Terraform: Module Outputs

[eddytrex]

Terraform Version

terraform version
Terraform v0.12.5
+ provider.null v2.1.2

Terraform Configuration Files

module 1:

provider "null" {}

resource "null_resource" "name" {
  count    = 1
  triggers = { description = var.description }

  provisioner "local-exec" { command = "echo name " }
}

variable "description" {
  type = string
}

output "ids" {
  value = [for s in null_resource.name : s.id]
}

moduel2:

provider "null" {}

resource "null_resource" "hi" {
  count    = length(var.resources)
  triggers = { t = "${var.resources[count.index]}" }

  provisioner "local-exec" {
    command = "echo greattings ${var.resources[count.index]}"
  }
}

variable "resources" {
  type = list(string)
}

module3:

module "r" {
  source      = "../module1"
  description = var.description
}

module "hi" {
  source    = "../module2"
  resources = module.r.ids
}

variable "description" {
  type = string
}

Debug Output

https://gist.github.com/eddytrex/43c5bd47b273607705ae0f9bf1931503

Crash Output

Error: Cycle: module.r.output.ids, module.hi.var.resources, module.hi.null_resource.hi (prepare state), module.hi.null_resource.hi[0] (destroy), module.r.null_resource.name[0] (destroy), module.r.null_resource.name[0]

Expected Behavior

Renew the null_resource in the module 1
pass the values to the module 2
then Renew null_resource in module 2

Actual Behavior

Error Cycle

Steps to Reproduce

Please list the full steps required to reproduce the issue, for example:

1. terraform init
2. terraform apply -var description=first module3
3. terraform apply -var description=second module3

Additional Context

tested with terraform 0.12.6 with the same behavior

References

[eddytrex]

Using the lifecycle crate before destroy on the null_resource name, it work fine. But maybe is not positive to use create before destroy in other resources.

[teamterraform]

Thanks for the added info @eddytrex, that might be very useful in tracking this down.
You are correct that you should not need to use create_before_destroy to fix a cycle like this, it just coincidentally happens to break the cycle when destroying the instances.

[eddytrex]

I'm testing in 0.12.7, interesting if there is not count/for_each in the null_resource.hi in the module 2 that depends on the variable resources it works with out cycle error.

[jbardin]

Closed by #22976

[hashibot[bot]]

I'm going to lock this issue because it has been closed for _30 days_ ⏳. This helps our maintainers find and focus on the active issues.

If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.