func CalculateNodePreferAvoidPodsPriorityMap(pod *v1.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (schedulerapi.HostPriority, error) {
node := nodeInfo.Node()
if node == nil {
return schedulerapi.HostPriority{}, fmt.Errorf("node not found")
}
controllerRef := priorityutil.GetControllerRef(pod)
if controllerRef != nil {
// Ignore pods that are owned by other controller than ReplicationController
// or ReplicaSet.
if controllerRef.Kind != "ReplicationController" && controllerRef.Kind != "ReplicaSet" {
controllerRef = nil
}
}
if controllerRef == nil {
return schedulerapi.HostPriority{Host: node.Name, Score: 10}, nil
}
avoids, err := v1.GetAvoidPodsFromNodeAnnotations(node.Annotations)
if err != nil {
// If we cannot get annotation, assume it's schedulable there.
return schedulerapi.HostPriority{Host: node.Name, Score: 10}, nil
}
for i := range avoids.PreferAvoidPods {
avoid := &avoids.PreferAvoidPods[i]
if controllerRef != nil {
if avoid.PodSignature.PodController.Kind == controllerRef.Kind && avoid.PodSignature.PodController.UID == controllerRef.UID {
return schedulerapi.HostPriority{Host: node.Name, Score: 0}, nil
}
}
}
return schedulerapi.HostPriority{Host: node.Name, Score: 10}, nil
}
// ComputeTaintTolerationPriority prepares the priority list for all the nodes based on the number of intolerable taints on the node
func ComputeTaintTolerationPriorityMap(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (schedulerapi.HostPriority, error) {
node := nodeInfo.Node()
if node == nil {
return schedulerapi.HostPriority{}, fmt.Errorf("node not found")
}
var tolerationList []api.Toleration
if priorityMeta, ok := meta.(*priorityMetadata); ok {
tolerationList = priorityMeta.podTolerations
} else {
var err error
tolerationList, err = getTolerationListFromPod(pod)
if err != nil {
return schedulerapi.HostPriority{}, err
}
}
taints, err := api.GetTaintsFromNodeAnnotations(node.Annotations)
if err != nil {
return schedulerapi.HostPriority{}, err
}
return schedulerapi.HostPriority{
Host: node.Name,
Score: countIntolerableTaintsPreferNoSchedule(taints, tolerationList),
}, nil
}
// CheckNodeDiskPressurePredicate checks if a pod can be scheduled on a node
// reporting disk pressure condition.
func CheckNodeDiskPressurePredicate(pod *v1.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
// is node under presure?
if nodeInfo.DiskPressureCondition() == v1.ConditionTrue {
return false, []algorithm.PredicateFailureReason{ErrNodeUnderDiskPressure}, nil
}
return true, nil, nil
}
// CheckNodeMemoryPressurePredicate checks if a pod can be scheduled on a node
// reporting memory pressure condition.
func CheckNodeMemoryPressurePredicate(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, error) {
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found")
}
var podBestEffort bool
predicateMeta, ok := meta.(*predicateMetadata)
if ok {
podBestEffort = predicateMeta.podBestEffort
} else {
// We couldn't parse metadata - fallback to computing it.
podBestEffort = isPodBestEffort(pod)
}
// pod is not BestEffort pod
if !podBestEffort {
return true, nil
}
// is node under presure?
for _, cond := range node.Status.Conditions {
if cond.Type == api.NodeMemoryPressure && cond.Status == api.ConditionTrue {
return false, ErrNodeUnderMemoryPressure
}
}
return true, nil
}
func (c *MaxPDVolumeCountChecker) predicate(pod *api.Pod, nodeName string, nodeInfo *schedulercache.NodeInfo) (bool, error) {
newVolumes := make(map[string]bool)
if err := c.filterVolumes(pod.Spec.Volumes, pod.Namespace, newVolumes); err != nil {
return false, err
}
// quick return
if len(newVolumes) == 0 {
return true, nil
}
// count unique volumes
existingVolumes := make(map[string]bool)
for _, existingPod := range nodeInfo.Pods() {
if err := c.filterVolumes(existingPod.Spec.Volumes, existingPod.Namespace, existingVolumes); err != nil {
return false, err
}
}
numExistingVolumes := len(existingVolumes)
// filter out already-mounted volumes
for k := range existingVolumes {
if _, ok := newVolumes[k]; ok {
delete(newVolumes, k)
}
}
numNewVolumes := len(newVolumes)
if numExistingVolumes+numNewVolumes > c.maxVolumes {
return false, nil
}
return true, nil
}
// Calculates score for all pods and returns podInfo structure.
// Score is defined as cpu_sum/node_capacity + mem_sum/node_capacity.
// Pods that have bigger requirements should be processed first, thus have higher scores.
func calculatePodScore(pods []*apiv1.Pod, nodeTemplate *schedulercache.NodeInfo) []*podInfo {
podInfos := make([]*podInfo, 0, len(pods))
for _, pod := range pods {
cpuSum := resource.Quantity{}
memorySum := resource.Quantity{}
for _, container := range pod.Spec.Containers {
if request, ok := container.Resources.Requests[apiv1.ResourceCPU]; ok {
cpuSum.Add(request)
}
if request, ok := container.Resources.Requests[apiv1.ResourceMemory]; ok {
memorySum.Add(request)
}
}
score := float64(0)
if cpuAllocatable, ok := nodeTemplate.Node().Status.Allocatable[apiv1.ResourceCPU]; ok && cpuAllocatable.MilliValue() > 0 {
score += float64(cpuSum.MilliValue()) / float64(cpuAllocatable.MilliValue())
}
if memAllocatable, ok := nodeTemplate.Node().Status.Allocatable[apiv1.ResourceMemory]; ok && memAllocatable.Value() > 0 {
score += float64(memorySum.Value()) / float64(memAllocatable.Value())
}
podInfos = append(podInfos, &podInfo{
score: score,
pod: pod,
})
}
return podInfos
}
func (checker *PodAffinityChecker) InterPodAffinityMatches(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, error) {
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found")
}
allPods, err := checker.podLister.List(labels.Everything())
if err != nil {
return false, err
}
affinity, err := api.GetAffinityFromPodAnnotations(pod.Annotations)
if err != nil {
return false, err
}
// Check if the current node match the inter-pod affinity scheduling constraints.
// Hard inter-pod affinity is not symmetric, check only when affinity.PodAffinity exists.
if affinity.PodAffinity != nil {
if !checker.NodeMatchesHardPodAffinity(pod, allPods, node, affinity.PodAffinity) {
return false, ErrPodAffinityNotMatch
}
}
// Hard inter-pod anti-affinity is symmetric, we should always check it.
if !checker.NodeMatchesHardPodAntiAffinity(pod, allPods, node, affinity.PodAntiAffinity) {
return false, ErrPodAffinityNotMatch
}
return true, nil
}
func (c *PodAffinityChecker) InterPodAffinityMatches(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
node := nodeInfo.Node()
if node == nil {
return false, nil, fmt.Errorf("node not found")
}
if !c.satisfiesExistingPodsAntiAffinity(pod, meta, node) {
return false, []algorithm.PredicateFailureReason{ErrPodAffinityNotMatch}, nil
}
// Now check if <pod> requirements will be satisfied on this node.
affinity, err := api.GetAffinityFromPodAnnotations(pod.Annotations)
if err != nil {
return false, nil, err
}
if affinity == nil || (affinity.PodAffinity == nil && affinity.PodAntiAffinity == nil) {
return true, nil, nil
}
if !c.satisfiesPodsAffinityAntiAffinity(pod, node, affinity) {
return false, []algorithm.PredicateFailureReason{ErrPodAffinityNotMatch}, nil
}
if glog.V(10) {
// We explicitly don't do glog.V(10).Infof() to avoid computing all the parameters if this is
// not logged. There is visible performance gain from it.
glog.Infof("Schedule Pod %+v on Node %+v is allowed, pod (anti)affinity constraints satisfied",
podName(pod), node.Name)
}
return true, nil, nil
}
// Calculate the resource occupancy on a node. 'node' has information about the resources on the node.
// 'pods' is a list of pods currently scheduled on the node.
func calculateResourceOccupancy(pod *api.Pod, node api.Node, nodeInfo *schedulercache.NodeInfo) schedulerapi.HostPriority {
totalMilliCPU := nodeInfo.NonZeroRequest().MilliCPU
totalMemory := nodeInfo.NonZeroRequest().Memory
capacityMilliCPU := node.Status.Allocatable.Cpu().MilliValue()
capacityMemory := node.Status.Allocatable.Memory().Value()
// Add the resources requested by the current pod being scheduled.
// This also helps differentiate between differently sized, but empty, nodes.
for _, container := range pod.Spec.Containers {
cpu, memory := priorityutil.GetNonzeroRequests(&container.Resources.Requests)
totalMilliCPU += cpu
totalMemory += memory
}
cpuScore := calculateScore(totalMilliCPU, capacityMilliCPU, node.Name)
memoryScore := calculateScore(totalMemory, capacityMemory, node.Name)
glog.V(10).Infof(
"%v -> %v: Least Requested Priority, capacity %d millicores %d memory bytes, total request %d millicores %d memory bytes, score %d CPU %d memory",
pod.Name, node.Name,
capacityMilliCPU, capacityMemory,
totalMilliCPU, totalMemory,
cpuScore, memoryScore,
)
return schedulerapi.HostPriority{
Host: node.Name,
Score: int((cpuScore + memoryScore) / 2),
}
}
// FastGetPodsToMove returns a list of pods that should be moved elsewhere if the node
// is drained. Raises error if there is an unreplicated pod and force option was not specified.
// Based on kubectl drain code. It makes an assumption that RC, DS, Jobs and RS were deleted
// along with their pods (no abandoned pods with dangling created-by annotation). Usefull for fast
// checks. Doesn't check i
func FastGetPodsToMove(nodeInfo *schedulercache.NodeInfo, skipNodesWithSystemPods bool, skipNodesWithLocalStorage bool) ([]*api.Pod, error) {
return drain.GetPodsForDeletionOnNodeDrain(
nodeInfo.Pods(),
api.Codecs.UniversalDecoder(),
skipNodesWithSystemPods,
skipNodesWithLocalStorage,
false,
nil,
0)
}
// NoDiskConflict evaluates if a pod can fit due to the volumes it requests, and those that
// are already mounted. If there is already a volume mounted on that node, another pod that uses the same volume
// can't be scheduled there.
// This is GCE, Amazon EBS, and Ceph RBD specific for now:
// - GCE PD allows multiple mounts as long as they're all read-only
// - AWS EBS forbids any two pods mounting the same volume ID
// - Ceph RBD forbids if any two pods share at least same monitor, and match pool and image.
// TODO: migrate this into some per-volume specific code?
func NoDiskConflict(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, error) {
for _, v := range pod.Spec.Volumes {
for _, ev := range nodeInfo.Pods() {
if isVolumeConflict(v, ev) {
return false, ErrDiskConflict
}
}
}
return true, nil
}
func PodSelectorMatches(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, error) {
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found")
}
if podMatchesNodeLabels(pod, node) {
return true, nil
}
return false, ErrNodeSelectorNotMatch
}
// NoDiskConflict evaluates if a pod can fit due to the volumes it requests, and those that
// are already mounted. If there is already a volume mounted on that node, another pod that uses the same volume
// can't be scheduled there.
// This is GCE, Amazon EBS, and Ceph RBD specific for now:
// - GCE PD allows multiple mounts as long as they're all read-only
// - AWS EBS forbids any two pods mounting the same volume ID
// - Ceph RBD forbids if any two pods share at least same monitor, and match pool and image.
// TODO: migrate this into some per-volume specific code?
func NoDiskConflict(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
for _, v := range pod.Spec.Volumes {
for _, ev := range nodeInfo.Pods() {
if isVolumeConflict(v, ev) {
return false, []algorithm.PredicateFailureReason{ErrDiskConflict}, nil
}
}
}
return true, nil, nil
}
// NoDiskConflict evaluates if a pod can fit due to the volumes it requests, and those that
// are already mounted. If there is already a volume mounted on that node, another pod that uses the same volume
// can't be scheduled there.
// This is GCE, Amazon EBS, and Ceph RBD specific for now:
// - GCE PD allows multiple mounts as long as they're all read-only
// - AWS EBS forbids any two pods mounting the same volume ID
// - Ceph RBD forbids if any two pods share at least same monitor, and match pool and image.
// TODO: migrate this into some per-volume specific code?
func NoDiskConflict(pod *api.Pod, nodeName string, nodeInfo *schedulercache.NodeInfo) (bool, error) {
for _, v := range pod.Spec.Volumes {
for _, ev := range nodeInfo.Pods() {
if isVolumeConflict(v, ev) {
return false, nil
}
}
}
return true, nil
}
func matchesPredicate(pod *api.Pod, nodeInfo *schedulercache.NodeInfo) (bool, error) {
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found")
}
if pod.Name == node.Name {
return true, nil
}
return false, algorithmpredicates.ErrFakePredicate
}
func PodSelectorMatches(pod *api.Pod, nodeName string, nodeInfo *schedulercache.NodeInfo) (bool, error) {
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found: %q", nodeName)
}
if PodMatchesNodeLabels(pod, node) {
return true, nil
}
return false, ErrNodeSelectorNotMatch
}
func PodSelectorMatches(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
node := nodeInfo.Node()
if node == nil {
return false, nil, fmt.Errorf("node not found")
}
if podMatchesNodeLabels(pod, node) {
return true, nil, nil
}
return false, []algorithm.PredicateFailureReason{ErrNodeSelectorNotMatch}, nil
}
func matchesPredicate(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
node := nodeInfo.Node()
if node == nil {
return false, nil, fmt.Errorf("node not found")
}
if pod.Name == node.Name {
return true, nil, nil
}
return false, []algorithm.PredicateFailureReason{algorithmpredicates.ErrFakePredicate}, nil
}
// FastGetPodsToMove returns a list of pods that should be moved elsewhere if the node
// is drained. Raises error if there is an unreplicated pod and force option was not specified.
// Based on kubectl drain code. It makes an assumption that RC, DS, Jobs and RS were deleted
// along with their pods (no abandoned pods with dangling created-by annotation). Usefull for fast
// checks.
func FastGetPodsToMove(nodeInfo *schedulercache.NodeInfo, force bool,
skipNodesWithSystemPods bool, skipNodesWithLocalStorage bool, decoder runtime.Decoder) ([]*api.Pod, error) {
pods := make([]*api.Pod, 0)
unreplicatedPodNames := []string{}
for _, pod := range nodeInfo.Pods() {
_, found := pod.ObjectMeta.Annotations[types.ConfigMirrorAnnotationKey]
if found {
// Skip mirror pod
continue
}
replicated := false
daemonsetPod := false
creatorRef, found := pod.ObjectMeta.Annotations[controller.CreatedByAnnotation]
if found {
var sr api.SerializedReference
if err := runtime.DecodeInto(decoder, []byte(creatorRef), &sr); err != nil {
return []*api.Pod{}, err
}
if sr.Reference.Kind == "ReplicationController" {
replicated = true
} else if sr.Reference.Kind == "DaemonSet" {
daemonsetPod = true
} else if sr.Reference.Kind == "Job" {
replicated = true
} else if sr.Reference.Kind == "ReplicaSet" {
replicated = true
}
}
if !daemonsetPod && pod.Namespace == "kube-system" && skipNodesWithSystemPods {
return []*api.Pod{}, fmt.Errorf("non-deamons set, non-mirrored, kube-system pod present: %s", pod.Name)
}
if !daemonsetPod && hasLocalStorage(pod) && skipNodesWithLocalStorage {
return []*api.Pod{}, fmt.Errorf("pod with local storage present: %s", pod.Name)
}
switch {
case daemonsetPod:
break
case !replicated:
unreplicatedPodNames = append(unreplicatedPodNames, pod.Name)
if force {
pods = append(pods, pod)
}
default:
pods = append(pods, pod)
}
}
if !force && len(unreplicatedPodNames) > 0 {
return []*api.Pod{}, fmt.Errorf("unreplicated pods present")
}
return pods, nil
}
// DetailedGetPodsForMove returns a list of pods that should be moved elsewhere if the node
// is drained. Raises error if there is an unreplicated pod and force option was not specified.
// Based on kubectl drain code. It checks whether RC, DS, Jobs and RS that created these pods
// still exist.
func DetailedGetPodsForMove(nodeInfo *schedulercache.NodeInfo, skipNodesWithSystemPods bool,
skipNodesWithLocalStorage bool, client *unversionedclient.Client, minReplicaCount int32) ([]*api.Pod, error) {
return drain.GetPodsForDeletionOnNodeDrain(
nodeInfo.Pods(),
api.Codecs.UniversalDecoder(),
skipNodesWithSystemPods,
skipNodesWithLocalStorage,
true,
client,
minReplicaCount)
}
func PodFitsHost(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, error) {
if len(pod.Spec.NodeName) == 0 {
return true, nil
}
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found")
}
if pod.Spec.NodeName == node.Name {
return true, nil
}
return false, ErrPodNotMatchHostName
}
func PodFitsHost(pod *v1.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
if len(pod.Spec.NodeName) == 0 {
return true, nil, nil
}
node := nodeInfo.Node()
if node == nil {
return false, nil, fmt.Errorf("node not found")
}
if pod.Spec.NodeName == node.Name {
return true, nil, nil
}
return false, []algorithm.PredicateFailureReason{ErrPodNotMatchHostName}, nil
}
// FastGetPodsToMove returns a list of pods that should be moved elsewhere if the node
// is drained. Raises error if there is an unreplicated pod and force option was not specified.
// Based on kubectl drain code. It makes an assumption that RC, DS, Jobs and RS were deleted
// along with their pods (no abandoned pods with dangling created-by annotation). Usefull for fast
// checks.
func FastGetPodsToMove(nodeInfo *schedulercache.NodeInfo, force bool,
skipNodesWithSystemPods bool, skipNodesWithLocalStorage bool) ([]*api.Pod, error) {
pods := make([]*api.Pod, 0)
unreplicatedPodNames := []string{}
for _, pod := range nodeInfo.Pods() {
if IsMirrorPod(pod) {
continue
}
replicated := false
daemonsetPod := false
creatorKind, err := CreatorRefKind(pod)
if err != nil {
return []*api.Pod{}, err
}
if creatorKind == "ReplicationController" {
replicated = true
} else if creatorKind == "DaemonSet" {
daemonsetPod = true
} else if creatorKind == "Job" {
replicated = true
} else if creatorKind == "ReplicaSet" {
replicated = true
}
if !daemonsetPod && pod.Namespace == "kube-system" && skipNodesWithSystemPods {
return []*api.Pod{}, fmt.Errorf("non-deamons set, non-mirrored, kube-system pod present: %s", pod.Name)
}
if !daemonsetPod && hasLocalStorage(pod) && skipNodesWithLocalStorage {
return []*api.Pod{}, fmt.Errorf("pod with local storage present: %s", pod.Name)
}
switch {
case daemonsetPod:
break
case !replicated:
unreplicatedPodNames = append(unreplicatedPodNames, pod.Name)
if force {
pods = append(pods, pod)
}
default:
pods = append(pods, pod)
}
}
if !force && len(unreplicatedPodNames) > 0 {
return []*api.Pod{}, fmt.Errorf("unreplicated pods present")
}
return pods, nil
}
func (r *NodeStatus) PodFitsResources(pod *api.Pod, nodeName string, nodeInfo *schedulercache.NodeInfo) (bool, error) {
info, err := r.info.GetNodeInfo(nodeName)
if err != nil {
return false, err
}
// TODO: move the following podNumber check to podFitsResourcesInternal when Kubelet allows podNumber check (See #20263).
allocatable := info.Status.Allocatable
allowedPodNumber := allocatable.Pods().Value()
if int64(len(nodeInfo.Pods()))+1 > allowedPodNumber {
return false,
newInsufficientResourceError(podCountResourceName, 1, int64(len(nodeInfo.Pods())), allowedPodNumber)
}
return podFitsResourcesInternal(pod, nodeName, nodeInfo, info)
}
// Calculate the resource used on a node. 'node' has information about the resources on the node.
// 'pods' is a list of pods currently scheduled on the node.
func calculateUsedPriority(pod *api.Pod, podRequests *schedulercache.Resource, nodeInfo *schedulercache.NodeInfo) (schedulerapi.HostPriority, error) {
node := nodeInfo.Node()
if node == nil {
return schedulerapi.HostPriority{}, fmt.Errorf("node not found")
}
allocatableResources := nodeInfo.AllocatableResource()
totalResources := *podRequests
totalResources.MilliCPU += nodeInfo.NonZeroRequest().MilliCPU
totalResources.Memory += nodeInfo.NonZeroRequest().Memory
cpuScore := calculateUsedScore(totalResources.MilliCPU, allocatableResources.MilliCPU, node.Name)
memoryScore := calculateUsedScore(totalResources.Memory, allocatableResources.Memory, node.Name)
if glog.V(10) {
// We explicitly don't do glog.V(10).Infof() to avoid computing all the parameters if this is
// not logged. There is visible performance gain from it.
glog.V(10).Infof(
"%v -> %v: Most Requested Priority, capacity %d millicores %d memory bytes, total request %d millicores %d memory bytes, score %d CPU %d memory",
pod.Name, node.Name,
allocatableResources.MilliCPU, allocatableResources.Memory,
totalResources.MilliCPU, totalResources.Memory,
cpuScore, memoryScore,
)
}
return schedulerapi.HostPriority{
Host: node.Name,
Score: int((cpuScore + memoryScore) / 2),
}, nil
}
func (checker *PodAffinityChecker) InterPodAffinityMatches(pod *api.Pod, nodeInfo *schedulercache.NodeInfo) (bool, error) {
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found")
}
allPods, err := checker.podLister.List(labels.Everything())
if err != nil {
return false, err
}
if checker.NodeMatchPodAffinityAntiAffinity(pod, allPods, node) {
return true, nil
}
return false, ErrPodAffinityNotMatch
}
// ImageLocalityPriority is a priority function that favors nodes that already have requested pod container's images.
// It will detect whether the requested images are present on a node, and then calculate a score ranging from 0 to 10
// based on the total size of those images.
// - If none of the images are present, this node will be given the lowest priority.
// - If some of the images are present on a node, the larger their sizes' sum, the higher the node's priority.
func ImageLocalityPriorityMap(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (schedulerapi.HostPriority, error) {
node := nodeInfo.Node()
if node == nil {
return schedulerapi.HostPriority{}, fmt.Errorf("node not found")
}
var sumSize int64
for i := range pod.Spec.Containers {
sumSize += checkContainerImageOnNode(node, &pod.Spec.Containers[i])
}
return schedulerapi.HostPriority{
Host: node.Name,
Score: calculateScoreFromSize(sumSize),
}, nil
}
// CheckPredicates Checks if the given pod can be placed on the given node.
func (p *PredicateChecker) CheckPredicates(pod *kube_api.Pod, nodeInfo *schedulercache.NodeInfo) error {
// TODO(fgrzadkowski): Use full list of predicates.
match, err := predicates.GeneralPredicates(pod, nodeInfo)
nodename := "unknown"
if nodeInfo.Node() != nil {
nodename = nodeInfo.Node().Name
}
if err != nil {
return fmt.Errorf("cannot put %s on %s due to %v", pod.Name, nodename, err)
}
if !match {
return fmt.Errorf("cannot put %s on %s", pod.Name, nodename)
}
return nil
}
// CheckNodeInodePressurePredicate checks if a pod can be scheduled on a node
// reporting inode pressure condition.
func CheckNodeInodePressurePredicate(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
node := nodeInfo.Node()
if node == nil {
return false, nil, fmt.Errorf("node not found")
}
// is node under presure?
for _, cond := range node.Status.Conditions {
if cond.Type == api.NodeInodePressure && cond.Status == api.ConditionTrue {
return false, []algorithm.PredicateFailureReason{ErrNodeUnderInodePressure}, nil
}
}
return true, nil, nil
}
// CheckNodeLabelPresence checks whether all of the specified labels exists on a node or not, regardless of their value
// If "presence" is false, then returns false if any of the requested labels matches any of the node's labels,
// otherwise returns true.
// If "presence" is true, then returns false if any of the requested labels does not match any of the node's labels,
// otherwise returns true.
//
// Consider the cases where the nodes are placed in regions/zones/racks and these are identified by labels
// In some cases, it is required that only nodes that are part of ANY of the defined regions/zones/racks be selected
//
// Alternately, eliminating nodes that have a certain label, regardless of value, is also useful
// A node may have a label with "retiring" as key and the date as the value
// and it may be desirable to avoid scheduling new pods on this node
func (n *NodeLabelChecker) CheckNodeLabelPresence(pod *api.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, error) {
node := nodeInfo.Node()
if node == nil {
return false, fmt.Errorf("node not found")
}
var exists bool
nodeLabels := labels.Set(node.Labels)
for _, label := range n.labels {
exists = nodeLabels.Has(label)
if (exists && !n.presence) || (!exists && n.presence) {
return false, ErrNodeLabelPresenceViolated
}
}
return true, nil
}