Azure App Service Instance getting too slow with large number of requests - causing azure front door to stop sending requests to it

Question

Hi,

I tried to create an azure support request but it wouldn't work, so creating a ticket here. Please reach out to me on teams for internal details that we may not be able to share publicly (such as troubleshooting portal links).

Today at 27-07-2026 00:40 AM we got alerted that azure front door detected a drop in availability from our app service in Australia. Debugging it, it seems like the health checks were timing out.

On app service we noticed that the traffic distribution between instances seemed normal, as well as cpu and memory. However, one single instance had a spike in thread count and response time.

Requests:

Response time:

front door health percentage (purple bar is the affected app service):

Connection timeouts on front door health check to the app service:

Answer 1

The evidence points to one App Service worker slowing while other instances had normal CPU and memory. Front Door counts a probe without a valid response as failed, so the availability drop likely reflects that worker's delay, not a Front Door routing fault.Correlate the incident time with per-instance metrics and Application Insights telemetry. Before restarting, capture Process Explorer details and a thread dump; high thread count with normal CPU can mean requests are waiting on locks or downstream calls. Then use Health Check's Instances view or Advanced Application Restart to restart only that worker.

Enable App Service Health Check on a lightweight endpoint that validates critical dependencies and returns 200 only when ready. With at least two instances, App Service can remove an unhealthy worker from rotation. If it recurs, configure Auto-Heal and open Microsoft Support with timestamps, instance ID, probe failures, and the dump.