App service plan linux - Why the Http request queue is always set to 1

Question

I've an app service plan with 4 apps running. When I'm doing a stress test on those apps, I would

expect the http request queue to grow ( the CPU is spiking at 100 %).

But the only thing I see is the memory and CPU growing. All the network metrics are stuck to 0.

Is there a reason for that ?

Answer 1

@Chabbey Francois Julien (extern) In Azure App Service, the HTTP request queue metric is a VM-level metric, which means it should be checked at the App Service Plan level. If you’re observing the CPU spiking to 100% but not seeing an increase in the HTTP request queue, it could be due to several reasons:

1. Application Insights: If you have Application Insights connected, ensure that you’re looking at the correct metric. The “Http Requests in Application Queue” metric should show the number of requests waiting to be processed. However, if this metric always shows 0, it might not be reflecting the actual queue length.
2. Metric Definitions: There are two different metrics to consider:
   • Requests In Application Queue: This shows the current number of requests the app service is processing.
   • Http Queue Length: This represents the average number of HTTP requests that had to wait in the queue before being fulfilled. A high or increasing HTTP Queue length indicates a plan under heavy load.
3. Autoscaling: If you have autoscaling set up based on the HttpQueueLength, and it scales up when the length is greater than 1, it’s possible that the autoscaling is effectively managing the load, preventing the queue from growing.

If you’re consistently seeing the network metrics stuck at 0, it could be a monitoring or configuration issue. You might want to review the configuration settings for your App Service Plan and ensure that the monitoring tools are set up correctly to reflect the actual performance and load on your applications. Additionally, checking the server logs and diagnostic information can provide more insights into the behavior you’re observing.

Let me know if you have any further questions by replying to this message.