mxnet

Documentation

Overview

Package controller provides a Kubernetes controller for a MXJob resource.

Package controller provides a Kubernetes controller for a MXJob resource.

Package controller provides a Kubernetes controller for a MXJob resource.

Package controller provides a Kubernetes controller for a MXJob resource.

Package controller provides a Kubernetes controller for a MXJob resource.

Index

• Variables
• func ContainSchedulerSpec(mxJob *mxv1beta1.MXJob) bool
• func GetPortFromMXJob(mxJob *mxv1beta1.MXJob, rtype mxv1beta1.MXReplicaType) (int32, error)
• func NewUnstructuredMXJobInformer(restConfig *restclientset.Config, namespace string) mxjobinformersv1beta1.MXJobInformer
• type ClusterSpec
• type LabelsSpec
• type MXConfig
• type MXController
  • func NewMXController(mxJobInformer mxjobinformersv1beta1.MXJobInformer, ...) *MXController
  • func (tc *MXController) ControllerName() string
  • func (tc *MXController) GetAPIGroupVersion() schema.GroupVersion
  • func (tc *MXController) GetAPIGroupVersionKind() schema.GroupVersionKind
  • func (tc *MXController) GetGroupNameLabelKey() string
  • func (tc *MXController) GetGroupNameLabelValue() string
  • func (tc *MXController) GetJobFromAPIClient(namespace, name string) (metav1.Object, error)
  • func (tc *MXController) GetJobFromInformerCache(namespace, name string) (metav1.Object, error)
  • func (tc *MXController) GetJobNameLabelKey() string
  • func (tc *MXController) GetJobRoleKey() string
  • func (tc *MXController) GetReplicaIndexLabelKey() string
  • func (tc *MXController) GetReplicaTypeLabelKey() string
  • func (tc *MXController) NewMXJobInformer(mxJobInformerFactory mxjobinformers.SharedInformerFactory) mxjobinformersv1beta1.MXJobInformer
  • func (tc *MXController) Run(threadiness int, stopCh <-chan struct{}) error
• type TaskSpec
• type Url_Port

Variables

var (
	// KeyFunc is the short name to DeletionHandlingMetaNamespaceKeyFunc.
	// IndexerInformer uses a delta queue, therefore for deletes we have to use this
	// key function but it should be just fine for non delete events.
	KeyFunc = cache.DeletionHandlingMetaNamespaceKeyFunc

	// DefaultMXControllerConfiguration is the suggested mxnet-operator configuration for production.
	DefaultMXControllerConfiguration = jobcontroller.JobControllerConfiguration{
		ReconcilerSyncLoopPeriod: metav1.Duration{Duration: 15 * time.Second},
		EnableGangScheduling:     false,
	}
)

Functions

func ContainSchedulerSpec

func ContainSchedulerSpec(mxJob *mxv1beta1.MXJob) bool

func GetPortFromMXJob

func GetPortFromMXJob(mxJob *mxv1beta1.MXJob, rtype mxv1beta1.MXReplicaType) (int32, error)

GetPortFromMXJob gets the port of mxnet container.

func NewUnstructuredMXJobInformer

func NewUnstructuredMXJobInformer(restConfig *restclientset.Config, namespace string) mxjobinformersv1beta1.MXJobInformer

Types

type ClusterSpec

type ClusterSpec map[string][]Url_Port

ClusterSpec represents a cluster Mxnet specification.

type LabelsSpec

type LabelsSpec map[string]string

LabelsSpec represents a label specification.

type MXConfig

type MXConfig struct {
	// Cluster represents a Mxnet ClusterSpec.
	Cluster ClusterSpec `json:"cluster"`
	// Labels include all label of task.
	Labels LabelsSpec `json:"labels"`
	// Task include information of current node.
	Task TaskSpec `json:"task"`
}

MXConfig is a struct representing the distributed Mxnet config. This struct is turned into an environment variable MX_CONFIG which is used by Mxnet processes to configure themselves.

type MXController

type MXController struct {
	jobcontroller.JobController
	// contains filtered or unexported fields
}

MXController is the type for MXJob Controller, which manages the lifecycle of MXJobs.

func NewMXController

func NewMXController(

	mxJobInformer mxjobinformersv1beta1.MXJobInformer,
	kubeClientSet kubeclientset.Interface,
	mxJobClientSet mxjobclientset.Interface,
	kubeBatchClientSet kubebatchclient.Interface,
	kubeInformerFactory kubeinformers.SharedInformerFactory,

	mxJobInformerFactory mxjobinformers.SharedInformerFactory,
	option options.ServerOption) *MXController

NewMXController returns a new MXJob controller.

func (*MXController) ControllerName

func (tc *MXController) ControllerName() string

func (*MXController) GetAPIGroupVersion

func (tc *MXController) GetAPIGroupVersion() schema.GroupVersion

func (*MXController) GetAPIGroupVersionKind

func (tc *MXController) GetAPIGroupVersionKind() schema.GroupVersionKind

func (*MXController) GetGroupNameLabelKey

func (tc *MXController) GetGroupNameLabelKey() string

func (*MXController) GetGroupNameLabelValue

func (tc *MXController) GetGroupNameLabelValue() string

func (*MXController) GetJobFromAPIClient

func (tc *MXController) GetJobFromAPIClient(namespace, name string) (metav1.Object, error)

func (*MXController) GetJobFromInformerCache

func (tc *MXController) GetJobFromInformerCache(namespace, name string) (metav1.Object, error)

func (*MXController) GetJobNameLabelKey

func (tc *MXController) GetJobNameLabelKey() string

func (*MXController) GetJobRoleKey

func (tc *MXController) GetJobRoleKey() string

func (*MXController) GetReplicaIndexLabelKey

func (tc *MXController) GetReplicaIndexLabelKey() string

func (*MXController) GetReplicaTypeLabelKey

func (tc *MXController) GetReplicaTypeLabelKey() string

func (*MXController) NewMXJobInformer

func (tc *MXController) NewMXJobInformer(mxJobInformerFactory mxjobinformers.SharedInformerFactory) mxjobinformersv1beta1.MXJobInformer

NewMXJobInformer returns MXJobInformer from the given factory.

func (*MXController) Run

func (tc *MXController) Run(threadiness int, stopCh <-chan struct{}) error

Run will set up the event handlers for types we are interested in, as well as syncing informer caches and starting workers. It will block until stopCh is closed, at which point it will shutdown the workqueue and wait for workers to finish processing their current work items.

type TaskSpec

type TaskSpec struct {
	Type  string `json:"type"`
	Index int    `json:"index"`
}

TaskSpec is the specification for a task (server or worker ...) of the MXJob.

type Url_Port

type Url_Port struct {
	Url  string `json:"url"`
	Port int    `json:"port"`
}