Troubleshoot errors with Gemini assistance

This document describes how you can use Gemini, an AI-powered collaborator in Google Cloud, to help you to troubleshoot errors in Error Reporting.

Error Reporting aggregates errors that the Error Reporting API collects and errors that it infers from common text patterns in log entries. Error Reporting creates error groups that contain errors that seem to have the same root cause. You can use Gemini to explain a selected error group.

This documentation assumes that your Google Cloud project already collects the errors that you want to explore.

For more information about which errors Error Reporting aggregates and how Error Reporting groups the errors, see Error Reporting overview.

Gemini doesn't use your prompts or its responses as data to train its models. For more information, see How Gemini for Google Cloud uses your data.

Before you begin

  1. Ensure that Gemini is set up for your Google Cloud user account and project.

  2. In the Google Cloud console, on the project selector page, select your Google Cloud project.

    Go to project selector

Interpret an error

Error Reporting sends only the text log entry that matches the error group to Gemini, so responses are based only on that text and don't include context such as other data in the log, or resources in your Google Cloud project. For information about how Gemini handles data, see Data governance in Gemini.

To troubleshoot an error with Gemini assistance, do the following:

  1. In the navigation panel of the Google Cloud console, select Error Reporting, and then select your Google Cloud project:

    Go to Error Reporting

  2. In the list of error groups, click the name of the error group that you want to troubleshoot.

  3. Next to the error group title, click Troubleshooting suggestions.

    Error Reporting sends the text of the error to Gemini with a prompt to ask for troubleshooting suggestions. Gemini responds with a generated list of suggestions in the Gemini pane. The suggestions are based only on the text of the error without any additional context.

What's next