Azure VMSS Autoscale Triggering Multiple Events and Causing Multiple Function Executions

I am facing an issue with Azure VMSS autoscale notifications (through webbhook) triggering multiple executions of an Azure Function. This causes my function, which includes a pipeline trigger and a DNS update, to be executed multiple times in parallel, leading to unintended behavior, such as triggering my Azure DevOps pipeline and DNS updates three times instead of once.

VMSS Configuration: The VMSS has autoscale enabled, and I am monitoring autoscale notifications to trigger actions after VM scale events (both scale-in and scale-out).

Azure Function: Language: PowerShell 7.4 Trigger: HTTP Trigger Purpose: The function processes autoscale notifications, checks the provisioning state of VMSS instances, triggers an Azure DevOps pipeline, and updates DNS records. Expected Behavior: Trigger the function once for each scale event (either scale-in or scale-out) and perform the VMSS provisioning check, trigger the pipeline once, and update DNS once.

When I manually execute the function code (outside the autoscale event), everything works as expected. The function runs only once, triggers the pipeline once, and performs the DNS update correctly.

The issue only occurs when the function is triggered automatically via the autoscale notification from VMSS, where it runs three times instead of once.

The same event is logged three times in Azure Function logs. Here is an example of the log entries:

2024-09-17T06:50:26.311 [Information] INFORMATION: Triggering DNS update function... 2024-09-17T06:51:00.320 [Information] INFORMATION: DNS update function triggered. Response: 2024-09-17T06:51:00.323 [Information] Executed 'Functions.ansible_pipeline' (Succeeded, Id=78b2f41a-6b7c-412d-b147-2f3632d7835f, Duration=66337ms) 2024-09-17T06:51:00.730 [Information] INFORMATION: DNS update function triggered. Response: 2024-09-17T06:51:00.730 [Information] Executed 'Functions.ansible_pipeline' (Succeeded, Id=b2048c9c-1c28-4ea7-b6d0-9f8217fa6f0d, Duration=86803ms) 2024-09-17T06:51:01.164 [Information] INFORMATION: DNS update function triggered. Response: 2024-09-17T06:51:01.165 [Information] Executed 'Functions.ansible_pipeline' (Succeeded, Id=8e0db316-ea98-4376-8af3-10dd8e080665, Duration=107182ms) INFORMATION: {"version": "1.0","status": "Activated","operation": "Scale Out","context": {"timestamp": "2024-09-16T15:42:41.0770601Z","id": .... INFORMATION: {"version": "1.0","status": "Activated","operation": "Scale Out","context": {"timestamp": "2024-09-16T15:42:41.0770601Z","id": .... INFORMATION: {"version": "1.0","status": "Activated","operation": "Scale Out","context": {"timestamp": "2024-09-16T15:42:41.0770601Z","id": ....

In the Azure Function logs, it shows that the autoscale event (e.g., Scale Out) is received multiple times with the same timestamp and details, causing the function to execute multiple times.

This is the json payload sent out by VMSS.

{
    "version": "1.0",
    "status": "Activated",
    "operation": "Scale Out",
    "context": {
        "timestamp": "2024-09-16T15:16:40.5377227Z",
        "id": "/subscriptions/sid/resourceGroups/rg-name/providers/microsoft.insights/autoscalesettings/auto-scale-set-vmss-name",
        "name": "auto-scale-set-vmss-name",
        "details": "Autoscale successfully started scale operation for resource 'vmss-name' from capacity '1' to capacity '2'",
        "subscriptionId": "sid",
        "resourceGroupName": "rg-name",
        "resourceName": "vmss-name",
        "resourceType": "microsoft.compute/virtualmachinescalesets",
        "resourceId": "/subscriptions/sid/resourceGroups/rg-name/providers/Microsoft.Compute/virtualMachineScaleSets/vmss-name",
        "portalLink": "https://portal.azure.com/#resource/subscriptions/sid/resourceGroups/rg/providers/Microsoft.Compute/virtualMachineScaleSets/vmss-name",
        "resourceRegion": "eastus",
        "oldCapacity": "1",
        "newCapacity": "2"
    },
    "properties": {}
}

Probably the VMSS sent 3 notification like "scale up initiated", "scale up in progress" and "scale up completed".

Replication steps:

1. VMSS auto scales (scales out) based on autoscale rules.
2. An HTTP-triggered Azure Function is set to process the autoscale notification.
3. The function logs multiple identical notifications for the same autoscale event, causing it to be executed three times for a single scale-out event.
4. As a result, the Azure DevOps pipeline and DNS update operations are also triggered three times, which is undesirable.