Terraform-provider-aws: credit specification is updated after no changes in the code

[Jonathan-7]

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
• Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
• If you are interested in working on this issue or have submitted a pull request, please leave a comment

Hello,
I ran into this a couple of times today while trying out the t3.micro instances. Even if I didn't change anything or I changed an unrelated resource (e.g the security group name) it would still update the credit specification.

Terraform Version

Terraform v0.11.8

Affected Resource(s)

• aws_instance

Terraform Configuration Files

provider "aws" {
  region = "eu-west-1"
}

resource "aws_instance" "example" {
  ami                    = "ami-0bdb1d6c15a40392c"
  instance_type          = "t3.micro"
  vpc_security_group_ids = ["${aws_security_group.allow_all.id}"]
  key_name = "jons-ubuntu"

  user_data = <<-EOF
              #!/bin/bash
              yum -y install httpd
              systemctl start httpd.service
              systemctl enable httpd
              echo "Hello, World" > /var/www/html/index.html
              EOF

  tags {
    Name = "terraform-example"
  }
}

resource "aws_security_group" "allow_all" {
  name        = "allow_everything"
  description = "Allow all traffic"

  ingress {
    from_port   = 0
    to_port     = 0
    protocol    = "-1"
    cidr_blocks = ["0.0.0.0/0"]
  }
}

Debug Output

https://gist.github.com/Jonathan-7/6c33ec9315158eeb1e4278132e2ced1d

Panic Output

Expected Behavior

Don't update anything.

Actual Behavior

Changed credit specification:
Terraform will perform the following actions:

~ aws_instance.example
credit_specification.0.cpu_credits: "unlimited" => "standard"

Steps to Reproduce

run terraform apply twice

1. terraform apply
2. terraform apply

Important Factoids

References

• #0000

[bflad]

According to the T3 instance product documentation:

T3 instances start in Unlimited mode by default

I believe we default the cpu_credits attribute to standard since that matched the previous behavior with T2 instances.

If you would like to continue to not configure the instance credit_specification (leaving the default), you should be able to workaround this issue with the below until a proper code fix is released:

resource "aws_instance" "example" {
  # ... other configuration ...
  lifecycle {
    ignore_changes = ["credit_specification.0.cpu_credits"]
  }
}

[jgreen-zynga]

This issue seems related to mine:

* aws_instance.jenkins: aws_instance.jenkins: diffs didn't match during apply. This is a bug with Terraform and should be reported as a GitHub Issue.

Please include the following information in your report:

    Terraform Version: 0.11.8
    Resource ID: aws_instance.jenkins
    Mismatch reason: attribute mismatch: credit_specification.0.cpu_credits

Diff One: 
credit_specification.0.cpu_credits":*terraform.ResourceAttrDiff{Old:"unlimited", New:"standard", NewComputed:false, NewRemoved:false, NewExtra:interface {}(nil), RequiresNew:false, Sensitive:false, Type:0x0}

Diff Two:
lacks credit_specification

I can provide the full diff on request but I'll have to clean it up.

This occurs when changing the user_data on an aws_instance with instance_type = t3.medium
resource:

resource "aws_instance" "jenkins" {
  ami             = "ami-09261ff1cf8893662"
  instance_type   = "t2.medium"
  subnet_id       = "subnet-64f47202"
  security_groups = ["${aws_security_group.instance.id}"]
  user_data       = "${file("${path.module}/user_data.sh")}"
}

[nhoughto]

Tried the ignore_changes workaround and are still seeing the diffs didn't match during apply error. So workaround doesn't work in all cases? Going back to t2s in the interim.

[kmotrebski]

I've encountered this with t3.* instances too. My version is Terraform v0.11.1 + provider.aws v1.28.0.

@Jonathan-7 As my workaround I've decided to lock the cpu_credits:

resource "aws_instance" "example" {
  # ...
  credit_specification = {
    cpu_credits = "unlimited"
  }
}

So this is quite similar to ignore_changes solution, or am I wrong here?

[Jonathan-7]

@kmotrebski your workaround should be fine. I expect them to have the credits as unlimited by default in an upcoming release.

[saravanan30erd]

@bflad I will work on this issue.

[rlees85]

not sure if its relevant but I set

  credit_specification = {
    cpu_credits = "standard"
  }

explicitly in code.... but Terraform still builds t3 instances with t.unlimited enabled. Only a second apply then corrects the credit specification

probably related and would be nice to have fixed

[bflad]

The fix for this has been merged into master for the aws_instance resource and will release with version 1.37.0 of the AWS provider, likely tomorrow.

[bflad]

This has been released in version 1.37.0 of the AWS provider. Please see the Terraform documentation on provider versioning or reach out if you need any assistance upgrading.

[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 feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. Thanks!