Azure Durable Function Storage Account cost increase using deployment slots with a lot of ClientOtherErrors

[Mivaweb]

Description

Using deployment slots on Azure Durable Functions we see an increase of Storage account costs on the following API's:

• Queues v2 - LRS - EU West / Class 2 Operations
• Blob Storage - Hot LRS - EU West / Hot LRS Write Operations

We have 1 slot which is called 'prelive' and we set the starter functions to disabled ( TimerTrigger / ServiceBusTrigger ) to not run the orchestrations so we don't have conflicts with the 'production' slot.

This results in:

• No '*-applease' containers to be created on the storage account
• No subfolders to be created for the prelive function apps under azure-webjobs-hosts container
• No queues to be created for the prelive function apps

But after viewing the diagnostic logs on the blob and queue services of the storage account we see a lot of ClientOtherErrors on those locations which is causing an increase of costs by trying and retrying to find these locations.

Expected behavior

Missing containers / folders and queues to be created even if the starter functions are set to disabled.

Actual behavior

Missing containers / folders which results in a lot of ClientOtherErrors.
Storage egress increased which writes a lot of GB's.

Relevant source code snippets

Known workarounds

• Removed deployments slots for now

App Details

• Microsoft.Azure.Functions.Worker.Extensions.DurableTask (v1.2.2):
• Microsoft.Azure.Functions.Worker.Sdk (v2.0.0):
• .NET:

Screenshots

If deployed to Azure

• Timeframe issue: last 7 days until yesterday 4PM CEST where we dropped the deployment slots
• Azure region: West Europe

[Mivaweb]

Seems related to #3114 and #2981