Here are the steps to use NRP with your UCSD email.

Your email will be registered by us.

1. Installations

After you have been registered, do the following:

1. Go to https://portal.nrp-nautilus.io/

2. Click Login

3. Change the Identity Provider (default is ORCID) to UCSD, and then click Log On

4. Go to the Home Page and at the very bottom, accept the Use Policy

5. Click Get Config and sign in with your UCSD email. A config file should be downloaded.

6. Go to https://kubernetes.io/docs/tasks/tools/

   Depending on your OS, click one of these three links.

1. For

   1. Windows users: Click on the Kubernetes release page

      

      Scroll down to find this exact version and click on the kubectl.exe

      

      It will download kubectl.exe to your computer.

   2. MacOS Users: Ask Girish for the step-by-step guide

   3. not too sure

2. In your home directory, make a new directory and name it .kube . Put the config and kubectl.exe in that .kube directory.

2. Making .yaml files

1. In the .kube directory, make a new file call it mypod.yaml file

2. Copy-paste this code into mypod.yaml

   # mypod.yaml
   apiVersion: v1
   kind: Pod
   metadata:
     name: philip-pod     # change philip to your name
   spec:
     containers:
     - name: philip-pod   # also change to your name here
       image: tensorflow/tensorflow:latest-gpu
       volumeMounts:
       - name: signal-processing-volume
         mountPath: /mnt/
       command: ["sleep", "infinity"]
       resources:
         limits:
           memory: 4Gi
           cpu: 4000m
           nvidia.com/gpu: 1
         requests:
           memory: 3Gi
           cpu: 3000m
           nvidia.com/gpu: 1
     volumes:
       - name: signal-processing-volume
         persistentVolumeClaim:
           claimName: signal-processing-volume
   

3. Make a new file myjob.yaml and put this code into it:

   # myjob.yaml
   apiVersion: batch/v1
   kind: Job
   metadata:
     name: philip-job     # change philip to your name
   spec:
     template:
       spec:
         containers:
         - name: philip-job    # also change to your name here
           image: tensorflow/tensorflow:latest-gpu
           volumeMounts:
             - name: signal-processing-volume
               mountPath: /mnt/
           resources:
             limits:
               memory: 16Gi
               cpu: 1280m
               nvidia.com/gpu: 1
             requests:
               memory: 16Gi
               cpu: 1280m
               nvidia.com/gpu: 1           
           command: ["echo", "Hello", "World!"]
           # We will figure out something in the future for what to put for command
         volumes:
           - name: signal-processing-volume
             persistentVolumeClaim:
               claimName: signal-processing-volume
         restartPolicy: Never
     backoffLimit: 5