Hey folks! Anyone who use Container Apps with a custom DNS address experiencing this issue? https://github.com/pulumi/pulumi-azure-native/issues/2117

I looked at your issue. I don't have a suggestion to fix it but I literally have the same setup for something that I setup couple of months ago. I am actually going to be creating a new stack and deploying everything again over the next couple of weeks and hope to not run into that issue. By the way, small aside there about the

ManagedEnvironment

resource. The

runtimeSubnetId

property is not used. Essentially, all your container apps will run in the

infrastructureSubnetId

for that environment. See https://github.com/microsoft/azure-container-apps/issues/374.

Thanks for the note about

runtimeSubnetId

! This issue is presenting even from a complete rebuild of all the resources in the project, sadly. I can manually intervene and create the TXT record using the verification ID that the error response contains, so worst-case scenario, you can do that too, until the issue is resolved. This behaviour was working on Azure Native provider version

1.83.1

as of ~1 month ago

If it once worked with

1.85.0

, I feel it might not be an issue with the provider or Pulumi. I think there was a Pulumi SDK bump with that version so any behavior changes should have affected your first time execution too. The Pulumi SDK that the provider depends upon was again bumped in the next patch version. So I am really curious if

1.88.1

fixes it for you.

Sadly it didn't. This indeed does seem to point to a breaking change in the Azure API. It could be worked around with some human intervention in the provider, but I think that goes against the intent of the provider being auto-generated from the Azure API spec

Ah bummer!

I updated the issue report with this information too. I wonder if, in the meantime, I could wrap the Container App resource creation with a try-catch and grab the verification ID from the error message?

Hmm I don't think you could do that. Even if it worked, you'd still have to re-run your infrastructure app because the creation of the container app would have failed. It will likely get a bit hacky after that.