Start here. This is the direct spoken answer to practice first.
Why this question matters
An empty App Service log stream does not mean the failure left no evidence. Application Insights, Azure Monitor, deployment output, platform diagnostics, web server logs, and failed-request traces may each hold a different part of the incident.
I would not stop just because log stream is empty. I would check whether application logging is enabled, whether logs go to Application Insights or Azure Monitor instead of the file stream, and whether the failing path logs exceptions at the right level. I would also inspect App Service diagnostics, deployment logs, web server logs, and failed request tracing where available. For production clients, I would keep detailed errors out of responses and put diagnostic detail in safe internal logs.