| { |
| "apiVersion": "kubeflow.org/v1", |
| "kind": "TFJob", |
| "metadata": { |
| "creationTimestamp": "2022-07-30T15:53:52.000+08:00", |
| "generation": 1, |
| "labels": { |
| "submarine-experiment-name": "tensorflow-dist-mnist" |
| }, |
| "name": "experiment-1659167632755-0001", |
| "namespace": "default", |
| "resourceVersion": "39556", |
| "uid": "d9b3c2dd-ce17-400a-a4f7-2781294ff3d5" |
| }, |
| "spec": { |
| "tfReplicaSpecs": { |
| "Ps": { |
| "replicas": 1, |
| "template": { |
| "metadata": { |
| "annotations": { |
| "sidecar.istio.io/inject": "false" |
| } |
| }, |
| "spec": { |
| "containers": [ |
| { |
| "command": [ |
| "python", |
| "/var/tf_mnist/mnist_with_summaries.py", |
| "--log_dir\u003d/train/log", |
| "--learning_rate\u003d0.01", |
| "--batch_size\u003d150" |
| ], |
| "env": [ |
| { |
| "name": "ENV_1", |
| "value": "ENV1" |
| } |
| ], |
| "image": "apache/submarine:tf-mnist-with-summaries-1.0", |
| "name": "tensorflow", |
| "resources": { |
| "limits": { |
| "cpu": "4", |
| "memory": "4096M" |
| }, |
| "requests": { |
| "cpu": "4", |
| "memory": "2048M" |
| } |
| }, |
| "volumeMounts": [ |
| { |
| "mountPath": "/logs", |
| "name": "volume", |
| "subPath": "submarine-tensorboard/tensorflow-dist-mnist" |
| } |
| ] |
| } |
| ], |
| "volumes": [ |
| { |
| "name": "volume", |
| "persistentVolumeClaim": { |
| "claimName": "submarine-tensorboard-pvc" |
| } |
| } |
| ] |
| } |
| }, |
| "restartPolicy": "OnFailure" |
| }, |
| "Worker": { |
| "replicas": 2, |
| "template": { |
| "metadata": { |
| "annotations": { |
| "sidecar.istio.io/inject": "false" |
| } |
| }, |
| "spec": { |
| "containers": [ |
| { |
| "command": [ |
| "python", |
| "/var/tf_mnist/mnist_with_summaries.py", |
| "--log_dir\u003d/train/log", |
| "--learning_rate\u003d0.01", |
| "--batch_size\u003d150" |
| ], |
| "env": [ |
| { |
| "name": "ENV_1", |
| "value": "ENV1" |
| } |
| ], |
| "image": "apache/submarine:tf-mnist-with-summaries-1.0", |
| "name": "tensorflow", |
| "resources": { |
| "limits": { |
| "cpu": "2", |
| "memory": "2048M", |
| "nvidia.com/gpu": "1" |
| }, |
| "requests": { |
| "cpu": "2", |
| "memory": "1024M", |
| "nvidia.com/gpu": "1" |
| } |
| }, |
| "volumeMounts": [ |
| { |
| "mountPath": "/logs", |
| "name": "volume", |
| "subPath": "submarine-tensorboard/tensorflow-dist-mnist" |
| } |
| ] |
| } |
| ], |
| "volumes": [ |
| { |
| "name": "volume", |
| "persistentVolumeClaim": { |
| "claimName": "submarine-tensorboard-pvc" |
| } |
| } |
| ] |
| } |
| }, |
| "restartPolicy": "OnFailure" |
| } |
| }, |
| "backoffLimit": 3 |
| }, |
| "status": { |
| "conditions": [ |
| { |
| "lastTransitionTime": "2022-07-30T07:59:01Z", |
| "lastUpdateTime": "2022-07-30T07:59:01Z", |
| "message": "TFJob experiment-1659167632755-0001 is created.", |
| "reason": "TFJobCreated", |
| "status": "True", |
| "type": "Created" |
| }, |
| { |
| "lastTransitionTime": "2022-07-30T07:59:02Z", |
| "lastUpdateTime": "2022-07-30T07:59:02Z", |
| "message": "TFJob default/experiment-1659167632755-0001 is running.", |
| "reason": "TFJobRunning", |
| "status": "True", |
| "type": "Running" |
| } |
| ], |
| "replicaStatuses": { |
| "PS": { |
| "active": 1 |
| }, |
| "Worker": { |
| "active": 2 |
| } |
| } |
| } |
| } |
