orchestrator

Documentation

Index

• func New(client *provreqclient.ProvisioningRequestClient, classes []ProvisioningClass) *provReqOrchestrator
• type ProvisioningClass
• type WrapperOrchestrator
  • func NewWrapperOrchestrator(provReqOrchestrator scaleup.Orchestrator) *WrapperOrchestrator
  • func (o *WrapperOrchestrator) Initialize(autoscalingContext *context.AutoscalingContext, ...)
  • func (o *WrapperOrchestrator) ScaleUp(unschedulablePods []*apiv1.Pod, nodes []*apiv1.Node, ...) (*status.ScaleUpStatus, errors.AutoscalerError)
  • func (o *WrapperOrchestrator) ScaleUpToNodeGroupMinSize(nodes []*apiv1.Node, nodeInfos map[string]*framework.NodeInfo) (*status.ScaleUpStatus, errors.AutoscalerError)

Functions

func New

func New(client *provreqclient.ProvisioningRequestClient, classes []ProvisioningClass) *provReqOrchestrator

New return new orchestrator.

Types

type ProvisioningClass

type ProvisioningClass interface {
	Provision([]*apiv1.Pod, []*apiv1.Node, []*appsv1.DaemonSet,
		map[string]*framework.NodeInfo) (*status.ScaleUpStatus, ca_errors.AutoscalerError)
	Initialize(*context.AutoscalingContext, *ca_processors.AutoscalingProcessors, *clusterstate.ClusterStateRegistry,
		estimator.EstimatorBuilder, taints.TaintConfig, *scheduling.HintingSimulator)
}

ProvisioningClass is an interface for ProvisioningRequests.

type WrapperOrchestrator

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

WrapperOrchestrator is an orchestrator which wraps Scale Up for ProvisioningRequests and regular pods. Each loop WrapperOrchestrator split out regular and pods from ProvisioningRequest, pick one group that wasn't picked in the last loop and run ScaleUp for it.

func NewWrapperOrchestrator

func NewWrapperOrchestrator(provReqOrchestrator scaleup.Orchestrator) *WrapperOrchestrator

NewWrapperOrchestrator return WrapperOrchestrator

func (*WrapperOrchestrator) Initialize

func (o *WrapperOrchestrator) Initialize(
	autoscalingContext *context.AutoscalingContext,
	processors *ca_processors.AutoscalingProcessors,
	clusterStateRegistry *clusterstate.ClusterStateRegistry,
	estimatorBuilder estimator.EstimatorBuilder,
	taintConfig taints.TaintConfig,
)

Initialize initializes the orchestrator object with required fields.

func (*WrapperOrchestrator) ScaleUp

func (o *WrapperOrchestrator) ScaleUp(
	unschedulablePods []*apiv1.Pod,
	nodes []*apiv1.Node,
	daemonSets []*appsv1.DaemonSet,
	nodeInfos map[string]*framework.NodeInfo,
	allOrNothing bool,
) (*status.ScaleUpStatus, errors.AutoscalerError)

ScaleUp run scaleUp function for regular pods of pods from ProvisioningRequest.

func (*WrapperOrchestrator) ScaleUpToNodeGroupMinSize

func (o *WrapperOrchestrator) ScaleUpToNodeGroupMinSize(
	nodes []*apiv1.Node,
	nodeInfos map[string]*framework.NodeInfo,
) (*status.ScaleUpStatus, errors.AutoscalerError)

ScaleUpToNodeGroupMinSize tries to scale up node groups that have less nodes than the configured min size. The source of truth for the current node group size is the TargetSize queried directly from cloud providers. Returns appropriate status or error if an unexpected error occurred.