Replacing Existing Infrastructure

Now we'll see that some times we change re-creates resources instead of updates in place. In this case, we'll change the AMI used.

 1 # File cloudcasts.tf
 2  
 3 resource "aws_instance" "cloudcasts_web" {
 4-  ami           = data.aws_ami.app.id 
 5+  ami           = "ami-0a91cd140a1fc148a" 
 6   instance_type = "t3.small"
 7  
 8   root_block_device {
 9     volume_size = 8 # GB
10     volume_type = "gp3"
11   }
12  
13+  lifecycle { 
14+    create_before_destroy = true
15+  } 
16  
17   tags = {
18     Name        = "cloudcasts-staging-web"
19     Project     = "cloudcasts.io"
20     Environment = "Staging"
21     ManagedBy   = "Terraform"
22   }
23 }
24  
25 resource "aws_eip" "cloudcasts_addr" {
26   # We're not doing this directly
27   # instance = aws_instance.cloudcasts_web.id
28   vpc      = true
29  
30   tags = {
31     Name        = "cloudcasts-staging-web-address"
32     Project     = "cloudcasts.io"
33     Environment = "staging"
34     ManagedBy   = "terraform"
35   }
36 }
37  
38 resource "aws_eip_association" "eip_assoc" {
39   instance_id   = aws_instance.cloudcasts_web.id
40   allocation_id = aws_eip.cloudcasts_addr.id
41 }

Note that the EIP remained associated with the new instance. Terraform handled it seamlessly!

We also added a new lifecycle hook:

1lifecycle {
2    create_before_destroy = true
3}

This tells Terraform to create some infrastructure BEFORE destroying the infrastructure it's replacing.

1terraform plan
2terraform apply