Phase is wrong unexpected TfJob phase: Done

[jlewi]

In the job below phase as reported as Failed but the state reports succeeded. This is a bug.

apiVersion: mlkube.io/v1beta1
kind: TfJob
metadata:
  clusterName: ""
  creationTimestamp: 2017-10-31T22:06:13Z
  generation: 0
  name: cifar10-171031-220613
  namespace: default
  resourceVersion: "58104"
  selfLink: /apis/mlkube.io/v1beta1/namespaces/default/tfjobs/cifar10-171031-220613
  uid: b5c82c47-be87-11e7-823a-42010a8e007e
spec:
  RuntimeId: zcuu
  replicaSpecs:
  - IsDefaultPS: false
    replicas: 1
    template:
      metadata:
        creationTimestamp: null
      spec:
        containers:
        - command:
          - python
          - /tensorflow_models/tutorials/image/cifar10_estimator/cifar10_main.py
          - --data-dir=gs://cloud-ml-dev_jlewi/cifar10/data
          - --job-dir=gs://cloud-ml-dev_jlewi/cifar10/jobs/cifar10-171031-220613
          - --train-steps=100000
          - --log-device-placement
          - --num-gpus=4
          image: gcr.io/cloud-ml-dev/tf-models-gpu:591ca2e-dirty-b75d293
          name: tensorflow
          resources:
            limits:
              nvidia.com/gpu: "4"
        restartPolicy: OnFailure
    tfPort: 2222
    tfReplicaType: MASTER
  tensorboard:
    logDir: gs://cloud-ml-dev_jlewi/cifar10/jobs/cifar10-171031-220613
    serviceType: ""
    volumeMounts: null
    volumes: null
  tfImage: gcr.io/cloud-ml-dev/tf-models-cpu:591ca2e-dirty-b75d293
status:
  conditions: null
  controlPaused: false
  phase: Failed
  reason: 'unexpected TfJob phase: Done'
  replicaStatuses:
  - ReplicasStates:
      Succeeded: 1
    state: Succeeded
    tf_replica_type: MASTER
  state: Succeeded

[gaocegege]

We do not use phase now, thus close the issue soon 😄