CRITICAL: aws_instance.root_block_device.volume_size not working in AWS "Govcloud" region

[herbrandson]

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

Terraform Version

Terraform v0.11.7

• provider.aws v1.17.0

Affected Resource(s)

• aws_instance

Terraform Configuration Files

provider "aws" {
  profile = "govcloud"
  region = "us-gov-west-1"
}

resource "aws_vpc" "default" {
    cidr_block = "10.10.0.0/16"
}

resource "aws_subnet" "default" {
    vpc_id = "${aws_vpc.default.id}"
    cidr_block = "10.10.0.0/20"
}

resource "aws_instance" "worker" {
    instance_type = "r4.xlarge"
    key_name = "liveearth"
    subnet_id = "${aws_subnet.default.id}"
    ami = "ami-e0aa3f81"

    root_block_device {
        volume_size = 50
    }
}

Expected Behavior

This should create an instance with a root device of 50gb

Actual Behavior

Terraform creates a new instance with 2 devices.

• /dev/xvda (8 gb), which is the root device
• /dev/xvdb (50 gb), which is not used

Steps to Reproduce

1. terraform apply
2. Run terraform apply a second time and see that Terraform wants to delete and recreate the instance because the root device size is not correct

Important Factoids

This is only happening when run in the AWS "Govcloud" region. The "East 1" regions works fine.

[herbrandson]

Sorry for the "CRITICAL" title, but this is causing huge issues for us right now. We currently can't run our Terraform scripts because it will delete and recreate all of our EC2 instances every time. We had scheduled some downtime last night and rebuilt our entire production cluster. Today, our servers have been crashing due to being out of disk space. I'm having to rebuild them manually one by one :(

Please help

[bflad]

The fix for this will be released in v1.18.0 of the AWS provider, later today. This bug has been present for a very long time but only surfaces under very specific conditions depending on EC2 API caching of different BlockDeviceMappings response ordering from DescribeImages. Some more details about that investigation can be seen in #4489.

[herbrandson]

Thanks so much. You guys are awesome!

[bflad]

This has been released in version 1.18.0 of the AWS provider. Please see the Terraform documentation on provider versioning or reach out if you need any assistance upgrading.

[herbrandson]

I have confirmed that this issue is resolved for us with v1.18.0. Thanks again!

[ghost]

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!