Deployment fails from time to time as the actions fails to find the function App

Hi, I am deploying a dotnet-isolated function app using .NET 6 and from time to time the deployment fails with the following error

Using RBAC for authentication, GitHub Action will perform resource validation.
##[debug]try-get AzureCLIAuthorizer
##[debug]"/usr/bin/az" account show
##[debug]"/usr/bin/az" cloud show
##[debug]"/usr/bin/az" account get-access-token
::add-mask::***
##[debug][GET] https://management.azure.com/subscriptions/***/resources?$filter=resourceType EQ 'Microsoft.Web%2FSites' AND name EQ '___FUNCTION__NAME___'&api-version=2016-07-01
Error: Execution Exception (state: ValidateAzureResource) (step: Invocation)
Error:   Resource ___FUNCTION__NAME___ doesn't exist.
Error:     Error: Resource ___FUNCTION__NAME___ doesn't exist.
    at Function.<anonymous> (/home/runner/work/_actions/Azure/functions-action/v1/node_modules/azure-actions-appservice-rest/Utilities/AzureResourceFilterUtility.js:21:23)
    at Generator.next (<anonymous>)
    at fulfilled (/home/runner/work/_actions/Azure/functions-action/v1/node_modules/azure-actions-appservice-rest/Utilities/AzureResourceFilterUtility.js:5:58)
    at processTicksAndRejections (internal/process/task_queues.js:93:5)
Error: Deployment Failed!
##[debug]Node Action run completed with exit code 1

I am using RBAC authentication via azure/login@v1. The service principal has contributor permission on the whole subscription. When I use Re-run all jobs for the workflow run, then deployment usually works fine.

I am deploying the function app via terraform, before deployment runs. But it also fails, if the function app is already there and exists for hours. So doesn’t seem to be some kind of “race condition”.

As I deploy IaC via terraform using publish-profiles would be overly complicated I would say. There seems to be no native support in terraform’s azurerm to get the profile.

[KernelPryanic]

Wow, no movement since February, really?

[Joaolfelicio]

Had the same issue, deleting the functions and re-running the pipeline did it, not the best solution in any case…