Your application is running on Compute Engine and experiencing sustained failures for a small subset of requests. You've identified the issue to a specific Compute Engine instance, but the instance is unresponsive to SSH. What is the next step you should take?