Kubeflow – TFJob v1

Docs: TFJob Common

Mon, 01 Jan 0001 00:00:00 +0000

Out of date

This guide contains outdated information pertaining to Kubeflow 1.0. This guide needs to be updated for Kubeflow 1.1.

Packages:

kubeflow.org

kubeflow.org

Package v1 is the v1 version of the API.

Resource Types:

CleanPodPolicy (`string` alias)

CleanPodPolicy describes how to deal with pods when the job is finished. Can be one of: All, Running, or None.

JobCondition

(Appears on: JobStatus)

JobCondition describes the state of the job at a certain point.

Field	Description
`type` JobConditionType	Type of job condition.
`status` Kubernetes core/v1.ConditionStatus	Status of the condition, one of True, False, or Unknown.
`reason` string	The reason for the condition’s last transition.
`message` string	A readable message indicating details about the transition.
`lastUpdateTime` Kubernetes meta/v1.Time	The last time this condition was updated.
`lastTransitionTime` Kubernetes meta/v1.Time	Last time the condition transitioned from one status to another.

JobConditionType (`string` alias)

(Appears on: JobCondition)

JobConditionType defines all possible types of JobStatus. Can be one of: Created, Running, Restarting, Succeeded, or Failed.

JobStatus

JobStatus represents the current observed state of the training job.

Field	Description
`conditions` [][]github.com/kubeflow/tf-operator/pkg/apis/common/v1.JobCondition	An array of current observed job conditions.
`replicaStatuses` map[github.com/kubeflow/tf-operator/pkg/apis/common/v1.ReplicaType]github.com/kubeflow/tf-operator/pkg/apis/common/v1.ReplicaStatus*	A map from ReplicaType (key) to ReplicaStatus (value), specifying the status of each replica.
`startTime` Kubernetes meta/v1.Time	Represents the 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.
`completionTime` Kubernetes meta/v1.Time	Represents the 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.
`lastReconcileTime` Kubernetes meta/v1.Time	Represents the 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.

ReplicaSpec

ReplicaSpec is a description of the job replica.

Field	Description
`replicas` int32	The desired number of replicas of the given template. If unspecified, defaults to 1.
`template` Kubernetes core/v1.PodTemplateSpec	Describes the pod that will be created for this replica. Note that RestartPolicy in PodTemplateSpec will be overidden by RestartPolicy in ReplicaSpec.
`restartPolicy` RestartPolicy	Restart policy for all replicas within the job. One of Always, OnFailure, Never, or ExitCode. Defaults to Never.

ReplicaStatus

(Appears on: JobStatus)

ReplicaStatus represents the current observed state of the replica.

Field	Description
`active` int32	The number of actively running pods.
`succeeded` int32	The number of pods which reached phase Succeeded.
`failed` int32	The number of pods which reached phase Failed.

ReplicaType (`string` alias)

ReplicaType represents the type of the job replica. Each operator (e.g. TensorFlow, PyTorch) needs to define its own set of ReplicaTypes.

RestartPolicy (`string` alias)

(Appears on: ReplicaSpec)

RestartPolicy describes how the replicas should be restarted. Can be one of: Always, OnFailure, Never, or ExitCode. If none of the following policies is specified, the default one is RestartPolicyAlways.

Generated with gen-crd-api-reference-docs on git commit fd76deec.

Docs: TFJob TensorFlow

Mon, 01 Jan 0001 00:00:00 +0000

Out of date

This guide contains outdated information pertaining to Kubeflow 1.0. This guide needs to be updated for Kubeflow 1.1.

Packages:

kubeflow.org

kubeflow.org

Package v1 is the v1 version of the API.

Resource Types:

TFJob

TFJob

Represents a TFJob resource.

Field Description

apiVersion
string kubeflow.org/v1

kind
string TFJob

metadata
Kubernetes meta/v1.ObjectMeta

Standard Kubernetes object’s metadata.

Refer to the Kubernetes API documentation for the fields of the metadata field.

spec
TFJobSpec

Specification of the desired state of the TFJob.

`activeDeadlineSeconds` int64	(Optional) Specifies the duration (in seconds) since startTime during which the job can remain active before it is terminated. Must be a positive integer. This setting applies only to pods where restartPolicy is OnFailure or Always.
`backoffLimit` int32	(Optional) Number of retries before marking this job as failed.
`cleanPodPolicy` common/v1.CleanPodPolicy	Defines the policy for cleaning up pods after the TFJob completes. Defaults to Running.
`ttlSecondsAfterFinished` int32	Defines the TTL for cleaning up finished TFJobs (temporary before kubernetes adds the cleanup controller). It may take extra ReconcilePeriod seconds for the cleanup, since reconcile gets called periodically. Defaults to infinite.
`tfReplicaSpecs` map[github.com/kubeflow/tf-operator/pkg/apis/tensorflow/v1.TFReplicaType]github.com/kubeflow/tf-operator/pkg/apis/common/v1.ReplicaSpec*	A map of TFReplicaType (type) to ReplicaSpec (value). Specifies the TF cluster configuration. For example, { “PS”: ReplicaSpec, “Worker”: ReplicaSpec, }

status
common/v1.JobStatus

Most recently observed status of the TFJob. Read-only (modified by the system).

TFJobSpec

(Appears on: TFJob)

TFJobSpec is a desired state description of the TFJob.

Field	Description
`activeDeadlineSeconds` int64	(Optional) Specifies the duration (in seconds) since startTime during which the job can remain active before it is terminated. Must be a positive integer. This setting applies only to pods where restartPolicy is OnFailure or Always.
`backoffLimit` int32	(Optional) Number of retries before marking this job as failed.
`cleanPodPolicy` common/v1.CleanPodPolicy	Defines the policy for cleaning up pods after the TFJob completes. Defaults to Running.
`ttlSecondsAfterFinished` int32	Defines the TTL for cleaning up finished TFJobs (temporary before kubernetes adds the cleanup controller). It may take extra ReconcilePeriod seconds for the cleanup, since reconcile gets called periodically. Defaults to infinite.
`tfReplicaSpecs` map[github.com/kubeflow/tf-operator/pkg/apis/tensorflow/v1.TFReplicaType]github.com/kubeflow/tf-operator/pkg/apis/common/v1.ReplicaSpec*	A map of TFReplicaType (type) to ReplicaSpec (value). Specifies the TF cluster configuration. For example, { “PS”: ReplicaSpec, “Worker”: ReplicaSpec, }

TFReplicaType (`string` alias)

TFReplicaType is the type for TFReplica. Can be one of: “Chief”/“Master” (semantically equivalent), “Worker”, “PS”, or “Evaluator”.

Generated with gen-crd-api-reference-docs on git commit fd76deec.

Kubeflow – TFJob v1

Docs: TFJob Common

Out of date

kubeflow.org

CleanPodPolicy (string alias)

JobCondition

JobConditionType (string alias)

JobStatus

ReplicaSpec

ReplicaStatus

ReplicaType (string alias)

RestartPolicy (string alias)

Docs: TFJob TensorFlow

Out of date

kubeflow.org

TFJob

TFJobSpec

TFReplicaType (string alias)

CleanPodPolicy (`string` alias)

JobConditionType (`string` alias)

ReplicaType (`string` alias)

RestartPolicy (`string` alias)

TFReplicaType (`string` alias)