mirror of https://github.com/kubeflow/examples.git
example mnist upgrade to v1alpha2 (#246)
* example mnist upgrade to v1alpha2 * Remove cleanPodPolicy * Fix kubeflow branch to v0.2.4
This commit is contained in:
Inki Hwang
k8s-ci-robot
parent
d878462bc5
commit
8e30631c54
|
|
@ -103,15 +103,15 @@ APP_NAME=my-kubeflow
|
|||
ks init ${APP_NAME}
|
||||
cd ${APP_NAME}
|
||||
|
||||
ks registry add kubeflow github.com/kubeflow/kubeflow/tree/master/kubeflow
|
||||
ks registry add kubeflow github.com/kubeflow/kubeflow/tree/v0.2.4/kubeflow
|
||||
|
||||
ks pkg install kubeflow/core@v0.1.2
|
||||
ks pkg install kubeflow/core@v0.2.4
|
||||
ks pkg install kubeflow/argo
|
||||
|
||||
# Deploy TF Operator and Argo
|
||||
kubectl create namespace ${NAMESPACE}
|
||||
ks generate core kubeflow-core --name=kubeflow-core --namespace=${NAMESPACE}
|
||||
ks generate argo kubeflow-argo --name=kubeflow-argo --namespace=${NAMESPACE} --imageTag=v2.1.0
|
||||
ks generate argo kubeflow-argo --name=kubeflow-argo --namespace=${NAMESPACE}
|
||||
|
||||
ks apply default -c kubeflow-core
|
||||
ks apply default -c kubeflow-argo
|
||||
|
|
|
|||
|
|
@ -112,147 +112,143 @@ spec:
|
|||
resource:
|
||||
action: apply
|
||||
# NOTE: need to detect master node complete
|
||||
successCondition: status.state == Succeeded
|
||||
successCondition: status.tfReplicaStatuses.Master.succeeded == 1
|
||||
manifest: |
|
||||
apiVersion: "kubeflow.org/v1alpha1"
|
||||
apiVersion: "kubeflow.org/v1alpha2"
|
||||
kind: "TFJob"
|
||||
metadata:
|
||||
name: {{workflow.parameters.job-name}}
|
||||
namespace: {{workflow.parameters.namespace}}
|
||||
spec:
|
||||
replicaSpecs:
|
||||
- replicas: 1
|
||||
tfReplicaType: MASTER
|
||||
template:
|
||||
spec:
|
||||
serviceAccountName: tf-job-operator
|
||||
containers:
|
||||
- image: {{workflow.parameters.tf-model-image}}
|
||||
name: tensorflow
|
||||
imagePullPolicy: Always
|
||||
env:
|
||||
- name: TF_MODEL_DIR
|
||||
value: {{inputs.parameters.s3-model-url}}
|
||||
- name: TF_EXPORT_DIR
|
||||
value: {{workflow.parameters.model-name}}
|
||||
- name: TF_TRAIN_STEPS
|
||||
value: "{{workflow.parameters.model-train-steps}}"
|
||||
- name: TF_TF_BATCH_SIZE
|
||||
value: "{{workflow.parameters.model-batch-size}}"
|
||||
- name: TF_LEARNING_RATE
|
||||
value: "{{workflow.parameters.model-learning-rate}}"
|
||||
- name: AWS_ACCESS_KEY_ID
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsAccessKeyID
|
||||
- name: AWS_SECRET_ACCESS_KEY
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsSecretAccessKey
|
||||
- name: AWS_DEFAULT_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: AWS_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_USE_HTTPS
|
||||
value: "{{workflow.parameters.s3-use-https}}"
|
||||
- name: S3_VERIFY_SSL
|
||||
value: "{{workflow.parameters.s3-verify-ssl}}"
|
||||
- name: S3_ENDPOINT
|
||||
value: {{workflow.parameters.s3-endpoint}}
|
||||
restartPolicy: OnFailure
|
||||
- replicas: {{workflow.parameters.tf-worker}}
|
||||
tfReplicaType: WORKER
|
||||
template:
|
||||
spec:
|
||||
serviceAccountName: tf-job-operator
|
||||
containers:
|
||||
- image: {{workflow.parameters.tf-model-image}}
|
||||
name: tensorflow
|
||||
imagePullPolicy: Always
|
||||
env:
|
||||
- name: TF_MODEL_DIR
|
||||
value: {{inputs.parameters.s3-model-url}}
|
||||
- name: TF_EXPORT_DIR
|
||||
value: {{workflow.parameters.model-name}}
|
||||
- name: TF_TRAIN_STEPS
|
||||
value: "{{workflow.parameters.model-train-steps}}"
|
||||
- name: TF_TF_BATCH_SIZE
|
||||
value: "{{workflow.parameters.model-batch-size}}"
|
||||
- name: TF_LEARNING_RATE
|
||||
value: "{{workflow.parameters.model-learning-rate}}"
|
||||
- name: AWS_ACCESS_KEY_ID
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsAccessKeyID
|
||||
- name: AWS_SECRET_ACCESS_KEY
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsSecretAccessKey
|
||||
- name: AWS_DEFAULT_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: AWS_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_USE_HTTPS
|
||||
value: "{{workflow.parameters.s3-use-https}}"
|
||||
- name: S3_VERIFY_SSL
|
||||
value: "{{workflow.parameters.s3-verify-ssl}}"
|
||||
- name: S3_ENDPOINT
|
||||
value: {{workflow.parameters.s3-endpoint}}
|
||||
restartPolicy: OnFailure
|
||||
- replicas: {{workflow.parameters.tf-ps}}
|
||||
tfReplicaType: PS
|
||||
template:
|
||||
spec:
|
||||
containers:
|
||||
- image: {{workflow.parameters.tf-model-image}}
|
||||
name: tensorflow
|
||||
imagePullPolicy: Always
|
||||
env:
|
||||
- name: TF_MODEL_DIR
|
||||
value: {{inputs.parameters.s3-model-url}}
|
||||
- name: TF_EXPORT_DIR
|
||||
value: {{workflow.parameters.model-name}}
|
||||
- name: TF_TRAIN_STEPS
|
||||
value: "{{workflow.parameters.model-train-steps}}"
|
||||
- name: TF_TF_BATCH_SIZE
|
||||
value: "{{workflow.parameters.model-batch-size}}"
|
||||
- name: TF_LEARNING_RATE
|
||||
value: "{{workflow.parameters.model-learning-rate}}"
|
||||
- name: AWS_ACCESS_KEY_ID
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsAccessKeyID
|
||||
- name: AWS_SECRET_ACCESS_KEY
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsSecretAccessKey
|
||||
- name: AWS_DEFAULT_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: AWS_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_USE_HTTPS
|
||||
value: "{{workflow.parameters.s3-use-https}}"
|
||||
- name: S3_VERIFY_SSL
|
||||
value: "{{workflow.parameters.s3-verify-ssl}}"
|
||||
- name: S3_ENDPOINT
|
||||
value: {{workflow.parameters.s3-endpoint}}
|
||||
restartPolicy: OnFailure
|
||||
terminationPolicy:
|
||||
chief:
|
||||
replicaIndex: 0
|
||||
replicaName: MASTER
|
||||
tfReplicaSpecs:
|
||||
Master:
|
||||
replicas: 1
|
||||
template:
|
||||
spec:
|
||||
serviceAccountName: tf-job-operator
|
||||
containers:
|
||||
- image: {{workflow.parameters.tf-model-image}}
|
||||
name: tensorflow
|
||||
imagePullPolicy: Always
|
||||
env:
|
||||
- name: TF_MODEL_DIR
|
||||
value: {{inputs.parameters.s3-model-url}}
|
||||
- name: TF_EXPORT_DIR
|
||||
value: {{workflow.parameters.model-name}}
|
||||
- name: TF_TRAIN_STEPS
|
||||
value: "{{workflow.parameters.model-train-steps}}"
|
||||
- name: TF_BATCH_SIZE
|
||||
value: "{{workflow.parameters.model-batch-size}}"
|
||||
- name: TF_LEARNING_RATE
|
||||
value: "{{workflow.parameters.model-learning-rate}}"
|
||||
- name: AWS_ACCESS_KEY_ID
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsAccessKeyID
|
||||
- name: AWS_SECRET_ACCESS_KEY
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsSecretAccessKey
|
||||
- name: AWS_DEFAULT_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: AWS_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_USE_HTTPS
|
||||
value: "{{workflow.parameters.s3-use-https}}"
|
||||
- name: S3_VERIFY_SSL
|
||||
value: "{{workflow.parameters.s3-verify-ssl}}"
|
||||
- name: S3_ENDPOINT
|
||||
value: {{workflow.parameters.s3-endpoint}}
|
||||
restartPolicy: OnFailure
|
||||
Worker:
|
||||
replicas: {{workflow.parameters.tf-worker}}
|
||||
template:
|
||||
spec:
|
||||
serviceAccountName: tf-job-operator
|
||||
containers:
|
||||
- image: {{workflow.parameters.tf-model-image}}
|
||||
name: tensorflow
|
||||
imagePullPolicy: Always
|
||||
env:
|
||||
- name: TF_MODEL_DIR
|
||||
value: {{inputs.parameters.s3-model-url}}
|
||||
- name: TF_EXPORT_DIR
|
||||
value: {{workflow.parameters.model-name}}
|
||||
- name: TF_TRAIN_STEPS
|
||||
value: "{{workflow.parameters.model-train-steps}}"
|
||||
- name: TF_BATCH_SIZE
|
||||
value: "{{workflow.parameters.model-batch-size}}"
|
||||
- name: TF_LEARNING_RATE
|
||||
value: "{{workflow.parameters.model-learning-rate}}"
|
||||
- name: AWS_ACCESS_KEY_ID
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsAccessKeyID
|
||||
- name: AWS_SECRET_ACCESS_KEY
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsSecretAccessKey
|
||||
- name: AWS_DEFAULT_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: AWS_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_USE_HTTPS
|
||||
value: "{{workflow.parameters.s3-use-https}}"
|
||||
- name: S3_VERIFY_SSL
|
||||
value: "{{workflow.parameters.s3-verify-ssl}}"
|
||||
- name: S3_ENDPOINT
|
||||
value: {{workflow.parameters.s3-endpoint}}
|
||||
restartPolicy: OnFailure
|
||||
Ps:
|
||||
replicas: {{workflow.parameters.tf-ps}}
|
||||
template:
|
||||
spec:
|
||||
containers:
|
||||
- image: {{workflow.parameters.tf-model-image}}
|
||||
name: tensorflow
|
||||
imagePullPolicy: Always
|
||||
env:
|
||||
- name: TF_MODEL_DIR
|
||||
value: {{inputs.parameters.s3-model-url}}
|
||||
- name: TF_EXPORT_DIR
|
||||
value: {{workflow.parameters.model-name}}
|
||||
- name: TF_TRAIN_STEPS
|
||||
value: "{{workflow.parameters.model-train-steps}}"
|
||||
- name: TF_BATCH_SIZE
|
||||
value: "{{workflow.parameters.model-batch-size}}"
|
||||
- name: TF_LEARNING_RATE
|
||||
value: "{{workflow.parameters.model-learning-rate}}"
|
||||
- name: AWS_ACCESS_KEY_ID
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsAccessKeyID
|
||||
- name: AWS_SECRET_ACCESS_KEY
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: {{workflow.parameters.aws-secret}}
|
||||
key: awsSecretAccessKey
|
||||
- name: AWS_DEFAULT_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: AWS_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_REGION
|
||||
value: {{workflow.parameters.aws-region}}
|
||||
- name: S3_USE_HTTPS
|
||||
value: "{{workflow.parameters.s3-use-https}}"
|
||||
- name: S3_VERIFY_SSL
|
||||
value: "{{workflow.parameters.s3-verify-ssl}}"
|
||||
- name: S3_ENDPOINT
|
||||
value: {{workflow.parameters.s3-endpoint}}
|
||||
restartPolicy: OnFailure
|
||||
- name: tf-tensorboard
|
||||
inputs:
|
||||
parameters:
|
||||
|
|
|
|||
Loading…
Reference in New Issue