Fresh dep ensure

This commit is contained in:
Mike Cronce
2018-11-26 13:23:56 -05:00
parent 93cb8a04d7
commit 407478ab9a
9016 changed files with 551394 additions and 279685 deletions

View File

@ -0,0 +1,43 @@
load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test")
go_library(
name = "go_default_library",
srcs = ["scheduling_queue.go"],
importpath = "k8s.io/kubernetes/pkg/scheduler/internal/queue",
visibility = ["//pkg/scheduler:__subpackages__"],
deps = [
"//pkg/api/v1/pod:go_default_library",
"//pkg/scheduler/algorithm/predicates:go_default_library",
"//pkg/scheduler/algorithm/priorities/util:go_default_library",
"//pkg/scheduler/util:go_default_library",
"//staging/src/k8s.io/api/core/v1:go_default_library",
"//staging/src/k8s.io/apimachinery/pkg/apis/meta/v1:go_default_library",
"//staging/src/k8s.io/client-go/tools/cache:go_default_library",
"//vendor/k8s.io/klog:go_default_library",
],
)
go_test(
name = "go_default_test",
srcs = ["scheduling_queue_test.go"],
embed = [":go_default_library"],
deps = [
"//pkg/scheduler/util:go_default_library",
"//staging/src/k8s.io/api/core/v1:go_default_library",
"//staging/src/k8s.io/apimachinery/pkg/apis/meta/v1:go_default_library",
],
)
filegroup(
name = "package-srcs",
srcs = glob(["**"]),
tags = ["automanaged"],
visibility = ["//visibility:private"],
)
filegroup(
name = "all-srcs",
srcs = [":package-srcs"],
tags = ["automanaged"],
visibility = ["//visibility:public"],
)

View File

@ -0,0 +1,769 @@
/*
Copyright 2017 The Kubernetes Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
// This file contains structures that implement scheduling queue types.
// Scheduling queues hold pods waiting to be scheduled. This file has two types
// of scheduling queue: 1) a FIFO, which is mostly the same as cache.FIFO, 2) a
// priority queue which has two sub queues. One sub-queue holds pods that are
// being considered for scheduling. This is called activeQ. Another queue holds
// pods that are already tried and are determined to be unschedulable. The latter
// is called unschedulableQ.
// FIFO is here for flag-gating purposes and allows us to use the traditional
// scheduling queue when util.PodPriorityEnabled() returns false.
package queue
import (
"container/heap"
"fmt"
"reflect"
"sync"
"k8s.io/klog"
"k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/client-go/tools/cache"
podutil "k8s.io/kubernetes/pkg/api/v1/pod"
"k8s.io/kubernetes/pkg/scheduler/algorithm/predicates"
priorityutil "k8s.io/kubernetes/pkg/scheduler/algorithm/priorities/util"
"k8s.io/kubernetes/pkg/scheduler/util"
)
var (
queueClosed = "scheduling queue is closed"
)
// 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.
type SchedulingQueue interface {
Add(pod *v1.Pod) error
AddIfNotPresent(pod *v1.Pod) error
AddUnschedulableIfNotPresent(pod *v1.Pod) error
// Pop removes the head of the queue and returns it. It blocks if the
// queue is empty and waits until a new item is added to the queue.
Pop() (*v1.Pod, error)
Update(oldPod, newPod *v1.Pod) error
Delete(pod *v1.Pod) error
MoveAllToActiveQueue()
AssignedPodAdded(pod *v1.Pod)
AssignedPodUpdated(pod *v1.Pod)
WaitingPodsForNode(nodeName string) []*v1.Pod
WaitingPods() []*v1.Pod
// Close closes the SchedulingQueue so that the goroutine which is
// waiting to pop items can exit gracefully.
Close()
// DeleteNominatedPodIfExists deletes nominatedPod from internal cache
DeleteNominatedPodIfExists(pod *v1.Pod)
}
// NewSchedulingQueue initializes a new scheduling queue. If pod priority is
// enabled a priority queue is returned. If it is disabled, a FIFO is returned.
func NewSchedulingQueue() SchedulingQueue {
if util.PodPriorityEnabled() {
return NewPriorityQueue()
}
return NewFIFO()
}
// FIFO is basically a simple wrapper around cache.FIFO to make it compatible
// with the SchedulingQueue interface.
type FIFO struct {
*cache.FIFO
}
var _ = SchedulingQueue(&FIFO{}) // Making sure that FIFO implements SchedulingQueue.
// Add adds a pod to the FIFO.
func (f *FIFO) Add(pod *v1.Pod) error {
return f.FIFO.Add(pod)
}
// AddIfNotPresent adds a pod to the FIFO if it is absent in the FIFO.
func (f *FIFO) AddIfNotPresent(pod *v1.Pod) error {
return f.FIFO.AddIfNotPresent(pod)
}
// AddUnschedulableIfNotPresent adds an unschedulable pod back to the queue. In
// FIFO it is added to the end of the queue.
func (f *FIFO) AddUnschedulableIfNotPresent(pod *v1.Pod) error {
return f.FIFO.AddIfNotPresent(pod)
}
// Update updates a pod in the FIFO.
func (f *FIFO) Update(oldPod, newPod *v1.Pod) error {
return f.FIFO.Update(newPod)
}
// Delete deletes a pod in the FIFO.
func (f *FIFO) Delete(pod *v1.Pod) error {
return f.FIFO.Delete(pod)
}
// 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 (f *FIFO) Pop() (*v1.Pod, error) {
result, err := f.FIFO.Pop(func(obj interface{}) error { return nil })
if err == cache.FIFOClosedError {
return nil, fmt.Errorf(queueClosed)
}
return result.(*v1.Pod), err
}
// WaitingPods returns all the waiting pods in the queue.
func (f *FIFO) WaitingPods() []*v1.Pod {
result := []*v1.Pod{}
for _, pod := range f.FIFO.List() {
result = append(result, pod.(*v1.Pod))
}
return result
}
// FIFO does not need to react to events, as all pods are always in the active
// scheduling queue anyway.
// AssignedPodAdded does nothing here.
func (f *FIFO) AssignedPodAdded(pod *v1.Pod) {}
// AssignedPodUpdated does nothing here.
func (f *FIFO) AssignedPodUpdated(pod *v1.Pod) {}
// MoveAllToActiveQueue does nothing in FIFO as all pods are always in the active queue.
func (f *FIFO) MoveAllToActiveQueue() {}
// WaitingPodsForNode returns pods that are nominated to run on the given node,
// but FIFO does not support it.
func (f *FIFO) WaitingPodsForNode(nodeName string) []*v1.Pod {
return nil
}
// Close closes the FIFO queue.
func (f *FIFO) Close() {
f.FIFO.Close()
}
// DeleteNominatedPodIfExists does nothing in FIFO.
func (f *FIFO) DeleteNominatedPodIfExists(pod *v1.Pod) {}
// NewFIFO creates a FIFO object.
func NewFIFO() *FIFO {
return &FIFO{FIFO: cache.NewFIFO(cache.MetaNamespaceKeyFunc)}
}
// NominatedNodeName returns nominated node name of a Pod.
func NominatedNodeName(pod *v1.Pod) string {
return pod.Status.NominatedNodeName
}
// 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.
type PriorityQueue struct {
lock sync.RWMutex
cond sync.Cond
// activeQ is heap structure that scheduler actively looks at to find pods to
// schedule. Head of heap is the highest priority pod.
activeQ *Heap
// unschedulableQ holds pods that have been tried and determined unschedulable.
unschedulableQ *UnschedulablePodsMap
// nominatedPods is a map keyed by a node name and the value is a list of
// pods which are nominated to run on the node. These are pods which can be in
// the activeQ or unschedulableQ.
nominatedPods map[string][]*v1.Pod
// receivedMoveRequest is set to true whenever we receive a request to move a
// pod from the unschedulableQ to the activeQ, and is set to false, when we pop
// a pod from the activeQ. It indicates if we received a move request when a
// pod was in flight (we were trying to schedule it). In such a case, we put
// the pod back into the activeQ if it is determined unschedulable.
receivedMoveRequest bool
// closed indicates that the queue is closed.
// It is mainly used to let Pop() exit its control loop while waiting for an item.
closed bool
}
// Making sure that PriorityQueue implements SchedulingQueue.
var _ = SchedulingQueue(&PriorityQueue{})
// NewPriorityQueue creates a PriorityQueue object.
func NewPriorityQueue() *PriorityQueue {
pq := &PriorityQueue{
activeQ: newHeap(cache.MetaNamespaceKeyFunc, util.HigherPriorityPod),
unschedulableQ: newUnschedulablePodsMap(),
nominatedPods: map[string][]*v1.Pod{},
}
pq.cond.L = &pq.lock
return pq
}
// addNominatedPodIfNeeded adds a pod to nominatedPods if it has a NominatedNodeName and it does not
// already exist in the map. Adding an existing pod is not going to update the pod.
func (p *PriorityQueue) addNominatedPodIfNeeded(pod *v1.Pod) {
nnn := NominatedNodeName(pod)
if len(nnn) > 0 {
for _, np := range p.nominatedPods[nnn] {
if np.UID == pod.UID {
klog.V(4).Infof("Pod %v/%v already exists in the nominated map!", pod.Namespace, pod.Name)
return
}
}
p.nominatedPods[nnn] = append(p.nominatedPods[nnn], pod)
}
}
// deleteNominatedPodIfExists deletes a pod from the nominatedPods.
// NOTE: this function assumes lock has been acquired in caller.
func (p *PriorityQueue) deleteNominatedPodIfExists(pod *v1.Pod) {
nnn := NominatedNodeName(pod)
if len(nnn) > 0 {
for i, np := range p.nominatedPods[nnn] {
if np.UID == pod.UID {
p.nominatedPods[nnn] = append(p.nominatedPods[nnn][:i], p.nominatedPods[nnn][i+1:]...)
if len(p.nominatedPods[nnn]) == 0 {
delete(p.nominatedPods, nnn)
}
break
}
}
}
}
// updateNominatedPod updates a pod in the nominatedPods.
func (p *PriorityQueue) updateNominatedPod(oldPod, newPod *v1.Pod) {
// Even if the nominated node name of the Pod is not changed, we must delete and add it again
// to ensure that its pointer is updated.
p.deleteNominatedPodIfExists(oldPod)
p.addNominatedPodIfNeeded(newPod)
}
// 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 (p *PriorityQueue) Add(pod *v1.Pod) error {
p.lock.Lock()
defer p.lock.Unlock()
err := p.activeQ.Add(pod)
if err != nil {
klog.Errorf("Error adding pod %v/%v to the scheduling queue: %v", pod.Namespace, pod.Name, err)
} else {
if p.unschedulableQ.get(pod) != nil {
klog.Errorf("Error: pod %v/%v is already in the unschedulable queue.", pod.Namespace, pod.Name)
p.deleteNominatedPodIfExists(pod)
p.unschedulableQ.delete(pod)
}
p.addNominatedPodIfNeeded(pod)
p.cond.Broadcast()
}
return err
}
// 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 (p *PriorityQueue) AddIfNotPresent(pod *v1.Pod) error {
p.lock.Lock()
defer p.lock.Unlock()
if p.unschedulableQ.get(pod) != nil {
return nil
}
if _, exists, _ := p.activeQ.Get(pod); exists {
return nil
}
err := p.activeQ.Add(pod)
if err != nil {
klog.Errorf("Error adding pod %v/%v to the scheduling queue: %v", pod.Namespace, pod.Name, err)
} else {
p.addNominatedPodIfNeeded(pod)
p.cond.Broadcast()
}
return err
}
func isPodUnschedulable(pod *v1.Pod) bool {
_, cond := podutil.GetPodCondition(&pod.Status, v1.PodScheduled)
return cond != nil && cond.Status == v1.ConditionFalse && cond.Reason == v1.PodReasonUnschedulable
}
// 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 (p *PriorityQueue) AddUnschedulableIfNotPresent(pod *v1.Pod) error {
p.lock.Lock()
defer p.lock.Unlock()
if p.unschedulableQ.get(pod) != nil {
return fmt.Errorf("pod is already present in unschedulableQ")
}
if _, exists, _ := p.activeQ.Get(pod); exists {
return fmt.Errorf("pod is already present in the activeQ")
}
if !p.receivedMoveRequest && isPodUnschedulable(pod) {
p.unschedulableQ.addOrUpdate(pod)
p.addNominatedPodIfNeeded(pod)
return nil
}
err := p.activeQ.Add(pod)
if err == nil {
p.addNominatedPodIfNeeded(pod)
p.cond.Broadcast()
}
return err
}
// 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 (p *PriorityQueue) Pop() (*v1.Pod, error) {
p.lock.Lock()
defer p.lock.Unlock()
for len(p.activeQ.data.queue) == 0 {
// When the queue is empty, invocation of Pop() is blocked until new item is enqueued.
// When Close() is called, the p.closed is set and the condition is broadcast,
// which causes this loop to continue and return from the Pop().
if p.closed {
return nil, fmt.Errorf(queueClosed)
}
p.cond.Wait()
}
obj, err := p.activeQ.Pop()
if err != nil {
return nil, err
}
pod := obj.(*v1.Pod)
p.receivedMoveRequest = false
return pod, err
}
// isPodUpdated checks if the pod is updated in a way that it may have become
// schedulable. It drops status of the pod and compares it with old version.
func isPodUpdated(oldPod, newPod *v1.Pod) bool {
strip := func(pod *v1.Pod) *v1.Pod {
p := pod.DeepCopy()
p.ResourceVersion = ""
p.Generation = 0
p.Status = v1.PodStatus{}
return p
}
return !reflect.DeepEqual(strip(oldPod), strip(newPod))
}
// 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 (p *PriorityQueue) Update(oldPod, newPod *v1.Pod) error {
p.lock.Lock()
defer p.lock.Unlock()
// If the pod is already in the active queue, just update it there.
if _, exists, _ := p.activeQ.Get(newPod); exists {
p.updateNominatedPod(oldPod, newPod)
err := p.activeQ.Update(newPod)
return err
}
// If the pod is in the unschedulable queue, updating it may make it schedulable.
if usPod := p.unschedulableQ.get(newPod); usPod != nil {
p.updateNominatedPod(oldPod, newPod)
if isPodUpdated(oldPod, newPod) {
p.unschedulableQ.delete(usPod)
err := p.activeQ.Add(newPod)
if err == nil {
p.cond.Broadcast()
}
return err
}
p.unschedulableQ.addOrUpdate(newPod)
return nil
}
// If pod is not in any of the two queue, we put it in the active queue.
err := p.activeQ.Add(newPod)
if err == nil {
p.addNominatedPodIfNeeded(newPod)
p.cond.Broadcast()
}
return err
}
// Delete deletes the item from either of the two queues. It assumes the pod is
// only in one queue.
func (p *PriorityQueue) Delete(pod *v1.Pod) error {
p.lock.Lock()
defer p.lock.Unlock()
p.deleteNominatedPodIfExists(pod)
err := p.activeQ.Delete(pod)
if err != nil { // The item was probably not found in the activeQ.
p.unschedulableQ.delete(pod)
}
return nil
}
// AssignedPodAdded is called when a bound pod is added. Creation of this pod
// may make pending pods with matching affinity terms schedulable.
func (p *PriorityQueue) AssignedPodAdded(pod *v1.Pod) {
p.lock.Lock()
p.movePodsToActiveQueue(p.getUnschedulablePodsWithMatchingAffinityTerm(pod))
p.lock.Unlock()
}
// AssignedPodUpdated is called when a bound pod is updated. Change of labels
// may make pending pods with matching affinity terms schedulable.
func (p *PriorityQueue) AssignedPodUpdated(pod *v1.Pod) {
p.lock.Lock()
p.movePodsToActiveQueue(p.getUnschedulablePodsWithMatchingAffinityTerm(pod))
p.lock.Unlock()
}
// 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 (p *PriorityQueue) MoveAllToActiveQueue() {
p.lock.Lock()
defer p.lock.Unlock()
for _, pod := range p.unschedulableQ.pods {
if err := p.activeQ.Add(pod); err != nil {
klog.Errorf("Error adding pod %v/%v to the scheduling queue: %v", pod.Namespace, pod.Name, err)
}
}
p.unschedulableQ.clear()
p.receivedMoveRequest = true
p.cond.Broadcast()
}
// NOTE: this function assumes lock has been acquired in caller
func (p *PriorityQueue) movePodsToActiveQueue(pods []*v1.Pod) {
for _, pod := range pods {
if err := p.activeQ.Add(pod); err == nil {
p.unschedulableQ.delete(pod)
} else {
klog.Errorf("Error adding pod %v/%v to the scheduling queue: %v", pod.Namespace, pod.Name, err)
}
}
p.receivedMoveRequest = true
p.cond.Broadcast()
}
// getUnschedulablePodsWithMatchingAffinityTerm returns unschedulable pods which have
// any affinity term that matches "pod".
// NOTE: this function assumes lock has been acquired in caller.
func (p *PriorityQueue) getUnschedulablePodsWithMatchingAffinityTerm(pod *v1.Pod) []*v1.Pod {
var podsToMove []*v1.Pod
for _, up := range p.unschedulableQ.pods {
affinity := up.Spec.Affinity
if affinity != nil && affinity.PodAffinity != nil {
terms := predicates.GetPodAffinityTerms(affinity.PodAffinity)
for _, term := range terms {
namespaces := priorityutil.GetNamespacesFromPodAffinityTerm(up, &term)
selector, err := metav1.LabelSelectorAsSelector(term.LabelSelector)
if err != nil {
klog.Errorf("Error getting label selectors for pod: %v.", up.Name)
}
if priorityutil.PodMatchesTermsNamespaceAndSelector(pod, namespaces, selector) {
podsToMove = append(podsToMove, up)
break
}
}
}
}
return podsToMove
}
// 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.
func (p *PriorityQueue) WaitingPodsForNode(nodeName string) []*v1.Pod {
p.lock.RLock()
defer p.lock.RUnlock()
if list, ok := p.nominatedPods[nodeName]; ok {
return list
}
return nil
}
// WaitingPods returns all the waiting pods in the queue.
func (p *PriorityQueue) WaitingPods() []*v1.Pod {
p.lock.Lock()
defer p.lock.Unlock()
result := []*v1.Pod{}
for _, pod := range p.activeQ.List() {
result = append(result, pod.(*v1.Pod))
}
for _, pod := range p.unschedulableQ.pods {
result = append(result, pod)
}
return result
}
// Close closes the priority queue.
func (p *PriorityQueue) Close() {
p.lock.Lock()
defer p.lock.Unlock()
p.closed = true
p.cond.Broadcast()
}
// DeleteNominatedPodIfExists deletes pod from internal cache if it's a nominatedPod
func (p *PriorityQueue) DeleteNominatedPodIfExists(pod *v1.Pod) {
p.lock.Lock()
p.deleteNominatedPodIfExists(pod)
p.lock.Unlock()
}
// UnschedulablePodsMap holds pods that cannot be scheduled. This data structure
// is used to implement unschedulableQ.
type UnschedulablePodsMap struct {
// pods is a map key by a pod's full-name and the value is a pointer to the pod.
pods map[string]*v1.Pod
keyFunc func(*v1.Pod) string
}
// Add adds a pod to the unschedulable pods.
func (u *UnschedulablePodsMap) addOrUpdate(pod *v1.Pod) {
u.pods[u.keyFunc(pod)] = pod
}
// Delete deletes a pod from the unschedulable pods.
func (u *UnschedulablePodsMap) delete(pod *v1.Pod) {
delete(u.pods, u.keyFunc(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 (u *UnschedulablePodsMap) get(pod *v1.Pod) *v1.Pod {
podKey := u.keyFunc(pod)
if p, exists := u.pods[podKey]; exists {
return p
}
return nil
}
// Clear removes all the entries from the unschedulable maps.
func (u *UnschedulablePodsMap) clear() {
u.pods = make(map[string]*v1.Pod)
}
// newUnschedulablePodsMap initializes a new object of UnschedulablePodsMap.
func newUnschedulablePodsMap() *UnschedulablePodsMap {
return &UnschedulablePodsMap{
pods: make(map[string]*v1.Pod),
keyFunc: util.GetPodFullName,
}
}
// Below is the implementation of the a heap. The logic is pretty much the same
// as cache.heap, however, this heap does not perform synchronization. It leaves
// synchronization to the SchedulingQueue.
// LessFunc is a function type to compare two objects.
type LessFunc func(interface{}, interface{}) bool
// KeyFunc is a function type to get the key from an object.
type KeyFunc func(obj interface{}) (string, error)
type heapItem struct {
obj interface{} // The object which is stored in the heap.
index int // The index of the object's key in the Heap.queue.
}
type itemKeyValue struct {
key string
obj interface{}
}
// heapData is an internal struct that implements the standard heap interface
// and keeps the data stored in the heap.
type heapData struct {
// items is a map from key of the objects to the objects and their index.
// We depend on the property that items in the map are in the queue and vice versa.
items map[string]*heapItem
// queue implements a heap data structure and keeps the order of elements
// according to the heap invariant. The queue keeps the keys of objects stored
// in "items".
queue []string
// keyFunc is used to make the key used for queued item insertion and retrieval, and
// should be deterministic.
keyFunc KeyFunc
// lessFunc is used to compare two objects in the heap.
lessFunc LessFunc
}
var (
_ = heap.Interface(&heapData{}) // heapData is a standard heap
)
// Less compares two objects and returns true if the first one should go
// in front of the second one in the heap.
func (h *heapData) Less(i, j int) bool {
if i > len(h.queue) || j > len(h.queue) {
return false
}
itemi, ok := h.items[h.queue[i]]
if !ok {
return false
}
itemj, ok := h.items[h.queue[j]]
if !ok {
return false
}
return h.lessFunc(itemi.obj, itemj.obj)
}
// Len returns the number of items in the Heap.
func (h *heapData) Len() int { return len(h.queue) }
// Swap implements swapping of two elements in the heap. This is a part of standard
// heap interface and should never be called directly.
func (h *heapData) Swap(i, j int) {
h.queue[i], h.queue[j] = h.queue[j], h.queue[i]
item := h.items[h.queue[i]]
item.index = i
item = h.items[h.queue[j]]
item.index = j
}
// Push is supposed to be called by heap.Push only.
func (h *heapData) Push(kv interface{}) {
keyValue := kv.(*itemKeyValue)
n := len(h.queue)
h.items[keyValue.key] = &heapItem{keyValue.obj, n}
h.queue = append(h.queue, keyValue.key)
}
// Pop is supposed to be called by heap.Pop only.
func (h *heapData) Pop() interface{} {
key := h.queue[len(h.queue)-1]
h.queue = h.queue[0 : len(h.queue)-1]
item, ok := h.items[key]
if !ok {
// This is an error
return nil
}
delete(h.items, key)
return item.obj
}
// Heap is a producer/consumer queue that implements a heap data structure.
// It can be used to implement priority queues and similar data structures.
type Heap struct {
// data stores objects and has a queue that keeps their ordering according
// to the heap invariant.
data *heapData
}
// Add inserts an item, and puts it in the queue. The item is updated if it
// already exists.
func (h *Heap) Add(obj interface{}) error {
key, err := h.data.keyFunc(obj)
if err != nil {
return cache.KeyError{Obj: obj, Err: err}
}
if _, exists := h.data.items[key]; exists {
h.data.items[key].obj = obj
heap.Fix(h.data, h.data.items[key].index)
} else {
heap.Push(h.data, &itemKeyValue{key, obj})
}
return nil
}
// 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 (h *Heap) AddIfNotPresent(obj interface{}) error {
key, err := h.data.keyFunc(obj)
if err != nil {
return cache.KeyError{Obj: obj, Err: err}
}
if _, exists := h.data.items[key]; !exists {
heap.Push(h.data, &itemKeyValue{key, obj})
}
return nil
}
// Update is the same as Add in this implementation. When the item does not
// exist, it is added.
func (h *Heap) Update(obj interface{}) error {
return h.Add(obj)
}
// Delete removes an item.
func (h *Heap) Delete(obj interface{}) error {
key, err := h.data.keyFunc(obj)
if err != nil {
return cache.KeyError{Obj: obj, Err: err}
}
if item, ok := h.data.items[key]; ok {
heap.Remove(h.data, item.index)
return nil
}
return fmt.Errorf("object not found")
}
// Pop returns the head of the heap.
func (h *Heap) Pop() (interface{}, error) {
obj := heap.Pop(h.data)
if obj != nil {
return obj, nil
}
return nil, fmt.Errorf("object was removed from heap data")
}
// Get returns the requested item, or sets exists=false.
func (h *Heap) Get(obj interface{}) (interface{}, bool, error) {
key, err := h.data.keyFunc(obj)
if err != nil {
return nil, false, cache.KeyError{Obj: obj, Err: err}
}
return h.GetByKey(key)
}
// GetByKey returns the requested item, or sets exists=false.
func (h *Heap) GetByKey(key string) (interface{}, bool, error) {
item, exists := h.data.items[key]
if !exists {
return nil, false, nil
}
return item.obj, true, nil
}
// List returns a list of all the items.
func (h *Heap) List() []interface{} {
list := make([]interface{}, 0, len(h.data.items))
for _, item := range h.data.items {
list = append(list, item.obj)
}
return list
}
// newHeap returns a Heap which can be used to queue up items to process.
func newHeap(keyFn KeyFunc, lessFn LessFunc) *Heap {
return &Heap{
data: &heapData{
items: map[string]*heapItem{},
queue: []string{},
keyFunc: keyFn,
lessFunc: lessFn,
},
}
}

View File

@ -0,0 +1,514 @@
/*
Copyright 2017 The Kubernetes Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
package queue
import (
"fmt"
"reflect"
"sync"
"testing"
"k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/kubernetes/pkg/scheduler/util"
)
var negPriority, lowPriority, midPriority, highPriority, veryHighPriority = int32(-100), int32(0), int32(100), int32(1000), int32(10000)
var mediumPriority = (lowPriority + highPriority) / 2
var highPriorityPod, highPriNominatedPod, medPriorityPod, unschedulablePod = v1.Pod{
ObjectMeta: metav1.ObjectMeta{
Name: "hpp",
Namespace: "ns1",
UID: "hppns1",
},
Spec: v1.PodSpec{
Priority: &highPriority,
},
},
v1.Pod{
ObjectMeta: metav1.ObjectMeta{
Name: "hpp",
Namespace: "ns1",
UID: "hppns1",
},
Spec: v1.PodSpec{
Priority: &highPriority,
},
Status: v1.PodStatus{
NominatedNodeName: "node1",
},
},
v1.Pod{
ObjectMeta: metav1.ObjectMeta{
Name: "mpp",
Namespace: "ns2",
UID: "mppns2",
Annotations: map[string]string{
"annot2": "val2",
},
},
Spec: v1.PodSpec{
Priority: &mediumPriority,
},
Status: v1.PodStatus{
NominatedNodeName: "node1",
},
},
v1.Pod{
ObjectMeta: metav1.ObjectMeta{
Name: "up",
Namespace: "ns1",
UID: "upns1",
Annotations: map[string]string{
"annot2": "val2",
},
},
Spec: v1.PodSpec{
Priority: &lowPriority,
},
Status: v1.PodStatus{
Conditions: []v1.PodCondition{
{
Type: v1.PodScheduled,
Status: v1.ConditionFalse,
Reason: v1.PodReasonUnschedulable,
},
},
NominatedNodeName: "node1",
},
}
func TestPriorityQueue_Add(t *testing.T) {
q := NewPriorityQueue()
q.Add(&medPriorityPod)
q.Add(&unschedulablePod)
q.Add(&highPriorityPod)
expectedNominatedPods := map[string][]*v1.Pod{
"node1": {&medPriorityPod, &unschedulablePod},
}
if !reflect.DeepEqual(q.nominatedPods, expectedNominatedPods) {
t.Errorf("Unexpected nominated map after adding pods. Expected: %v, got: %v", expectedNominatedPods, q.nominatedPods)
}
if p, err := q.Pop(); err != nil || p != &highPriorityPod {
t.Errorf("Expected: %v after Pop, but got: %v", highPriorityPod.Name, p.Name)
}
if p, err := q.Pop(); err != nil || p != &medPriorityPod {
t.Errorf("Expected: %v after Pop, but got: %v", medPriorityPod.Name, p.Name)
}
if p, err := q.Pop(); err != nil || p != &unschedulablePod {
t.Errorf("Expected: %v after Pop, but got: %v", unschedulablePod.Name, p.Name)
}
if len(q.nominatedPods["node1"]) != 2 {
t.Errorf("Expected medPriorityPod and unschedulablePod to be still present in nomindatePods: %v", q.nominatedPods["node1"])
}
}
func TestPriorityQueue_AddIfNotPresent(t *testing.T) {
q := NewPriorityQueue()
q.unschedulableQ.addOrUpdate(&highPriNominatedPod)
q.AddIfNotPresent(&highPriNominatedPod) // Must not add anything.
q.AddIfNotPresent(&medPriorityPod)
q.AddIfNotPresent(&unschedulablePod)
expectedNominatedPods := map[string][]*v1.Pod{
"node1": {&medPriorityPod, &unschedulablePod},
}
if !reflect.DeepEqual(q.nominatedPods, expectedNominatedPods) {
t.Errorf("Unexpected nominated map after adding pods. Expected: %v, got: %v", expectedNominatedPods, q.nominatedPods)
}
if p, err := q.Pop(); err != nil || p != &medPriorityPod {
t.Errorf("Expected: %v after Pop, but got: %v", medPriorityPod.Name, p.Name)
}
if p, err := q.Pop(); err != nil || p != &unschedulablePod {
t.Errorf("Expected: %v after Pop, but got: %v", unschedulablePod.Name, p.Name)
}
if len(q.nominatedPods["node1"]) != 2 {
t.Errorf("Expected medPriorityPod and unschedulablePod to be still present in nomindatePods: %v", q.nominatedPods["node1"])
}
if q.unschedulableQ.get(&highPriNominatedPod) != &highPriNominatedPod {
t.Errorf("Pod %v was not found in the unschedulableQ.", highPriNominatedPod.Name)
}
}
func TestPriorityQueue_AddUnschedulableIfNotPresent(t *testing.T) {
q := NewPriorityQueue()
q.Add(&highPriNominatedPod)
q.AddUnschedulableIfNotPresent(&highPriNominatedPod) // Must not add anything.
q.AddUnschedulableIfNotPresent(&medPriorityPod) // This should go to activeQ.
q.AddUnschedulableIfNotPresent(&unschedulablePod)
expectedNominatedPods := map[string][]*v1.Pod{
"node1": {&highPriNominatedPod, &medPriorityPod, &unschedulablePod},
}
if !reflect.DeepEqual(q.nominatedPods, expectedNominatedPods) {
t.Errorf("Unexpected nominated map after adding pods. Expected: %v, got: %v", expectedNominatedPods, q.nominatedPods)
}
if p, err := q.Pop(); err != nil || p != &highPriNominatedPod {
t.Errorf("Expected: %v after Pop, but got: %v", highPriNominatedPod.Name, p.Name)
}
if p, err := q.Pop(); err != nil || p != &medPriorityPod {
t.Errorf("Expected: %v after Pop, but got: %v", medPriorityPod.Name, p.Name)
}
if len(q.nominatedPods) != 1 {
t.Errorf("Expected nomindatePods to have one element: %v", q.nominatedPods)
}
if q.unschedulableQ.get(&unschedulablePod) != &unschedulablePod {
t.Errorf("Pod %v was not found in the unschedulableQ.", unschedulablePod.Name)
}
}
func TestPriorityQueue_Pop(t *testing.T) {
q := NewPriorityQueue()
wg := sync.WaitGroup{}
wg.Add(1)
go func() {
defer wg.Done()
if p, err := q.Pop(); err != nil || p != &medPriorityPod {
t.Errorf("Expected: %v after Pop, but got: %v", medPriorityPod.Name, p.Name)
}
if len(q.nominatedPods["node1"]) != 1 {
t.Errorf("Expected medPriorityPod to be present in nomindatePods: %v", q.nominatedPods["node1"])
}
}()
q.Add(&medPriorityPod)
wg.Wait()
}
func TestPriorityQueue_Update(t *testing.T) {
q := NewPriorityQueue()
q.Update(nil, &highPriorityPod)
if _, exists, _ := q.activeQ.Get(&highPriorityPod); !exists {
t.Errorf("Expected %v to be added to activeQ.", highPriorityPod.Name)
}
if len(q.nominatedPods) != 0 {
t.Errorf("Expected nomindatePods to be empty: %v", q.nominatedPods)
}
// Update highPriorityPod and add a nominatedNodeName to it.
q.Update(&highPriorityPod, &highPriNominatedPod)
if q.activeQ.data.Len() != 1 {
t.Error("Expected only one item in activeQ.")
}
if len(q.nominatedPods) != 1 {
t.Errorf("Expected one item in nomindatePods map: %v", q.nominatedPods)
}
// Updating an unschedulable pod which is not in any of the two queues, should
// add the pod to activeQ.
q.Update(&unschedulablePod, &unschedulablePod)
if _, exists, _ := q.activeQ.Get(&unschedulablePod); !exists {
t.Errorf("Expected %v to be added to activeQ.", unschedulablePod.Name)
}
// Updating a pod that is already in activeQ, should not change it.
q.Update(&unschedulablePod, &unschedulablePod)
if len(q.unschedulableQ.pods) != 0 {
t.Error("Expected unschedulableQ to be empty.")
}
if _, exists, _ := q.activeQ.Get(&unschedulablePod); !exists {
t.Errorf("Expected: %v to be added to activeQ.", unschedulablePod.Name)
}
if p, err := q.Pop(); err != nil || p != &highPriNominatedPod {
t.Errorf("Expected: %v after Pop, but got: %v", highPriorityPod.Name, p.Name)
}
}
func TestPriorityQueue_Delete(t *testing.T) {
q := NewPriorityQueue()
q.Update(&highPriorityPod, &highPriNominatedPod)
q.Add(&unschedulablePod)
q.Delete(&highPriNominatedPod)
if _, exists, _ := q.activeQ.Get(&unschedulablePod); !exists {
t.Errorf("Expected %v to be in activeQ.", unschedulablePod.Name)
}
if _, exists, _ := q.activeQ.Get(&highPriNominatedPod); exists {
t.Errorf("Didn't expect %v to be in activeQ.", highPriorityPod.Name)
}
if len(q.nominatedPods) != 1 {
t.Errorf("Expected nomindatePods to have only 'unschedulablePod': %v", q.nominatedPods)
}
q.Delete(&unschedulablePod)
if len(q.nominatedPods) != 0 {
t.Errorf("Expected nomindatePods to be empty: %v", q.nominatedPods)
}
}
func TestPriorityQueue_MoveAllToActiveQueue(t *testing.T) {
q := NewPriorityQueue()
q.Add(&medPriorityPod)
q.unschedulableQ.addOrUpdate(&unschedulablePod)
q.unschedulableQ.addOrUpdate(&highPriorityPod)
q.MoveAllToActiveQueue()
if q.activeQ.data.Len() != 3 {
t.Error("Expected all items to be in activeQ.")
}
}
// TestPriorityQueue_AssignedPodAdded tests AssignedPodAdded. It checks that
// when a pod with pod affinity is in unschedulableQ and another pod with a
// matching label is added, the unschedulable pod is moved to activeQ.
func TestPriorityQueue_AssignedPodAdded(t *testing.T) {
affinityPod := unschedulablePod.DeepCopy()
affinityPod.Name = "afp"
affinityPod.Spec = v1.PodSpec{
Affinity: &v1.Affinity{
PodAffinity: &v1.PodAffinity{
RequiredDuringSchedulingIgnoredDuringExecution: []v1.PodAffinityTerm{
{
LabelSelector: &metav1.LabelSelector{
MatchExpressions: []metav1.LabelSelectorRequirement{
{
Key: "service",
Operator: metav1.LabelSelectorOpIn,
Values: []string{"securityscan", "value2"},
},
},
},
TopologyKey: "region",
},
},
},
},
Priority: &mediumPriority,
}
labelPod := v1.Pod{
ObjectMeta: metav1.ObjectMeta{
Name: "lbp",
Namespace: affinityPod.Namespace,
Labels: map[string]string{"service": "securityscan"},
},
Spec: v1.PodSpec{NodeName: "machine1"},
}
q := NewPriorityQueue()
q.Add(&medPriorityPod)
// Add a couple of pods to the unschedulableQ.
q.unschedulableQ.addOrUpdate(&unschedulablePod)
q.unschedulableQ.addOrUpdate(affinityPod)
// Simulate addition of an assigned pod. The pod has matching labels for
// affinityPod. So, affinityPod should go to activeQ.
q.AssignedPodAdded(&labelPod)
if q.unschedulableQ.get(affinityPod) != nil {
t.Error("affinityPod is still in the unschedulableQ.")
}
if _, exists, _ := q.activeQ.Get(affinityPod); !exists {
t.Error("affinityPod is not moved to activeQ.")
}
// Check that the other pod is still in the unschedulableQ.
if q.unschedulableQ.get(&unschedulablePod) == nil {
t.Error("unschedulablePod is not in the unschedulableQ.")
}
}
func TestPriorityQueue_WaitingPodsForNode(t *testing.T) {
q := NewPriorityQueue()
q.Add(&medPriorityPod)
q.Add(&unschedulablePod)
q.Add(&highPriorityPod)
if p, err := q.Pop(); err != nil || p != &highPriorityPod {
t.Errorf("Expected: %v after Pop, but got: %v", highPriorityPod.Name, p.Name)
}
expectedList := []*v1.Pod{&medPriorityPod, &unschedulablePod}
if !reflect.DeepEqual(expectedList, q.WaitingPodsForNode("node1")) {
t.Error("Unexpected list of nominated Pods for node.")
}
if q.WaitingPodsForNode("node2") != nil {
t.Error("Expected list of nominated Pods for node2 to be empty.")
}
}
func TestUnschedulablePodsMap(t *testing.T) {
var pods = []*v1.Pod{
{
ObjectMeta: metav1.ObjectMeta{
Name: "p0",
Namespace: "ns1",
Annotations: map[string]string{
"annot1": "val1",
},
},
Status: v1.PodStatus{
NominatedNodeName: "node1",
},
},
{
ObjectMeta: metav1.ObjectMeta{
Name: "p1",
Namespace: "ns1",
Annotations: map[string]string{
"annot": "val",
},
},
},
{
ObjectMeta: metav1.ObjectMeta{
Name: "p2",
Namespace: "ns2",
Annotations: map[string]string{
"annot2": "val2", "annot3": "val3",
},
},
Status: v1.PodStatus{
NominatedNodeName: "node3",
},
},
{
ObjectMeta: metav1.ObjectMeta{
Name: "p3",
Namespace: "ns4",
},
Status: v1.PodStatus{
NominatedNodeName: "node1",
},
},
}
var updatedPods = make([]*v1.Pod, len(pods))
updatedPods[0] = pods[0].DeepCopy()
updatedPods[1] = pods[1].DeepCopy()
updatedPods[3] = pods[3].DeepCopy()
tests := []struct {
name string
podsToAdd []*v1.Pod
expectedMapAfterAdd map[string]*v1.Pod
podsToUpdate []*v1.Pod
expectedMapAfterUpdate map[string]*v1.Pod
podsToDelete []*v1.Pod
expectedMapAfterDelete map[string]*v1.Pod
}{
{
name: "create, update, delete subset of pods",
podsToAdd: []*v1.Pod{pods[0], pods[1], pods[2], pods[3]},
expectedMapAfterAdd: map[string]*v1.Pod{
util.GetPodFullName(pods[0]): pods[0],
util.GetPodFullName(pods[1]): pods[1],
util.GetPodFullName(pods[2]): pods[2],
util.GetPodFullName(pods[3]): pods[3],
},
podsToUpdate: []*v1.Pod{updatedPods[0]},
expectedMapAfterUpdate: map[string]*v1.Pod{
util.GetPodFullName(pods[0]): updatedPods[0],
util.GetPodFullName(pods[1]): pods[1],
util.GetPodFullName(pods[2]): pods[2],
util.GetPodFullName(pods[3]): pods[3],
},
podsToDelete: []*v1.Pod{pods[0], pods[1]},
expectedMapAfterDelete: map[string]*v1.Pod{
util.GetPodFullName(pods[2]): pods[2],
util.GetPodFullName(pods[3]): pods[3],
},
},
{
name: "create, update, delete all",
podsToAdd: []*v1.Pod{pods[0], pods[3]},
expectedMapAfterAdd: map[string]*v1.Pod{
util.GetPodFullName(pods[0]): pods[0],
util.GetPodFullName(pods[3]): pods[3],
},
podsToUpdate: []*v1.Pod{updatedPods[3]},
expectedMapAfterUpdate: map[string]*v1.Pod{
util.GetPodFullName(pods[0]): pods[0],
util.GetPodFullName(pods[3]): updatedPods[3],
},
podsToDelete: []*v1.Pod{pods[0], pods[3]},
expectedMapAfterDelete: map[string]*v1.Pod{},
},
{
name: "delete non-existing and existing pods",
podsToAdd: []*v1.Pod{pods[1], pods[2]},
expectedMapAfterAdd: map[string]*v1.Pod{
util.GetPodFullName(pods[1]): pods[1],
util.GetPodFullName(pods[2]): pods[2],
},
podsToUpdate: []*v1.Pod{updatedPods[1]},
expectedMapAfterUpdate: map[string]*v1.Pod{
util.GetPodFullName(pods[1]): updatedPods[1],
util.GetPodFullName(pods[2]): pods[2],
},
podsToDelete: []*v1.Pod{pods[2], pods[3]},
expectedMapAfterDelete: map[string]*v1.Pod{
util.GetPodFullName(pods[1]): updatedPods[1],
},
},
}
for _, test := range tests {
t.Run(test.name, func(t *testing.T) {
upm := newUnschedulablePodsMap()
for _, p := range test.podsToAdd {
upm.addOrUpdate(p)
}
if !reflect.DeepEqual(upm.pods, test.expectedMapAfterAdd) {
t.Errorf("Unexpected map after adding pods. Expected: %v, got: %v",
test.expectedMapAfterAdd, upm.pods)
}
if len(test.podsToUpdate) > 0 {
for _, p := range test.podsToUpdate {
upm.addOrUpdate(p)
}
if !reflect.DeepEqual(upm.pods, test.expectedMapAfterUpdate) {
t.Errorf("Unexpected map after updating pods. Expected: %v, got: %v",
test.expectedMapAfterUpdate, upm.pods)
}
}
for _, p := range test.podsToDelete {
upm.delete(p)
}
if !reflect.DeepEqual(upm.pods, test.expectedMapAfterDelete) {
t.Errorf("Unexpected map after deleting pods. Expected: %v, got: %v",
test.expectedMapAfterDelete, upm.pods)
}
upm.clear()
if len(upm.pods) != 0 {
t.Errorf("Expected the map to be empty, but has %v elements.", len(upm.pods))
}
})
}
}
func TestSchedulingQueue_Close(t *testing.T) {
tests := []struct {
name string
q SchedulingQueue
expectedErr error
}{
{
name: "FIFO close",
q: NewFIFO(),
expectedErr: fmt.Errorf(queueClosed),
},
{
name: "PriorityQueue close",
q: NewPriorityQueue(),
expectedErr: fmt.Errorf(queueClosed),
},
}
for _, test := range tests {
t.Run(test.name, func(t *testing.T) {
wg := sync.WaitGroup{}
wg.Add(1)
go func() {
defer wg.Done()
pod, err := test.q.Pop()
if err.Error() != test.expectedErr.Error() {
t.Errorf("Expected err %q from Pop() if queue is closed, but got %q", test.expectedErr.Error(), err.Error())
}
if pod != nil {
t.Errorf("Expected pod nil from Pop() if queue is closed, but got: %v", pod)
}
}()
test.q.Close()
wg.Wait()
})
}
}