Parallel bug fixing & unit testing with Jules and Observability extensions for Gemini CLI

Key Concepts

• Google Cloud Observability Error Reporting: A service that aggregates and analyzes errors from applications running on Google Cloud.
• Observability Extension: A tool or plugin that enhances the capabilities of observability platforms, in this case, allowing interaction with Google Cloud's error reporting.
• Jules Agent: An AI-powered agent designed to assist with software development tasks, including error investigation and fixing.
• Stack Trace: A report of the active stack frames at a particular point in time, used to diagnose errors.
• Arithmetic Exception: An error that occurs during arithmetic operations, such as division by zero.
• Unit Test: A piece of code that tests a specific unit of functionality within an application to ensure it behaves as expected.
• Diff Patch: A file containing the differences between two versions of a file, used to apply changes.
• Pull Request (PR): A mechanism for proposing changes to a codebase, allowing for review and discussion before merging.

Error Investigation and Resolution Workflow

This demonstration illustrates a workflow for identifying and resolving the top error reported in Google Cloud Observability Error Reporting, leveraging an Observability Extension and an AI agent named Jules.

1. Identifying the Top Error:

• The process begins by querying Google Cloud Observability Error Reporting for the top error within the last 30 days.
• The query reveals an ArithmeticException that occurred 232 times.
• The full stack trace for this exception is accessible for detailed analysis.

2. Delegating Error Investigation to Jules Agent:

• The user requests Jules to work on the identified issue.
• Crucially, the user ensures that Jules has access to the complete stack trace.
• A key requirement is that Jules must also add a unit test to prevent the recurrence of this bug.

3. Jules Agent Task Execution:

• Jules initiates a new task to address the error.
• The task details are visible in the Jules console, confirming that all necessary information, including the stack trace and the requirement for a unit test, is included in the prompt.
• While Jules is working, the user can continue with other development tasks, such as working on a new feature locally.

4. Reviewing Jules's Fix:

• Upon completion of the task, Jules provides an update.
• The user can interact with Jules using natural language to understand the changes.
• Jules generates a diff patch containing the proposed code modifications.
• The changes are published to a new branch named "exception_fix" for a structured review.

5. Analyzing the Code Changes:

• The review of the "exception_fix" branch reveals that the root cause of the error was a division by zero.
• The user notes that the item price was set incorrectly, leading to this bug.
• It is also discovered that no unit test existed for the affected file, which Jules has now implemented.
• Other files in the diff appear to be generated artifacts, not directly related to the fix.

6. Merging the Fix:

• The code changes are deemed satisfactory after review.
• A pull request is created for the "exception_fix" branch.
• The pull request is subsequently merged, effectively resolving the top detected issue from Google Cloud Error Reporting.

Technical Details and Observations

• The example repository is the backend of an e-commerce application, primarily written in Java.
• The Google Cloud Observability Extension is a prerequisite for accessing the relevant MCP (Monitoring, Control, and Performance) tools.
• The specific error identified was an ArithmeticException occurring 232 times.
• The fix involved addressing a division by zero scenario caused by an incorrectly set item price.
• Jules's contribution included not only fixing the bug but also adding a missing unit test for the affected file.

Conclusion

This demonstration effectively showcases a streamlined process for leveraging Google Cloud's observability tools and AI agents to automate error detection, investigation, and resolution. By integrating with error reporting and empowering an agent like Jules to not only fix bugs but also implement preventative measures like unit tests, development teams can significantly improve their efficiency and code quality. The workflow emphasizes clear communication, structured code review, and the ability to work in parallel, ultimately leading to a faster and more robust development cycle.