Using gcloud CLI and Cloud Shell for Automation

Overview and What You Will Learn

In this lab, you will use Cloud Shell to run gcloud commands, write a small script that creates several resources in a loop, and use --format and filtering flags to extract exactly the data you need instead of scanning full output manually.

Why This Matters in Production

A team preparing infrastructure for a scheduled load test needs ten identical VMs. Clicking through the Console's VM creation wizard ten times invites naming inconsistencies and configuration mistakes; the same task as a gcloud script with a loop takes seconds, produces identical results every time, and can be reviewed and reused for the next test.

Core Principles

Cloud Shell is a browser-based terminal built directly into the Console with gcloud pre-installed and already authenticated to your account - no local setup needed to start practicing immediately, and it comes with a small amount of persistent storage that survives between sessions.

Every action available in the Console ultimately calls the same underlying APIs gcloud calls directly - this is why scripting a task with gcloud produces results identical to doing it manually through the Console, just faster and more consistently repeatable.

Detailed Step-by-Step Practical Lab

1. Open Cloud Shell from the Console's top navigation bar.

2. Confirm your active project and account:

3. Create a project for this lab:

4. Write a small script that creates three VMs in a loop, rather than running the same command three times by hand:

5. List all three VMs to confirm the loop worked identically each time:

6. Use --format to extract only the specific fields needed, instead of scanning full output manually:

Note

--format="value(field)" uses gcloud's own formatting syntax, conceptually similar to AWS CLI's --query (JMESPath) or Azure CLI's --query. This becomes essential once scripting against dozens of resources where manually parsing full JSON or table output would be slow and error-prone.

7. Clean up:

Production Best Practices & Common Pitfalls

Common Mistake

Using the Console's creation wizard to manually create many similar resources one at a time, "since it's just a few." Manual repetition is exactly where inconsistent naming, forgotten configuration options, and copy-paste mistakes creep in - a script produces the same correct result every single time.

Tip

Save frequently used gcloud commands as small shell scripts in Cloud Shell's persistent storage. Cloud Shell keeps a small file share between sessions specifically so your scripts and tools survive even after you close the browser tab.

• Use --format="table(...)" while exploring interactively, and --format="value(...)" or --format="json" when scripting. Table output is easier for a human to scan; a value or JSON format is what a script should actually parse programmatically.
• Test destructive commands (delete, especially with wildcards or filters) against a throwaway project first. A loop that deletes resources is exactly as fast and exactly as repeatable as a loop that creates them - verify the filter logic somewhere safe before running it against anything real.

Quick Reference & Troubleshooting Commands