v1

package
v0.4.7 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Apr 25, 2023 License: Apache-2.0 Imports: 7 Imported by: 47

Documentation

Overview

Package v1 is the v1 version of the API. +groupName=kubeflow.org

Index

Constants

View Source
const (

	// ReplicaIndexLabel represents the label key for the replica-index, e.g. 0, 1, 2.. etc
	ReplicaIndexLabel = "training.kubeflow.org/replica-index"

	// ReplicaTypeLabel represents the label key for the replica-type, e.g. ps, worker etc.
	ReplicaTypeLabel = "training.kubeflow.org/replica-type"

	// OperatorNameLabel represents the label key for the operator name, e.g. tf-operator, mpi-operator, etc.
	OperatorNameLabel = "training.kubeflow.org/operator-name"

	// JobNameLabel represents the label key for the job name, the value is the job name.
	JobNameLabel = "training.kubeflow.org/job-name"

	// JobRoleLabel represents the label key for the job role, e.g. master.
	JobRoleLabel = "training.kubeflow.org/job-role"
)

Variables

This section is empty.

Functions

func RegisterDefaults

func RegisterDefaults(scheme *runtime.Scheme) error

RegisterDefaults adds defaulters functions to the given scheme. Public to allow building arbitrary schemes. All generated defaulters are covering - they call all nested defaulters.

Types

type CleanPodPolicy

type CleanPodPolicy string

+k8s:openapi-gen=true CleanPodPolicy describes how to deal with pods when the job is finished.

const (
	CleanPodPolicyUndefined CleanPodPolicy = ""
	CleanPodPolicyAll       CleanPodPolicy = "All"
	CleanPodPolicyRunning   CleanPodPolicy = "Running"
	CleanPodPolicyNone      CleanPodPolicy = "None"
)

type ControllerInterface

type ControllerInterface interface {
	// Returns the Controller name
	ControllerName() string

	// Returns the GroupVersionKind of the API
	GetAPIGroupVersionKind() schema.GroupVersionKind

	// Returns the GroupVersion of the API
	GetAPIGroupVersion() schema.GroupVersion

	// Returns the Group Name(value) in the labels of the job
	GetGroupNameLabelValue() string

	// Returns the Job from Informer Cache
	GetJobFromInformerCache(namespace, name string) (metav1.Object, error)

	// Returns the Job from API server
	GetJobFromAPIClient(namespace, name string) (metav1.Object, error)

	// GetPodsForJob returns the pods managed by the job. This can be achieved by selecting pods using label key "job-name"
	// i.e. all pods created by the job will come with label "job-name" = <this_job_name>
	GetPodsForJob(job interface{}) ([]*v1.Pod, 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>
	GetServicesForJob(job interface{}) ([]*v1.Service, error)

	// DeleteJob deletes the job
	DeleteJob(job interface{}) error

	// UpdateJobStatus updates the job status and job conditions
	UpdateJobStatus(job interface{}, replicas map[ReplicaType]*ReplicaSpec, jobStatus *JobStatus) error

	// UpdateJobStatusInApiServer updates the job status in API server
	UpdateJobStatusInApiServer(job interface{}, jobStatus *JobStatus) error

	// SetClusterSpec sets the cluster spec for the pod
	SetClusterSpec(job interface{}, podTemplate *v1.PodTemplateSpec, rtype, index string) error

	// Returns the default container name in pod
	GetDefaultContainerName() string

	// Get the default container port name
	GetDefaultContainerPortName() string

	// Returns if this replica type with index specified is a master role.
	// MasterRole pod will have "job-role=master" set in its label
	IsMasterRole(replicas map[ReplicaType]*ReplicaSpec, rtype ReplicaType, index int) bool

	// ReconcileJobs checks and updates replicas for each given ReplicaSpec of a job.
	// Common implementation will be provided and User can still override this to implement their own reconcile logic
	ReconcileJobs(job interface{}, replicas map[ReplicaType]*ReplicaSpec, jobStatus JobStatus, runPolicy *RunPolicy) error

	// ReconcilePods checks and updates pods for each given ReplicaSpec.
	// It will requeue the job in case of an error while creating/deleting pods.
	// Common implementation will be provided and User can still override this to implement their own reconcile logic
	ReconcilePods(job interface{}, jobStatus *JobStatus, pods []*v1.Pod, rtype ReplicaType, spec *ReplicaSpec,
		replicas map[ReplicaType]*ReplicaSpec) error

	// ReconcileServices checks and updates services for each given ReplicaSpec.
	// It will requeue the job in case of an error while creating/deleting services.
	// Common implementation will be provided and User can still override this to implement their own reconcile logic
	ReconcileServices(job metav1.Object, services []*v1.Service, rtype ReplicaType, spec *ReplicaSpec) error
}

ControllerInterface defines the Interface to be implemented by custom operators. e.g. tf-operator needs to implement this interface

type JobCondition

type JobCondition struct {
	// Type of job condition.
	Type JobConditionType `json:"type"`
	// Status of the condition, one of True, False, Unknown.
	Status v1.ConditionStatus `json:"status"`
	// The reason for the condition's last transition.
	Reason string `json:"reason,omitempty"`
	// A human readable message indicating details about the transition.
	Message string `json:"message,omitempty"`
	// The last time this condition was updated.
	LastUpdateTime metav1.Time `json:"lastUpdateTime,omitempty"`
	// Last time the condition transitioned from one status to another.
	LastTransitionTime metav1.Time `json:"lastTransitionTime,omitempty"`
}

+k8s:openapi-gen=true +k8s:deepcopy-gen=true JobCondition describes the state of the job at a certain point.

func (*JobCondition) DeepCopy

func (in *JobCondition) DeepCopy() *JobCondition

DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new JobCondition.

func (*JobCondition) DeepCopyInto

func (in *JobCondition) DeepCopyInto(out *JobCondition)

DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.

type JobConditionType

type JobConditionType string

+k8s:openapi-gen=true JobConditionType defines all kinds of types of JobStatus.

const (
	// JobCreated means the job has been accepted by the system,
	// but one or more of the pods/services has not been started.
	// This includes time before pods being scheduled and launched.
	JobCreated JobConditionType = "Created"

	// JobRunning means all sub-resources (e.g. services/pods) of this job
	// have been successfully scheduled and launched.
	// The training is running without error.
	JobRunning JobConditionType = "Running"

	// JobRestarting means one or more sub-resources (e.g. services/pods) of this job
	// reached phase failed but maybe restarted according to it's restart policy
	// which specified by user in v1.PodTemplateSpec.
	// The training is freezing/pending.
	JobRestarting JobConditionType = "Restarting"

	// JobSucceeded means all sub-resources (e.g. services/pods) of this job
	// reached phase have terminated in success.
	// The training is complete without error.
	JobSucceeded JobConditionType = "Succeeded"

	// JobFailed means one or more sub-resources (e.g. services/pods) of this job
	// reached phase failed with no restarting.
	// The training has failed its execution.
	JobFailed JobConditionType = "Failed"
)

type JobStatus

type JobStatus struct {
	// Conditions is an array of current observed job conditions.
	Conditions []JobCondition `json:"conditions"`

	// ReplicaStatuses is map of ReplicaType and ReplicaStatus,
	// specifies the status of each replica.
	ReplicaStatuses map[ReplicaType]*ReplicaStatus `json:"replicaStatuses"`

	// Represents time when the job was acknowledged by the job controller.
	// It is not guaranteed to be set in happens-before order across separate operations.
	// It is represented in RFC3339 form and is in UTC.
	StartTime *metav1.Time `json:"startTime,omitempty"`

	// Represents time when the job was completed. It is not guaranteed to
	// be set in happens-before order across separate operations.
	// It is represented in RFC3339 form and is in UTC.
	CompletionTime *metav1.Time `json:"completionTime,omitempty"`

	// Represents last time when the job was reconciled. It is not guaranteed to
	// be set in happens-before order across separate operations.
	// It is represented in RFC3339 form and is in UTC.
	LastReconcileTime *metav1.Time `json:"lastReconcileTime,omitempty"`
}

+k8s:openapi-gen=true +k8s:deepcopy-gen=true JobStatus represents the current observed state of the training Job.

func (*JobStatus) DeepCopy

func (in *JobStatus) DeepCopy() *JobStatus

DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new JobStatus.

func (*JobStatus) DeepCopyInto

func (in *JobStatus) DeepCopyInto(out *JobStatus)

DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.

type ReplicaSpec

type ReplicaSpec struct {
	// Replicas is the desired number of replicas of the given template.
	// If unspecified, defaults to 1.
	Replicas *int32 `json:"replicas,omitempty"`

	// Template is the object that describes the pod that
	// will be created for this replica. RestartPolicy in PodTemplateSpec
	// will be overide by RestartPolicy in ReplicaSpec
	Template v1.PodTemplateSpec `json:"template,omitempty"`

	// Restart policy for all replicas within the job.
	// One of Always, OnFailure, Never and ExitCode.
	// Default to Never.
	RestartPolicy RestartPolicy `json:"restartPolicy,omitempty"`
}

+k8s:openapi-gen=true +k8s:deepcopy-gen=true ReplicaSpec is a description of the replica

func (*ReplicaSpec) DeepCopy

func (in *ReplicaSpec) DeepCopy() *ReplicaSpec

DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new ReplicaSpec.

func (*ReplicaSpec) DeepCopyInto

func (in *ReplicaSpec) DeepCopyInto(out *ReplicaSpec)

DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.

type ReplicaStatus

type ReplicaStatus struct {
	// The number of actively running pods.
	Active int32 `json:"active,omitempty"`

	// The number of pods which reached phase Succeeded.
	Succeeded int32 `json:"succeeded,omitempty"`

	// The number of pods which reached phase Failed.
	Failed int32 `json:"failed,omitempty"`

	// Deprecated: Use Selector instead
	LabelSelector *metav1.LabelSelector `json:"labelSelector,omitempty"`

	// A Selector is a label query over a set of resources. The result of matchLabels and
	// matchExpressions are ANDed. An empty Selector matches all objects. A null
	// Selector matches no objects.
	Selector string `json:"selector,omitempty"`
}

+k8s:openapi-gen=true ReplicaStatus represents the current observed state of the replica.

func (*ReplicaStatus) DeepCopy

func (in *ReplicaStatus) DeepCopy() *ReplicaStatus

DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new ReplicaStatus.

func (*ReplicaStatus) DeepCopyInto

func (in *ReplicaStatus) DeepCopyInto(out *ReplicaStatus)

DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.

type ReplicaType

type ReplicaType string

+k8s:openapi-gen=true ReplicaType represents the type of the replica. Each operator needs to define its own set of ReplicaTypes.

type RestartPolicy

type RestartPolicy string

+k8s:openapi-gen=true RestartPolicy describes how the replicas should be restarted. Only one of the following restart policies may be specified. If none of the following policies is specified, the default one is RestartPolicyAlways.

const (
	RestartPolicyAlways    RestartPolicy = "Always"
	RestartPolicyOnFailure RestartPolicy = "OnFailure"
	RestartPolicyNever     RestartPolicy = "Never"

	// RestartPolicyExitCode policy means that user should add exit code by themselves,
	// The job operator will check these exit codes to
	// determine the behavior when an error occurs:
	// - 1-127: permanent error, do not restart.
	// - 128-255: retryable error, will restart the pod.
	RestartPolicyExitCode RestartPolicy = "ExitCode"
)

type RunPolicy

type RunPolicy struct {
	// CleanPodPolicy defines the policy to kill pods after the job completes.
	// Default to None.
	CleanPodPolicy *CleanPodPolicy `json:"cleanPodPolicy,omitempty"`

	// TTLSecondsAfterFinished is the TTL to clean up jobs.
	// It may take extra ReconcilePeriod seconds for the cleanup, since
	// reconcile gets called periodically.
	// Default to infinite.
	TTLSecondsAfterFinished *int32 `json:"ttlSecondsAfterFinished,omitempty"`

	// Specifies the duration in seconds relative to the startTime that the job may be active
	// before the system tries to terminate it; value must be positive integer.
	// +optional
	ActiveDeadlineSeconds *int64 `json:"activeDeadlineSeconds,omitempty"`

	// Optional number of retries before marking this job failed.
	// +optional
	BackoffLimit *int32 `json:"backoffLimit,omitempty"`

	// SchedulingPolicy defines the policy related to scheduling, e.g. gang-scheduling
	// +optional
	SchedulingPolicy *SchedulingPolicy `json:"schedulingPolicy,omitempty"`
}

+k8s:openapi-gen=true +k8s:deepcopy-gen=true RunPolicy encapsulates various runtime policies of the distributed training job, for example how to clean up resources and how long the job can stay active.

func (*RunPolicy) DeepCopy

func (in *RunPolicy) DeepCopy() *RunPolicy

DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new RunPolicy.

func (*RunPolicy) DeepCopyInto

func (in *RunPolicy) DeepCopyInto(out *RunPolicy)

DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.

type SchedulingPolicy

type SchedulingPolicy struct {
	MinAvailable           *int32           `json:"minAvailable,omitempty"`
	Queue                  string           `json:"queue,omitempty"`
	MinResources           *v1.ResourceList `json:"minResources,omitempty"`
	PriorityClass          string           `json:"priorityClass,omitempty"`
	ScheduleTimeoutSeconds *int32           `json:"scheduleTimeoutSeconds,omitempty"`
}

+k8s:openapi-gen=true SchedulingPolicy encapsulates various scheduling policies of the distributed training job, for example `minAvailable` for gang-scheduling.

func (*SchedulingPolicy) DeepCopy

func (in *SchedulingPolicy) DeepCopy() *SchedulingPolicy

DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new SchedulingPolicy.

func (*SchedulingPolicy) DeepCopyInto

func (in *SchedulingPolicy) DeepCopyInto(out *SchedulingPolicy)

DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL