0

I've got a docker based (linux) azure functions app running in Azure App service which we've recently updated to using .net 7. Since this update we've seen a massive spike in the amount of application insight logs for performance counters.

Previously we were getting around 1 performance counter per name per minute (so 60 "Private Bytes" entries per hour, 60 "% Processor Time" entries per hour and 60 "% Processor Time Normalized" entries per hour) All of these entries have the instance of "??APP_WIN32_PROC??".

Now we're getting upto over 6000 entries per hour for each counter name.

We're using a classic application insights resource configured in the functions app using the APPINSIGHTS_INSTRUMENTATIONKEY configuration key.

I've been trying to work out where these counters are coming from and why they have suddenly increased.

Kelvin
  • 255
  • 1
  • 2
  • 7

1 Answers1

1

We think we've found that this was a case of the functions app was actually failing to start with some exceptions and our best guess is that these performance counters are the app constantly rebooting but never starting cleanly.

Kelvin
  • 255
  • 1
  • 2
  • 7
  • Performance counters are a Windows OS feature and have to be collected by some other process, eg Performance Monitor on Windows. It's Azure itself that reads those counters when the application starts. – Panagiotis Kanavos Nov 28 '22 at 12:35