for_each-tf

Small demo of using for_each with resources.

Shout out to Dave Chappelle for uncovering this one.

The tfstate file has been included so you can see what happens when you try to delete a resource in a list. Essentially terraform has to re-index all of the resources which causes alot of unneeded deleting/creating. When you use a for_each instead of count with a set, terraform chooses a better ID for each resource, so when you remove something from the middle of a set it knows exactly what part of the state to target.

Here is an example when we run terraform plan -var 'my_set_of_resources=["resource A", "resource C", "resource D"]' (removing resource B). As you can see the resources created with count all get shuffled and there is alot of unneeded work.

An execution plan has been generated and is shown below.
Resource actions are indicated with the following symbols:
  - destroy
-/+ destroy and then create replacement

Terraform will perform the following actions:

  # null_resource.count[1] must be replaced
-/+ resource "null_resource" "count" {
      ~ id       = "7917027815834816187" -> (known after apply)
      ~ triggers = { # forces replacement
          ~ "name" = "resource B" -> "resource C"
        }
    }

  # null_resource.count[2] must be replaced
-/+ resource "null_resource" "count" {
      ~ id       = "1630229190969434012" -> (known after apply)
      ~ triggers = { # forces replacement
          ~ "name" = "resource C" -> "resource D"
        }
    }

  # null_resource.count[3] will be destroyed
  - resource "null_resource" "count" {
      - id       = "4994212249103566959" -> null
      - triggers = {
          - "name" = "resource D"
        } -> null
    }

  # null_resource.for_each["resource B"] will be destroyed
  - resource "null_resource" "for_each" {
      - id       = "1163336361708868214" -> null
      - triggers = {
          - "name" = "resource B"
        } -> null
    }

Plan: 2 to add, 0 to change, 4 to destroy.