Token expires during longer Pulumi programs

[ekrengel]

@metral and I were both trying to provision an AKS cluster which took a significant amount of time (~20-40mins). While pulumi up was running, our Authorization tokens expired resulting in this error:

error: Error building AzureRM Client: Error populating Client ID from the Azure CLI: No Authorization Tokens were found - please re-authenticate using `az login`.

This seems to be related to this TF error.

[stack72]

Linking this issue... https://github.com/terraform-providers/terraform-provider-azurerm/issues/2602

[stack72]

Reproduction code https://pulumi.io/quickstart/kubernetes/tutorial-azure-kubernetes-service/

[mikhailshilkov]

While the upstream issue is open for half-a-year, what's the workaround to apply in Pulumi programs?

Today, I got the failure while creating an instance of Azure API Management. The next run of pulumi up suggests creating a new one. Should I adopt the created-but-reported-as-failed resource somehow?

[praneetloke]

It looks like this may be fixed with the provider version 1.20. I see this note in the docs:

Prior to version 1.20 the AzureRM Provider used a different method of authorizing via the Azure CLI where credentials reset after an hour - as such we'd recommend upgrading to version 1.20 or later of the AzureRM Provider.

[stack72]

In the latest Terraform AzureRM provider that is due for release, the following is included in it

https://github.com/terraform-providers/terraform-provider-azurerm/pull/3850

I don't believe they are going to support anything other than Service Principal for the foreseeable future