Creating a script asset with CPDCTL in Watson Machine Learning

Learn to create a script asset with Cloud Pak for Data command-line interface (CPDCTL).

1. Install the CPDCTL library.

   CPDCTL_VERSION = ""
   
   ! curl -L https://github.com/IBM/cpdctl/releases/download/{CPDCTL_VERSION}/cpdctl_linux_amd64.tar.gz --output cpdctl_linux_amd64.tar.gz
   

   Check for the available Cloud Pak for Data versions on the GitHub releases page.

2. Extract downloaded archive.

   ! tar xzf cpdctl_linux_amd64.tar.gz
   

3. Ensure that the cpdctl archive is in the current working directory and the system PATH.

   %%capture
   %env PATH={CWD}:{PATH}
   

4. Check that CLI is accessible.

   ! cpdctl version
   

5. Optional: The following steps are not required if you are running the script in a Cloud Pak for Data environment runtime.

   URL = "<<your cluster URL>>"
   USERNAME = "<<CPD user name>>"
   APIKEY = "<<CPD user API key>>"
   
   ! cpdctl config context cpd-demo --url {URL} --username {USERNAME} --apikey {APIKEY}
   
   project_id = os.getenv('PROJECT_ID')
   

6. Open your script.

   with open('<test_script>', 'w') as f:
       f.write("echo OK")
   

7. Create the new script asset.

   ! cpdctl asset script create --file <test_script> --project-id {project_id}
   

   When you run this line of code, you receive the asset ID in its output.

8. List the script asset in the project.

   ! cpdctl asset search --type-name script --query 'test.sh' --project-id {project_id}
   

9. Optional: You can delete the script with the following code.

   SCRIPT_ASSET_ID = ""
   
   ! cpdctl asset delete --asset-id {SCRIPT_ASSET_ID} --project-id {project_id}
   

Parent topic: Managing AI lifecycle with CPDCTL