Troubleshooting
Start with telemetry, narrow the failing request path, and then decide whether the issue is instrumentation, configuration, or a backend dependency.
Most support work on Simple Chat starts with one question: is the failure in the app layer, in telemetry wiring, or in a downstream Azure dependency? This page gives the shortest path to answer that question.
OpenTelemetry settings
Use the official Azure Monitor and OpenTelemetry references when instrumentation variables or exporter settings are in question.
Failing backend calls
Trace the failed request in Application Insights first, capture the `operation_Id`, and pivot from requests into exceptions.
Startup instrumentation errors
If Flask instrumentation itself is breaking startup, disable it explicitly with an environment variable and restart the app.
OpenTelemetry Settings
Backend Call Failing
Simple Chat uses Flask instrumentation by default, and backend calls are logged to Application Insights. Start with the requests table to find the failing call, capture the operation_Id, and use that identifier to pivot into related exceptions.
Query failed requests
requests
| where success == false
Query most recent exceptions
exceptions
| top 10 by timestamp
Query exceptions associated with a specific operation_Id
exceptions
| where operation_Id == '61a97b6a6ddc11b465b5289738bddcf1'
Flask Instrumentation Startup Error
If startup logs show an error while Flask instrumentation is initializing, disable it with the DISABLE_FLASK_INSTRUMENTATION environment variable. Set the value to 1 or true, then restart the app service so the process starts cleanly without the instrumentation hook.