How to manage second EBS volume from AMI

1. Import the Existing Data Volume and Attachment:

First, import the existing data volume and its attachment to Terraform. This is crucial to ensure Terraform manages it properly.

terraform import aws_volume_attachment.old_data_volume_attachment vol-xxxxxxxx
terraform import aws_ebs_volume.old_data_volume vol-xxxxxxxx

2. Define the Old and New Data Volumes in Terraform:

Define both the old and new data volumes in your Terraform configuration. Ensure that you create a dependency between the new instance and the detachment of the new data volume.

resource "aws_ebs_volume" "old_data_volume" {
  volume_id = "vol-xxxxxxxx"  # The existing data volume ID
}

resource "aws_volume_attachment" "attach_old_data_volume" {
  device_name = "/dev/sdf"
  volume_id   = aws_ebs_volume.old_data_volume.id
  instance_id = aws_instance.new_instance.id

  depends_on = [aws_volume_attachment.detach_new_data_volume]
}

resource "aws_volume_attachment" "detach_new_data_volume" {
  device_name = "/dev/sdf"
  volume_id   = aws_instance.new_instance.root_block_device[0].volume_id
  instance_id = aws_instance.new_instance.id

  force_detach = true
  lifecycle {
    create_before_destroy = true
  }
}

3. Lifecycle Rules for Detachment:

Use the lifecycle block to control the order of operations. Specifically, the create_before_destroy setting ensures that the detachment of the new data volume occurs before the old data volume is attached.

resource "aws_volume_attachment" "detach_new_data_volume" {
  device_name = "/dev/sdf"
  volume_id   = aws_instance.new_instance.root_block_device[0].volume_id
  instance_id = aws_instance.new_instance.id

  force_detach = true

  lifecycle {
    create_before_destroy = true
  }
}

4. Manage Instance Recreation:

Use the terraform taint command to force the recreation of the instance. Terraform will first detach the new (empty) data volume, then attach the old data volume to the new instance.

terraform taint aws_instance.new_instance
terraform apply

5. Handling Device Name Conflicts:

The depends_on block ensures that the attachment of the old data volume only occurs after the new data volume has been detached, avoiding device name conflicts.

6. Cleanup Unused Volumes:

After the new instance is up and running with the old data volume, you may want to delete the unused new data volume. This can be handled manually or by adding a null_resource or similar mechanism in Terraform.

Example Terraform Configuration

Here's a simplified version of the Terraform configuration:

resource "aws_instance" "new_instance" {
  ami           = "ami-xxxxxxxx"
  instance_type = "t3.medium"

  root_block_device {
    volume_size = 50
  }

  # Other instance configurations
}

resource "aws_ebs_volume" "old_data_volume" {
  volume_id = "vol-xxxxxxxx"  # Replace with your old data volume ID
}

resource "aws_volume_attachment" "attach_old_data_volume" {
  device_name = "/dev/sdf"
  volume_id   = aws_ebs_volume.old_data_volume.id
  instance_id = aws_instance.new_instance.id

  depends_on = [aws_volume_attachment.detach_new_data_volume]
}

resource "aws_volume_attachment" "detach_new_data_volume" {
  device_name = "/dev/sdf"
  volume_id   = aws_instance.new_instance.root_block_device[0].volume_id
  instance_id = aws_instance.new_instance.id

  force_detach = true

  lifecycle {
    create_before_destroy = true
  }
}

Useful Documents https://docs.aws.amazon.com/prescriptive-guidance/latest/backup-recovery/new-ebs-volume-backups.html

https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/creating-an-ami-ebs.html