	podQueue SchedulingQueue,
	predicates map[string]algorithm.FitPredicate,
	predicateMetaProducer algorithm.PredicateMetadataProducer,
	prioritizers []algorithm.PriorityConfig,
	priorityMetaProducer algorithm.PriorityMetadataProducer,
	extenders []algorithm.SchedulerExtender,
	volumeBinder *volumebinder.VolumeBinder,
	pvcLister corelisters.PersistentVolumeClaimLister,
	alwaysCheckAllPredicates bool) algorithm.ScheduleAlgorithm

func NominatedNodeName

func NominatedNodeName(pod *v1.Pod) string

func PrioritizeNodes

func PrioritizeNodes(
	pod *v1.Pod,
	nodeNameToInfo map[string]*schedulercache.NodeInfo,
	meta interface{},
	priorityConfigs []algorithm.PriorityConfig,
	nodes []*v1.Node,
	extenders []algorithm.SchedulerExtender,
) (schedulerapi.HostPriorityList, error)

Prioritizes the nodes by running the individual priority functions in parallel. Each priority function is expected to set a score of 0-10 0 is the lowest priority score (least preferred node) and 10 is the highest Each priority function can also have its own weight The node scores returned by the priority function are multiplied by the weights to get weighted scores All scores are finally combined (added) to get the total weighted scores of all nodes

Types

type AlgorithmCache

type AlgorithmCache struct {
	// contains filtered or unexported fields
}

The AlgorithmCache stores PredicateMap with predicate name as key

type EquivalenceCache

type EquivalenceCache struct {
	sync.RWMutex
	// contains filtered or unexported fields
}

EquivalenceCache holds: 1. a map of AlgorithmCache with node name as key 2. function to get equivalence pod

func NewEquivalenceCache

func NewEquivalenceCache(getEquivalencePodFunc algorithm.GetEquivalencePodFunc) *EquivalenceCache

func (*EquivalenceCache) InvalidateAllCachedPredicateItemOfNode

func (ec *EquivalenceCache) InvalidateAllCachedPredicateItemOfNode(nodeName string)

InvalidateAllCachedPredicateItemOfNode marks all cached items on given node as invalid

func (*EquivalenceCache) InvalidateCachedPredicateItem

func (ec *EquivalenceCache) InvalidateCachedPredicateItem(nodeName string, predicateKeys sets.String)

InvalidateCachedPredicateItem marks all items of given predicateKeys, of all pods, on the given node as invalid

func (*EquivalenceCache) InvalidateCachedPredicateItemForPodAdd

func (ec *EquivalenceCache) InvalidateCachedPredicateItemForPodAdd(pod *v1.Pod, nodeName string)

InvalidateCachedPredicateItemForPodAdd is a wrapper of InvalidateCachedPredicateItem for pod add case

func (*EquivalenceCache) InvalidateCachedPredicateItemOfAllNodes

func (ec *EquivalenceCache) InvalidateCachedPredicateItemOfAllNodes(predicateKeys sets.String)

InvalidateCachedPredicateItemOfAllNodes marks all items of given predicateKeys, of all pods, on all node as invalid

func (*EquivalenceCache) PredicateWithECache

func (ec *EquivalenceCache) PredicateWithECache(
	podName, nodeName, predicateKey string,
	equivalenceHash uint64,
) (bool, []algorithm.PredicateFailureReason, bool)

PredicateWithECache returns: 1. if fit 2. reasons if not fit 3. if this cache is invalid based on cached predicate results

func (*EquivalenceCache) UpdateCachedPredicateItem

func (ec *EquivalenceCache) UpdateCachedPredicateItem(
	podName, nodeName, predicateKey string,
	fit bool,
	reasons []algorithm.PredicateFailureReason,
	equivalenceHash uint64,
)

UpdateCachedPredicateItem updates pod predicate for equivalence class

type FIFO

type FIFO struct {
	*cache.FIFO
}

FIFO is basically a simple wrapper around cache.FIFO to make it compatible with the SchedulingQueue interface.

func NewFIFO

func NewFIFO() *FIFO

func (*FIFO) Add

func (f *FIFO) Add(pod *v1.Pod) error

func (*FIFO) AddIfNotPresent

func (f *FIFO) AddIfNotPresent(pod *v1.Pod) error

func (*FIFO) AddUnschedulableIfNotPresent

func (f *FIFO) AddUnschedulableIfNotPresent(pod *v1.Pod) error

AddUnschedulableIfNotPresent adds an unschedulable pod back to the queue. In FIFO it is added to the end of the queue.

func (*FIFO) AssignedPodAdded

func (f *FIFO) AssignedPodAdded(pod *v1.Pod)

FIFO does not need to react to events, as all pods are always in the active scheduling queue anyway.

func (*FIFO) AssignedPodUpdated

func (f *FIFO) AssignedPodUpdated(pod *v1.Pod)

func (*FIFO) Delete

func (f *FIFO) Delete(pod *v1.Pod) error

func (*FIFO) MoveAllToActiveQueue

func (f *FIFO) MoveAllToActiveQueue()

MoveAllToActiveQueue does nothing in FIFO as all pods are always in the active queue.

func (*FIFO) Pop

func (f *FIFO) Pop() (*v1.Pod, error)

Pop removes the head of FIFO and returns it. This is just a copy/paste of cache.Pop(queue Queue) from fifo.go that scheduler has always been using. There is a comment in that file saying that this method shouldn't be used in production code, but scheduler has always been using it. This function does minimal error checking.

func (*FIFO) Update

func (f *FIFO) Update(pod *v1.Pod) error

func (*FIFO) WaitingPodsForNode

func (f *FIFO) WaitingPodsForNode(nodeName string) []*v1.Pod

WaitingPodsForNode returns pods that are nominated to run on the given node, but FIFO does not support it.

type FailedPredicateMap

type FailedPredicateMap map[string][]algorithm.PredicateFailureReason

type FitError

type FitError struct {
	Pod              *v1.Pod
	NumAllNodes      int
	FailedPredicates FailedPredicateMap
}

func (*FitError) Error

func (f *FitError) Error() string

Error returns detailed information of why the pod failed to fit on each node

type HTTPExtender

type HTTPExtender struct {
	// contains filtered or unexported fields
}

HTTPExtender implements the algorithm.SchedulerExtender interface.

func (*HTTPExtender) Bind

func (h *HTTPExtender) Bind(binding *v1.Binding) error

Bind delegates the action of binding a pod to a node to the extender.

func (*HTTPExtender) Filter

func (h *HTTPExtender) Filter(pod *v1.Pod, nodes []*v1.Node, nodeNameToInfo map[string]*schedulercache.NodeInfo) ([]*v1.Node, schedulerapi.FailedNodesMap, error)

Filter based on extender implemented predicate functions. The filtered list is expected to be a subset of the supplied list. failedNodesMap optionally contains the list of failed nodes and failure reasons.

func (*HTTPExtender) IsBinder

func (h *HTTPExtender) IsBinder() bool

IsBinder returns whether this extender is configured for the Bind method.

func (*HTTPExtender) Prioritize

func (h *HTTPExtender) Prioritize(pod *v1.Pod, nodes []*v1.Node) (*schedulerapi.HostPriorityList, int, error)

Prioritize based on extender implemented priority functions. Weight*priority is added up for each such priority function. The returned score is added to the score computed by Kubernetes scheduler. The total score is used to do the host selection.

type Heap

type Heap struct {
	// contains filtered or unexported fields
}

Heap is a producer/consumer queue that implements a heap data structure. It can be used to implement priority queues and similar data structures.

func (*Heap) Add

func (h *Heap) Add(obj interface{}) error

Add inserts an item, and puts it in the queue. The item is updated if it already exists.

func (*Heap) AddIfNotPresent

func (h *Heap) AddIfNotPresent(obj interface{}) error

AddIfNotPresent inserts an item, and puts it in the queue. If an item with the key is present in the map, no changes is made to the item.

func (*Heap) BulkAdd

func (h *Heap) BulkAdd(list []interface{}) error

BulkAdd adds all the items in the list to the queue.

func (*Heap) Delete

func (h *Heap) Delete(obj interface{}) error

Delete removes an item.

func (*Heap) Get

func (h *Heap) Get(obj interface{}) (interface{}, bool, error)

Get returns the requested item, or sets exists=false.

func (*Heap) GetByKey

func (h *Heap) GetByKey(key string) (interface{}, bool, error)

GetByKey returns the requested item, or sets exists=false.

func (*Heap) List

func (h *Heap) List() []interface{}

List returns a list of all the items.

func (*Heap) Pop

func (h *Heap) Pop() (interface{}, error)

Pop returns the head of the heap.

func (*Heap) Update

func (h *Heap) Update(obj interface{}) error

Update is the same as Add in this implementation. When the item does not exist, it is added.

type HostPredicate

type HostPredicate struct {
	Fit         bool
	FailReasons []algorithm.PredicateFailureReason
}

HostPredicate is the cached predicate result

type KeyFunc

type KeyFunc func(obj interface{}) (string, error)

type LessFunc

type LessFunc func(interface{}, interface{}) bool

type PredicateMap

type PredicateMap map[uint64]HostPredicate

PredicateMap stores HostPrediacte with equivalence hash as key

type PriorityQueue

type PriorityQueue struct {
	// contains filtered or unexported fields
}

PriorityQueue implements a scheduling queue. It is an alternative to FIFO. The head of PriorityQueue is the highest priority pending pod. This structure has two sub queues. One sub-queue holds pods that are being considered for scheduling. This is called activeQ and is a Heap. Another queue holds pods that are already tried and are determined to be unschedulable. The latter is called unschedulableQ.

func NewPriorityQueue

func NewPriorityQueue() *PriorityQueue

func (*PriorityQueue) Add

func (p *PriorityQueue) Add(pod *v1.Pod) error

Add adds a pod to the active queue. It should be called only when a new pod is added so there is no chance the pod is already in either queue.

func (*PriorityQueue) AddIfNotPresent

func (p *PriorityQueue) AddIfNotPresent(pod *v1.Pod) error

AddIfNotPresent adds a pod to the active queue if it is not present in any of the two queues. If it is present in any, it doesn't do any thing.

func (*PriorityQueue) AddUnschedulableIfNotPresent

func (p *PriorityQueue) AddUnschedulableIfNotPresent(pod *v1.Pod) error

AddUnschedulableIfNotPresent does nothing if the pod is present in either queue. Otherwise it adds the pod to the unschedulable queue if p.receivedMoveRequest is false, and to the activeQ if p.receivedMoveRequest is true.

func (*PriorityQueue) AssignedPodAdded

func (p *PriorityQueue) AssignedPodAdded(pod *v1.Pod)

AssignedPodAdded is called when a bound pod is added. Creation of this pod may make pending pods with matching affinity terms schedulable.

func (*PriorityQueue) AssignedPodUpdated

func (p *PriorityQueue) AssignedPodUpdated(pod *v1.Pod)

AssignedPodUpdated is called when a bound pod is updated. Change of labels may make pending pods with matching affinity terms schedulable.

func (*PriorityQueue) Delete

func (p *PriorityQueue) Delete(pod *v1.Pod) error

Delete deletes the item from either of the two queues. It assumes the pod is only in one queue.

func (*PriorityQueue) MoveAllToActiveQueue

func (p *PriorityQueue) MoveAllToActiveQueue()

MoveAllToActiveQueue moves all pods from unschedulableQ to activeQ. This function adds all pods and then signals the condition variable to ensure that if Pop() is waiting for an item, it receives it after all the pods are in the queue and the head is the highest priority pod. TODO(bsalamat): We should add a back-off mechanism here so that a high priority pod which is unschedulable does not go to the head of the queue frequently. For example in a cluster where a lot of pods being deleted, such a high priority pod can deprive other pods from getting scheduled.

func (*PriorityQueue) Pop

func (p *PriorityQueue) Pop() (*v1.Pod, error)

Pop removes the head of the active queue and returns it. It blocks if the activeQ is empty and waits until a new item is added to the queue. It also clears receivedMoveRequest to mark the beginning of a new scheduling cycle.

func (*PriorityQueue) Update

func (p *PriorityQueue) Update(pod *v1.Pod) error

Update updates a pod in the active queue if present. Otherwise, it removes the item from the unschedulable queue and adds the updated one to the active queue.

func (*PriorityQueue) WaitingPodsForNode

func (p *PriorityQueue) WaitingPodsForNode(nodeName string) []*v1.Pod

WaitingPodsForNode returns pods that are nominated to run on the given node, but they are waiting for other pods to be removed from the node before they can be actually scheduled.

type SchedulingQueue

type SchedulingQueue interface {
	Add(pod *v1.Pod) error
	AddIfNotPresent(pod *v1.Pod) error
	AddUnschedulableIfNotPresent(pod *v1.Pod) error
	Pop() (*v1.Pod, error)
	Update(pod *v1.Pod) error
	Delete(pod *v1.Pod) error
	MoveAllToActiveQueue()
	AssignedPodAdded(pod *v1.Pod)
	AssignedPodUpdated(pod *v1.Pod)
	WaitingPodsForNode(nodeName string) []*v1.Pod
}

SchedulingQueue is an interface for a queue to store pods waiting to be scheduled. The interface follows a pattern similar to cache.FIFO and cache.Heap and makes it easy to use those data structures as a SchedulingQueue.

func NewSchedulingQueue

func NewSchedulingQueue() SchedulingQueue

NewSchedulingQueue initializes a new scheduling queue. If pod priority is enabled a priority queue is returned. If it is disabled, a FIFO is returned.

type UnschedulablePods

type UnschedulablePods interface {
	Add(pod *v1.Pod)
	Delete(pod *v1.Pod)
	Update(pod *v1.Pod)
	GetPodsWaitingForNode(nodeName string) []*v1.Pod
	Get(pod *v1.Pod) *v1.Pod
	Clear()
}

UnschedulablePods is an interface for a queue that is used to keep unschedulable pods. These pods are not actively reevaluated for scheduling. They are moved to the active scheduling queue on certain events, such as termination of a pod in the cluster, addition of nodes, etc.

type UnschedulablePodsMap

type UnschedulablePodsMap struct {
	// contains filtered or unexported fields
}

UnschedulablePodsMap holds pods that cannot be scheduled. This data structure is used to implement unschedulableQ.

func (*UnschedulablePodsMap) Add

func (u *UnschedulablePodsMap) Add(pod *v1.Pod)

Add adds a pod to the unschedulable pods.

func (*UnschedulablePodsMap) Clear

func (u *UnschedulablePodsMap) Clear()

Clear removes all the entries from the unschedulable maps.

func (*UnschedulablePodsMap) Delete

func (u *UnschedulablePodsMap) Delete(pod *v1.Pod)

Delete deletes a pod from the unschedulable pods.

func (*UnschedulablePodsMap) Get

func (u *UnschedulablePodsMap) Get(pod *v1.Pod) *v1.Pod

Get returns the pod if a pod with the same key as the key of the given "pod" is found in the map. It returns nil otherwise.

func (*UnschedulablePodsMap) GetPodsWaitingForNode

func (u *UnschedulablePodsMap) GetPodsWaitingForNode(nodeName string) []*v1.Pod

GetPodsWaitingForNode returns a list of unschedulable pods whose NominatedNodeNames are equal to the given nodeName.

func (*UnschedulablePodsMap) Update

func (u *UnschedulablePodsMap) Update(pod *v1.Pod)

Update updates a pod in the unschedulable pods.

type Victims

type Victims struct {
	// contains filtered or unexported fields
}

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL