tensorflow

Documentation

Overview

Package tensorflow provides a Kubernetes controller for a TFJob resource.

Package controller provides a Kubernetes controller for a TFJob resource.

Package tensorflow provides a Kubernetes controller for a TFJob resource.

Index

• Constants
• func OnOwnerDeleteAndDeletionExpectationFunc(jc job_controller.JobController) func(e event.DeleteEvent) bool
• type ClusterSpec
• type TFConfig
• type TFJobReconciler
  • func NewReconciler(mgr ctrl.Manager, config options.JobControllerConfiguration) *TFJobReconciler
  • func (r *TFJobReconciler) ControllerName() string
  • func (r *TFJobReconciler) DeleteJob(job interface{}) error
  • func (r *TFJobReconciler) GetAPIGroupVersion() schema.GroupVersion
  • func (r *TFJobReconciler) GetAPIGroupVersionKind() schema.GroupVersionKind
  • func (r *TFJobReconciler) GetDefaultContainerName() string
  • func (r *TFJobReconciler) GetDefaultContainerPortName() string
  • func (r *TFJobReconciler) GetDefaultContainerPortNumber() int32
  • func (r *TFJobReconciler) GetGroupNameLabelValue() string
  • func (r *TFJobReconciler) GetJobFromAPIClient(namespace, name string) (metav1.Object, error)
  • func (r *TFJobReconciler) GetJobFromInformerCache(namespace, name string) (metav1.Object, error)
  • func (r *TFJobReconciler) GetNodeForModelOutput(pods []*corev1.Pod) string
  • func (r *TFJobReconciler) GetPodsForJob(obj interface{}) ([]*corev1.Pod, error)
  • func (r *TFJobReconciler) GetReconcileOrders() []v1.ReplicaType
  • func (r *TFJobReconciler) GetServicesForJob(obj interface{}) ([]*corev1.Service, error)
  • func (r *TFJobReconciler) IsMasterRole(replicas map[v1.ReplicaType]*v1.ReplicaSpec, rtype v1.ReplicaType, index int) bool
  • func (r *TFJobReconciler) Reconcile(_ context.Context, req ctrl.Request) (ctrl.Result, error)
  • func (r *TFJobReconciler) SetClusterSpec(ctx context.Context, job interface{}, podTemplateSpec *corev1.PodTemplateSpec, ...) error
  • func (r *TFJobReconciler) SetupWithManager(mgr ctrl.Manager) error
  • func (r *TFJobReconciler) UpdateJobStatus(job interface{}, replicas map[v1.ReplicaType]*v1.ReplicaSpec, ...) error
  • func (r *TFJobReconciler) UpdateJobStatusInApiServer(job interface{}, jobStatus *v1.JobStatus) error
• type TaskSpec

Constants

const (
	// EnvCustomClusterDomain is the custom defined cluster domain, such as "svc.cluster.local".
	// Ref: https://kubernetes.io/docs/concepts/services-networking/dns-pod-service/#a-records
	EnvCustomClusterDomain = "CUSTOM_CLUSTER_DOMAIN"
)

Functions

func OnOwnerDeleteAndDeletionExpectationFunc

func OnOwnerDeleteAndDeletionExpectationFunc(jc job_controller.JobController) func(e event.DeleteEvent) bool

Types

type ClusterSpec

type ClusterSpec map[string][]string

ClusterSpec represents a cluster TensorFlow specification. https://www.tensorflow.org/deploy/distributed#create_a_tftrainclusterspec_to_describe_the_cluster It is a map from job names to network addresses.

type TFConfig

type TFConfig struct {
	// Cluster represents a TensorFlow ClusterSpec.
	// See: https://www.tensorflow.org/api_docs/python/tf/train/ClusterSpec
	Cluster ClusterSpec `json:"cluster"`
	Task    TaskSpec    `json:"task"`
	// Environment is used by tensorflow.contrib.learn.python.learn in versions <= 1.3
	// TODO(jlewi): I don't think it is used in versions TF >- 1.4. So we can eventually get rid of it.
	Environment string `json:"environment"`
}

TFConfig is a struct representing the distributed TensorFlow config. This struct is turned into an environment variable TF_CONFIG which is used by TensorFlow processes to configure themselves. https://www.tensorflow.org/api_docs/python/tf/estimator/RunConfig#methods https://cloud.google.com/ml-engine/docs/tensorflow/distributed-training-details

type TFJobReconciler

type TFJobReconciler struct {
	client.Client

	utilruntime.EmptyScaleImpl
	// contains filtered or unexported fields
}

TFJobReconciler reconciles a TFJob object

func NewReconciler

func NewReconciler(mgr ctrl.Manager, config options.JobControllerConfiguration) *TFJobReconciler

NewReconciler returns a new reconcile.Reconciler

func (*TFJobReconciler) ControllerName

func (r *TFJobReconciler) ControllerName() string

ControllerName returns the Controller name

func (*TFJobReconciler) DeleteJob

func (r *TFJobReconciler) DeleteJob(job interface{}) error

DeleteJob deletes the job

func (*TFJobReconciler) GetAPIGroupVersion

func (r *TFJobReconciler) GetAPIGroupVersion() schema.GroupVersion

GetAPIGroupVersion returns the GroupVersion of the API

func (*TFJobReconciler) GetAPIGroupVersionKind

func (r *TFJobReconciler) GetAPIGroupVersionKind() schema.GroupVersionKind

GetAPIGroupVersionKind returns the GroupVersionKind of the API

func (*TFJobReconciler) GetDefaultContainerName

func (r *TFJobReconciler) GetDefaultContainerName() string

GetDefaultContainerName returns the default container name in pod

func (*TFJobReconciler) GetDefaultContainerPortName

func (r *TFJobReconciler) GetDefaultContainerPortName() string

GetDefaultContainerPortName Get the default container port name

func (*TFJobReconciler) GetDefaultContainerPortNumber

func (r *TFJobReconciler) GetDefaultContainerPortNumber() int32

GetDefaultContainerPortNumber get the default container port number

func (*TFJobReconciler) GetGroupNameLabelValue

func (r *TFJobReconciler) GetGroupNameLabelValue() string

GetGroupNameLabelValue returns the Group Name(value) in the labels of the job

func (*TFJobReconciler) GetJobFromAPIClient

func (r *TFJobReconciler) GetJobFromAPIClient(namespace, name string) (metav1.Object, error)

GetJobFromAPIClient returns the Job from API server

func (*TFJobReconciler) GetJobFromInformerCache

func (r *TFJobReconciler) GetJobFromInformerCache(namespace, name string) (metav1.Object, error)

GetJobFromInformerCache returns the Job from Informer Cache

func (*TFJobReconciler) GetNodeForModelOutput

func (r *TFJobReconciler) GetNodeForModelOutput(pods []*corev1.Pod) string

func (*TFJobReconciler) GetPodsForJob

func (r *TFJobReconciler) GetPodsForJob(obj interface{}) ([]*corev1.Pod, error)

GetPodsForJob returns the set of pods that this job should manage.

func (*TFJobReconciler) GetReconcileOrders

func (r *TFJobReconciler) GetReconcileOrders() []v1.ReplicaType

Get replicas reconcile orders so that replica type with higher priority can be created earlier.

func (*TFJobReconciler) GetServicesForJob

func (r *TFJobReconciler) GetServicesForJob(obj interface{}) ([]*corev1.Service, error)

GetServicesForJob returns the services managed by the job. This can be achieved by selecting services using label key "job-name" i.e. all services created by the job will come with label "job-name" = <this_job_name>

func (*TFJobReconciler) IsMasterRole

func (r *TFJobReconciler) IsMasterRole(replicas map[v1.ReplicaType]*v1.ReplicaSpec, rtype v1.ReplicaType, index int) bool

IsMasterRole returns if this replica type with index specified is a master role. MasterRole pod will have "job-role=master" set in its label

func (*TFJobReconciler) Reconcile

func (r *TFJobReconciler) Reconcile(_ context.Context, req ctrl.Request) (ctrl.Result, error)

func (*TFJobReconciler) SetClusterSpec

func (r *TFJobReconciler) SetClusterSpec(ctx context.Context, job interface{}, podTemplateSpec *corev1.PodTemplateSpec, rt, index string) error

SetClusterSpec generates and sets TF_CONFIG for the given podTemplateSpec.

func (*TFJobReconciler) SetupWithManager

func (r *TFJobReconciler) SetupWithManager(mgr ctrl.Manager) error

SetupWithManager setup reconciler to the Manager with default RBAC. The Manager will set fields on the Controller and Start it when the Manager is Started.

func (*TFJobReconciler) UpdateJobStatus

func (r *TFJobReconciler) UpdateJobStatus(job interface{}, replicas map[v1.ReplicaType]*v1.ReplicaSpec, jobStatus *v1.JobStatus, restart bool) error

UpdateJobStatus updates the job status and job conditions

func (*TFJobReconciler) UpdateJobStatusInApiServer

func (r *TFJobReconciler) UpdateJobStatusInApiServer(job interface{}, jobStatus *v1.JobStatus) error

UpdateJobStatusInApiServer updates the job status in API server

type TaskSpec

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

TaskSpec is the specification for a task (PS or worker) of the TFJob.