Hello Tung and the BoltOps Community,
I’m facing a situation where I think I’ve hit a limitation from pure Terraform HCL (and of course, as always the case, looking if Terraspace could help overcome this limitation).
I’m trying to deploy AWS AppStream stacks/fleets and configure its autoscaling policy. From https://docs.aws.amazon.com/appstream2/latest/developerguide/roles-required-for-appstream.html, it mentions a few roles that is needed for the AppStream autoscaling.
The thing with the roles is… I think they don’t exist until we try to utilize AppStream autoscaling (where it would create it automatically)… So what happens is, when I deploy my code the first time, it would fail (because the roles don’t exist). But when I try to explicitly create the role, it would error saying that the roles already exists… And what I’d like to do is cater for both scenarios without having to recode/reconfigure input. I.e. what I want is to evaluate whether the resource exists, and only create it if it’s not there…
However, according to https://github.com/hashicorp/terraform/issues/23178, it was never the intention for this conditional creation of resources to happen (because it blurs whether or not the object is managed by Terraform).
In any case, I was wondering if I’d able to do something like
data "aws_iam_role" "amazonappstreamserviceaccess" {
name = "AmazonAppStreamServiceAccess"
}
to query the role, and somehow put the result into a ruby variable, and use erb templating to evaluate the variable, and conditionally create/ignore the creation of the roles? Or perhaps, is there a better way to do this?
Interested to know if someone’s encountered a similar challenge and managed to work around around it
Many thanks,
James