Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  1. spin up a node

    Code Block
    sudo -u apache kubectl create -f /export2/temp/cdeep3m/cil-cdeep3m_py.yaml --validate=false


  2. wait for running state

    Code Block
    sudo -u apache kubectl get pods
    NAME                             READY     STATUS             RESTARTS   AGE
    gpu-pod-cil-cdeep3m-py-testing   1/1       Running            0          1m
    
    


  3. connect to the pod

    Code Block
    sudo -u apache kubectl exec -it gpu-pod-cil-cdeep3m-py-testing bashroot@gpubash
    root@gpu-pod-cil-cdeep3m-py-testing:/home/cdeep3m#


  4. test

    Code Block
    ./CDEEP3M_prp --augspeed 1 --models 1fm --overlay --enhance --pod_id=A 3515 10.7295/W9CDEEP3M3

    to rerun previously run jobs in ~/CDEEP3M_prp change line 130

    Code Block
     }else {
        print  LOG getLoggingTime()."\t(".__LINE__.') '."$pod_id is not the first!! $data\n";
        exit 0;
      }
      return;

    to

    Code Block
     }else {
        print  LOG getLoggingTime()."\t(".__LINE__.') '."$pod_id is not the first!! $data\n";
        leave (0);
      }
      return;

    and uncomment line 513

    Code Block
    sub leave{
      my $code =shift;
      #unset_pod_status();
      if ($code){
        post_url ("$image_api_uri/Image_process_rest/update_cdeep3m_error/stage/$crop_id",'error => true',$image_api_user,$image_api_pass);
      }

    then run the testing command, the first time through will reset the status and the second time it will run the prediction. 

  5. When finished exit the pod with "exit" or CTRL d then delete the running pod

    Code Block
    sudo -u apache kubectl delete  -f /export2/temp/cdeep3m/cil-cdeep3m_py.yaml


...