PyTorchJob

Reference documentation for PyTorchJob

This guide contains outdated information pertaining to Kubeflow 1.0.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

Packages:

kubeflow.org

Package v1beta2 is the v1beta2 version of the API.

Resource Types:

PyTorchJob represents the configuration of PyTorchJob

Field Description

apiVersion
string kubeflow.org/v1beta2

kind
string PyTorchJob

metadata
Kubernetes meta/v1.ObjectMeta

Standard object’s metadata.

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

spec
PyTorchJobSpec

Specification of the desired behavior of the PyTorchJob.

`activeDeadlineSeconds` int64	(Optional) 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. This method applies only to pods with restartPolicy == OnFailure or Always.
`backoffLimit` int32	(Optional) Optional number of retries before marking this job failed.
`cleanPodPolicy` common/v1beta2.CleanPodPolicy	CleanPodPolicy defines the policy to kill pods after PyTorchJob is succeeded. Default to Running.
`ttlSecondsAfterFinished` int32	TTLSecondsAfterFinished is the TTL to clean up pytorch-jobs (temporary before kubernetes adds the cleanup controller). It may take extra ReconcilePeriod seconds for the cleanup, since reconcile gets called periodically. Default to infinite.
`pytorchReplicaSpecs` map[github.com/kubeflow/pytorch-operator/pkg/apis/pytorch/v1beta2.PyTorchReplicaType]github.com/kubeflow/tf-operator/pkg/apis/common/v1beta2.ReplicaSpec*	PyTorchReplicaSpecs is map of PyTorchReplicaType and PyTorchReplicaSpec specifies the PyTorch replicas to run. For example, { “Master”: PyTorchReplicaSpec, “Worker”: PyTorchReplicaSpec, }

status
common/v1beta2.JobStatus

Most recently observed status of the PyTorchJob. This data may not be up to date. Populated by the system. Read-only.

(Appears on: PyTorchJob)

PyTorchJobSpec is a desired state description of the PyTorchJob.

Field	Description
`activeDeadlineSeconds` int64	(Optional) 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. This method applies only to pods with restartPolicy == OnFailure or Always.
`backoffLimit` int32	(Optional) Optional number of retries before marking this job failed.
`cleanPodPolicy` common/v1beta2.CleanPodPolicy	CleanPodPolicy defines the policy to kill pods after PyTorchJob is succeeded. Default to Running.
`ttlSecondsAfterFinished` int32	TTLSecondsAfterFinished is the TTL to clean up pytorch-jobs (temporary before kubernetes adds the cleanup controller). It may take extra ReconcilePeriod seconds for the cleanup, since reconcile gets called periodically. Default to infinite.
`pytorchReplicaSpecs` map[github.com/kubeflow/pytorch-operator/pkg/apis/pytorch/v1beta2.PyTorchReplicaType]github.com/kubeflow/tf-operator/pkg/apis/common/v1beta2.ReplicaSpec*	PyTorchReplicaSpecs is map of PyTorchReplicaType and PyTorchReplicaSpec specifies the PyTorch replicas to run. For example, { “Master”: PyTorchReplicaSpec, “Worker”: PyTorchReplicaSpec, }

PyTorchReplicaType is the type for PyTorchReplica.

Generated with gen-crd-api-reference-docs on git commit e775742.

Last modified 30.04.2021: fix broken link (#2672) (c1aba76b)