Cosmos Trigger does not start consumption app that has scaled all the way down

[ntaylor-ads]

I have a python function app with two functions that use two CosmosDB triggers to ingest data for the rest of the app. The data stream is not continuous with traffic for only a couple hours once a day. The function app runs correctly on initial start and then correctly scales to zero once the traffic subsides. The issue is that the app never restarts the following day once the traffic resumes. If I then manually start the app (for example, by connecting to the live telemetry) is scales up and correctly resumes processing on the change feed until it eventually scales to zero again.

Please provide the following:

• Function App version: 4.1042.100.2
• Region: West Europe and East US 2

Both function apps have func.json's that look like the following (with different values behind the redactions.

{
    "scriptFile": "__init__.py",
    "bindings": [
        {
            "name": "documents",
            "type": "cosmosDBTrigger",
            "direction": "in",
            "connection": "<REDACTED>CosmosConnection",
            "databaseName": "<REDACTED>",
            "containerName": "<REDACTED>",
            "leaseConnection": "COSMOS_CONNECTION_STRING",
            "leaseDatabaseName": "<REDACTED>Data",
            "leaseContainerName": "ingestLeases",
            "leaseContainerPrefix": "<REDACTED>",
            "createLeaseContainerIfNotExists": true
        }
    ]
}

Where CosmosConnection is referencing the following app setting:

<REDACTED>CosmosConnection__accountEndpoint   =  "https://<database_account_name>.documents.azure.com:443/"

Expected behavior

The cosmos trigger should continue to function even after the app has scaled all the way down. Upon changes in the monitored cosmos container, it should scale up the function app.

Actual behavior

The function app never starts.

Known workarounds

I haven't tried these yet, but I assume creating a "keep-alive" timer function that runs every ~10 mins or creating a health endpoint that is regularly pinged by an external system would keep the function app alive and, therefore, the cosmos trigger would continue to function.

[jviau]

@ntaylor-ads this typically indicates a configuration / app setup issue. Scale controller most likely cannot access your CosmosDB account to detect when an instance needs to be activated to process new work.

Is your CosmosDB connection relying on managed identity authentication? Are you using system assigned or user assigned identity?

[ntaylor-ads]

It's set up using a system assigned identity, i.e. we grant the function's system assigned identity the required permissions.

[jviau]

Can you try adding this app setting as well:

<REDACTED>CosmosConnection__credential = "managedidentity"

If that doesn't work, you will need to open a support ticket for our scale controller team to investigate.

[microsoft-github-policy-service]

This issue has been automatically marked as stale because it has been marked as requiring author feedback but has not had any activity for 4 days. It will be closed if no further activity occurs within 3 days of this comment.

If you are not the original author (@ntaylor-ads) and believe this issue is not stale, please comment with /bot not-stale and I will not close it.