creating multiple subnets per availability zone using terraform - amazon-web-services

Question

I am new to terraform. I want to create single public subnet and three private subnet per availability zone in particular region in AWS using terraform. I am able to create one private and public subnet per availability zone by referring the following link https://medium.com/#maneetkum/create-subnet-per-availability-zone-in-aws-through-terraform-ea81d1ec1883. However I need to split the one private subet created into another 2. Is that possible in terraform?
data "aws_availability_zones" "available" {}resource "aws_vpc" "myVpc" {
cidr_block = "10.20.0.0/16"
enable_dns_hostnames = true
tags {
Name = "myVpc"
}
}
resource "aws_subnet" "public_subnet" {
count = "${length(data.aws_availability_zones.available.names)}"
vpc_id = "${aws_vpc.myVpc.id}"
cidr_block = "10.20.${10+count.index}.0/24"
availability_zone = "${data.aws_availability_zones.available.names[count.index]}"
map_public_ip_on_launch = true
tags {
Name = "PublicSubnet"
}
}
resource "aws_subnet" "private_subnet" {
count = "${length(data.aws_availability_zones.available.names)}"
vpc_id = "${aws_vpc.myVpc.id}"
cidr_block = "10.20.${20+count.index}.0/24"
availability_zone= "${data.aws_availability_zones.available.names[count.index]}"
map_public_ip_on_launch = false
tags {
Name = "PrivateSubnet"
}
}
The above code is used to create the one private and public subnet per availability zone.

Answer 1

It is possible to create multiple subnets, and automatically place them into an Availability Zone, without duplicating code. Let's keep things DRY. To avoid duplicating the code, use the magic of Terraform meta-arguments and built-in functions. Specifically, use "count" and "cidrsubnet". The "count", will generate as many copies of your subnet as you want.
If you want to provide each subnet with unique values, such as the subnet's tag Name, you can give each subnet a unique, and mnemonic name, by creating a data dictionary with the names that you want assigned to each of the subnets. Then assign them as you create the subnets, by using "count.index". IF that is too much work, you could also just embed the count.index in the name as well.
Different regions have a different number of Availability Zones. To make sure that you are assigning your subnets to Availability Zones that actually exit, you should generate the list of the Availability Zones dynamically. This way, you know that all of the Availability Zones, that are in the list, are actually available in the region that you are working.
What happens if you have more subnets then the region has Availability Zones? Use modulo arithmetic to wrap your working index. Rather than using the index.count directly, do a modulo on the index.count, using the length of the list. This will wrap the index around, so your working index never overflows the length of the list of Availability Zones.
But, the real magic is "cidrsubnet" command. The example below will take the size of the base CIDR block that is passed (which happens to be a /16), add the 2nd parameter (4), and generate a /20 CIDR block. The third parameter indexes through the available CIDR block, thus ensuring that each subnet gets a different sub-CIDR block.
Note: The related cidrsubnets command is quite a bit different. So, be careful, and don't get the two functions confused.
resource "aws_subnet" "area_subnets" {
count = 4 # creates four subnets
vpc_id = var.area_vpc_id
map_public_ip_on_launch = var.map_public_ip_on_launch
cidr_block = cidrsubnet(var.area_subnet_cidr, 4, count.index)
availability_zone_id = data.aws_availability_zones.available.zone_ids[count.index % length(data.aws_availability_zones.available.zone_ids)]
tags = tomap({ "Name" = "${var.subnet_names[count.index]}" })
}
variable "subnet_names" {
type = list(string)
default = [
"Primary NAT Gateway Subnet",
"Secondary NAT Gateway Subnet",
"Channel A Subnet",
"Channel B Subnet"
]
}
variable "map_public_ip_on_launch" {
type = bool
default = true
}
variable "area_vpc_id"
documentation = "The Terraform ID of the containing VPC"
type = string
default = "vpc-abcdefghijklmno"
}
variable "area_subnet_cidr"
documentation = "The base CIDR that you are working with"
type = string
default = "10.0.0.0/16"
}
data "aws_availability_zones" "available" {
state = "available"
filter { # Only fetch Availability Zones (no Local Zones)
name = "opt-in-status"
values = ["opt-in-not-required"]
}
}

Answer 2

You can simply duplicate the private_subnet resource element to create the two new subnets in each AZ:
...
resource "aws_subnet" "private_subnet" {
count = "${length(data.aws_availability_zones.available.names)}"
vpc_id = "${aws_vpc.myVpc.id}"
cidr_block = "10.20.${20+count.index}.0/24"
availability_zone= "${data.aws_availability_zones.available.names[count.index]}"
map_public_ip_on_launch = false
tags {
Name = "PrivateSubnet"
}
}
resource "aws_subnet" "private_subnet_2" {
count = "${length(data.aws_availability_zones.available.names)}"
vpc_id = "${aws_vpc.myVpc.id}"
cidr_block = "10.30.${20+count.index}.0/24"
availability_zone= "${data.aws_availability_zones.available.names[count.index]}"
map_public_ip_on_launch = false
tags {
Name = "PrivateSubnet2"
}
}
You will need to modify the CIDR blocks for each subnet to make sure they don't overlap with each other.