📘 Terraform Series – Day 9

Meta Arguments: count vs for_each with Output

📝 Abstract

In Terraform, managing multiple resources efficiently is a common requirement in real-world DevOps projects. While creating multiple instances may seem straightforward, choosing the right approach can significantly impact scalability, flexibility, and maintainability.

This blog explores Terraform meta arguments, specifically count and for_each, and demonstrates how they behave differently when creating multiple resources. It also covers how outputs change based on these approaches and when to use depends_on for explicit dependencies.

🎯 Objectives

After completing this blog, you will be able to:

• Understand what meta arguments are in Terraform

• Create multiple resources using count

• Identify limitations of count

• Use for_each for better flexibility

• Handle outputs correctly for both approaches

• Understand the use of depends_on

🔷 Step 1: Understanding Meta Arguments

Meta arguments are special Terraform keywords that control resource behavior.

Common Meta Arguments:

• count

• for_each

• depends_on

• lifecycle

👉 In this tutorial, we focus on:

• count

• for_each

🔷 Step 2: Creating Multiple Resources using count

✅ Code

resource "aws_instance" "my_instance" {
  count         = 2
  ami           = var.ec2_ami_id
  instance_type = var.ec2_instance_type
}

🔍 Explanation

• Terraform creates 2 EC2 instances

• Indexed as:

aws_instance.my_instance[0]
aws_instance.my_instance[1]

🔷 Step 3: Output using count

output "ec2_public_ip" {
  value = aws_instance.my_instance[*].public_ip
}

🧠 Concept

• [*] → Splat Expression

• Used to extract values from multiple resources

🔷 Step 4: Limitation of count

Using count has restrictions:

• Same configuration for all instances

• Cannot assign:

  • Different names ❌

  • Different instance types ❌

🔷 Step 6: Using for_each (Better Approach)

resource "aws_instance" "my_instance" {
  for_each = {
    Apurv-1 = "t2.micro"
    Apurv-2 = "t2.medium"
  }

  ami           = var.ec2_ami_id
  instance_type = each.value

  tags = {
    Name = each.key
  }
}

🧠 Explanation

• each.key → Instance Name

• each.value → Instance Type

👉 Each instance becomes unique and configurable

🔷 apply terraform and check output

terraform apply 

🔷 Step 7: Why Old Output Fails

❌ Old Code

aws_instance.my_instance[*].public_ip

❗ Reason

• count → returns list

• for_each → returns map

👉 So [*] does not work here

🔷 Step 8: Correct Output with for_each

✅ Public IP

output "ec2_public_ip" {
  value = [
    for instance in aws_instance.my_instance :
    instance.public_ip
  ]
}

✅ Public DNS

output "ec2_public_dns" {
  value = [
    for instance in aws_instance.my_instance :
    instance.public_dns
  ]
}

✅ Private IP

output "ec2_private_ip" {
  value = [
    for instance in aws_instance.my_instance :
    instance.private_ip
  ]
}

🔷 Step 9: Final Result

After applying for_each:

• Different instance names ✅

• Different instance types ✅

• Correct outputs fetched ✅

🔷 Step 10: Understanding depends_on

✅ Definition

depends_on is used to explicitly define dependencies

⚙️ Example

depends_on = [
  aws_security_group.my_groups,
  aws_key_pair.my_key
]

🧠 When to Use

• When Terraform cannot detect dependency automatically

• When using hardcoded values

• When dependencies are indirect

❌ When NOT to Use

key_name = aws_key_pair.my_key.key_name

👉 Terraform automatically handles this (implicit dependency)

🚀 Conclusion

• Use count → for simple, identical resources

• Use for_each → for flexible and scalable setups

• Understand difference between list vs map outputs

• Use depends_on only when required

👨‍💻 About the Author

“A complete Terraform series covering everything from fundamentals to advanced real-world infrastructure automation in a DevOps environment.”

📬 Let's Stay Connected

• 📧 Email: gujjarapurv181@gmail.com

• 🐙 GitHub: github.com/ApurvGujjar07

• 💼 LinkedIn: linkedin.com/in/apurv-gujjar