You can view elastic jobs that are submitted by using Arena in AI Dashboard. This topic describes how to view the details of an elastic job in AI Dashboard.

Prerequisites

  1. Run the following command to submit a training job by using Arena:
    arena submit tf \
      --name=tf-git \
      --gpus=1 \
      --image=tensorflow/tensorflow:1.5.0-devel-gpu \
      --sync-mode=git \
      --sync-source=https://code.aliyun.com/xiaozhou/tensorflow-sample-code.git \
      "python code/tensorflow-sample-code/tfjob/docker/mnist/main.py --max_steps 10000 --data_dir=code/tensorflow-sample-code/data"
  2. Log on to AI Dashboard by using the credentials of the administrator.
  3. In the left-side navigation pane of AI Dashboard, choose Elastic Job > Job List.
  4. On the Training Job tab, you can view the training job submitted in Step 1.
  5. In the training job list, select the job that you want to view and click Detail in the Operator column.
    On the Job Cost page, you can view the following information about the job: the duration, the estimated actual cost, the estimated on-demand cost, and the estimated saved cost. You can also view the following information about each pod that runs the job: state, duration, resource type, instance type, and price.
Note You can submit inference jobs by using Arena or kubectl and then view the job details on the Inference Job tab.