Troubleshoot a failing cloud resource from the console
Start an investigation on a broken Google Cloud resource and get observations and likely causes drawn from logs, config and metrics.
The use case
Engineers debugging a failed Cloud Run deploy or an unhealthy GKE workload often hop between logs, metrics and configuration pages. Gemini Cloud Assist investigations take a description of the problem and the affected resources, then gather logs, configuration and metrics, and return observations and hypotheses about what is wrong with suggested next steps. The engineer still checks each observation, runs the fix, and can hand the investigation to Google support with the context attached if needed. It works best when the problem description names the resource, the time it started and what was expected, so the agent knows where to look first.
Example workflow
- Describe the issue and pick the affected resources
- Start an investigation
- Read observations and ranked hypotheses
- Verify the top hypothesis in logs or config
- Apply the fix or escalate with the investigation attached
Tools
- Gemini Cloud Assist
- Google Cloud
Risks and guardrails
- Hypotheses can be wrong; confirm before changing config
- Investigations need read access to logs; review IAM
- Do not paste secrets into the issue description
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- Google Cloud docs: Gemini Cloud Assist investigations: docs.cloud.google.com/cloud-assist/investigations