Master Terraform Meta-Arguments: count, for_each, depends_on & lifecycle

count: Replicating Resources

The count meta‑argument lets you create multiple identical resource instances based on a numeric value, simplifying bulk deployments such as a set of S3 buckets. Changing the count variable automatically adds or removes resources, and setting it to 0 or 1 can act as a toggle.

variable "bucket_count" {
  description = "Number of S3 buckets to create"
  type        = number
  default     = 3
}

resource "aws_s3_bucket" "example" {
  count  = var.bucket_count
  bucket = "my-app-bucket-${count.index}"
  tags = {
    Name = "My app bucket #${count.index + 1}"
  }
}

for_each: Tagging Resources Uniquely

Unlike count, which treats instances as a list, for_each iterates over a map or set of strings, assigning each instance a unique key. This prevents re‑indexing when an element is removed and is ideal for environment‑specific resources.

resource "aws_s3_bucket" "environment_buckets" {
  for_each = toset(["dev", "qa", "prod"])

  bucket = "my-app-${each.key}-bucket"
  tags = {
    Environment = each.key
  }
}

depends_on: Explicit Dependency Order

Terraform automatically infers many dependencies, but hidden or logical dependencies may require explicit declaration. The depends_on argument forces Terraform to create one resource only after another has been successfully provisioned.

resource "aws_db_instance" "database" {
  # ... database configuration ...
}

resource "aws_instance" "app_server" {
  # ... app server configuration ...
  depends_on = [aws_db_instance.database]
}

Use depends_on sparingly, as overuse can reduce readability and parallelism.

lifecycle: Controlling Resource Lifecycle

The lifecycle block offers fine‑grained control over creation, replacement, and destruction of resources. Key arguments include:

create_before_destroy : When set to true, Terraform creates the new resource before destroying the old one, avoiding service interruption.

prevent_destroy : Setting to true makes Terraform abort if a destroy operation is planned for the resource, protecting critical assets.

ignore_changes : Lists attributes that Terraform should ignore when they drift, useful when external processes modify them.

replace_triggered_by : Triggers a resource replacement whenever any referenced value changes.

resource "aws_instance" "example" {
  # ...
  lifecycle {
    ignore_changes = [
      tags,
    ]
  }
}

resource "aws_appautoscaling_target" "ecs_target" {
  # ...
  lifecycle {
    replace_triggered_by = [
      aws_ecs_service.svc.id
    ]
  }
}

Conclusion

Terraform's meta‑arguments— count, for_each, depends_on, and lifecycle —provide the building blocks for concise, scalable, and reliable infrastructure code. Mastering these directives moves you from merely using Terraform to truly mastering infrastructure as code.