Terraform count to for_each refactor - terraform

Trying to refactor count to for_each and having hard time with this.
local = {
routes = [
{
CIDR = "192.225.0.0/16"
Name = "rt-1"
},
{
CIDR = "192.225.0.0/16"
Name = "rt-2"
},
{
CIDR = "192.225.1.0/16"
Name = "rt-1"
},
{
CIDR = "192.225.1.0/16"
Name = "rt-2"
},
{
CIDR = "192.225.0.0/16"
Name = "rt-3"
}
]
}
Works:
resource "aws_route" "sample" {
count = length(local.routes)
route_table_id = local.routes[count.index].Name
destination_cidr_block = local.routes[count.index].CIDR
}
Don't Work:
resource "aws_route" "sample" {
for_each = local.routes
route_table_id = local.routes[count.index].Name
destination_cidr_block = local.routes[count.index].CIDR
}
Getting this error
The given "for_each" argument value is unsuitable: the "for_each" argument must be a map, or set of strings, and you have provided a value of type tuple.
If I change my local to map then it complains about duplicate, any help.

The for_each meta-argument accepts a map or a set of strings and not list and you are trying to pass a list. You could do something like this:
resource "aws_route" "sample" {
for_each = { for entry in local.routes : "${entry.Name}.${entry.CICR}" => entry }
route_table_id = each.value.Name
destination_cidr_block = each.value.CIDR
}
Read more on for_each: https://www.terraform.io/language/meta-arguments/for_each#basic-syntax

Related

Terraform how to access resource local name as a variable inside the resource block itself

resource "aws_subnet" "goodsubnet" {
vpc_id = "VPCID"
cidr_block = "x.x.x.x/x"
availability_zone = "xyz"
tags =
{
tagname1 = "$something"
}
}
I want the tag "tagname1" to dynamically have the value of the resource local name i.e "goodsubnet" Is there a variable I can use?
Thanks
That is not possible the way you want to do it. However, Terraform has a concept of variables, which you can use to assign values to arguments. That could be either a local value [1] or an input variable [2]. For example, an input variable definition:
variable "subnet_name_tag" {
type = string
description = "Tag name for a subnet."
}
Then, in your code you would do:
resource "aws_subnet" "goodsubnet" {
vpc_id = "VPCID"
cidr_block = "x.x.x.x/x"
availability_zone = "xyz"
tags =
{
tagname1 = var.subnet_name_tag
}
}
Alternatively, you could define a local value:
locals {
subnet_tag_name = "goodsubnet"
}
Followed by:
resource "aws_subnet" "goodsubnet" {
vpc_id = "VPCID"
cidr_block = "x.x.x.x/x"
availability_zone = "xyz"
tags =
{
tagname1 = local.subnet_name_tag
}
}
[1] https://developer.hashicorp.com/terraform/language/values/locals
[2] https://developer.hashicorp.com/terraform/language/values/variables

How to use Terraform for_each resource block and count resource block

I am having a azurerm_postgresql_flexible_server resource using count and azurerm_postgresql_flexible_server_configuration using fore each. Please find the below code.
config= [{
"name" = shared_preload_libraries,
"values" = ["EXAMPLE1", "EXAMPLE2"]
},
{
"name" = "azure.extensions"
"values" = ["EXAMPLE1", "EXAMPLE2", "EXAMPLE3"]
}]
locals {
flat_config = merge([
for single_config in var.config: {
for value in single_config.values:
"${single_config.name}-${value}" => {
"name" = single_config.name
"value" = value
}
}
]...)
}
Below is my for_each resource
resource "azurerm_postgresql_flexible_server_configuration" "example" {
for_each = local.flat_config
name = each.value.name
server_id = azurerm_postgresql_flexible_server.example.id
value = each.value.value
}
currently I am having two azurerm_postgresql_flexible_server resources. azurerm_postgresql_flexible_server.example[0] and azurerm_postgresql_flexible_server.example[1]. Could you please let me know if there is a possibility to include some alternative options like count?
Splat Expressions did not work.
If you want to apply your two azurerm_postgresql_flexible_server for each instance of azurerm_postgresql_flexible_server_configuration, you need one more extra level of flattening:
locals {
flat_config = merge(flatten([
for single_config in var.config: [
for value in single_config.values: {
for idx, server in azurerm_postgresql_flexible_server.example:
"${single_config.name}-${value}-${idx}" => {
"name" = single_config.name
"value" = value
"flexible_server" = server
}
}
]
])...)
}
then
resource "azurerm_postgresql_flexible_server_configuration" "example" {
for_each = local.flat_config
name = each.value.name
server_id = each.value.flexible_server.id
value = each.value.value
}

Multiple Vnets and subnets using terraform modules

I am new to terraform and I am trying to create module for multiple vnets and under that multiple subnets
I am able to create multiple vnets using the module but I am facing issue in creating multiple subnets on one or more of the vnets created.
Below is my code for the module and as well as main.tf
network resources module/main.tf
data "azurerm_resource_group" "network" {
name = var.resource_group_name
}
resource "azurerm_virtual_network" "vnets" {
count = length(var.vnet_names)
name = var.vnet_names[count.index]
resource_group_name = data.azurerm_resource_group.network.name
location = var.vnet_location != null ? var.vnet_location : data.azurerm_resource_group.network.location
address_space = [var.vnet_adress_spaces[count.index]]
}
network resources module/variables.tf
variable "vnet_names" {
description = "Name of the vnets to be created"
type = list(string)
default = ["vnet1","vnet2","vnet3"]
}
variable "vnet_adress_spaces" {
description = "Name of the vnets to be created"
type = list(string)
default = ["192.168.0.0/16" ,"10.0.0.0/16","10.80.0.0/16"]
}
variable "resource_group_name" {
description = "Name of the resource group to be imported."
type = string
}
variable "vnet_location" {
description = "The location of the vnet to create. Defaults to the location of the resource group."
type = string
default = null
}
variable "subnet_names" {
description = "The list of subnets which needs to be created"
type = list(list(string))
default = [[],["subnet1_vnet1","subnet2_vnet1"],["subnet1_vnet3","subnet2_vnet3","subnet3_vnet3"]]
}
variable "subnet_addresses" {
description = "The list of subnets which needs to be created"
type = list(list(string))
default = [[],["10.0.2.0/24","10.0.0.0/24"],["10.80.2.0/24","10.80.1.0/24","10.80.0.0/24"]]
}
main.tf
terraform {
required_providers {
azurerm = {
source = "hashicorp/azurerm"
version = "=2.98.0"
}
}
}
# Configure the Microsoft Azure Provider
provider "azurerm" {
features {}
}
resource "azurerm_resource_group" "rg1" {
name = "rg1"
location = "West Europe"
}
module network {
source = "./network_resources"
resource_group_name = azurerm_resource_group.rg1.name
}
3 vnets are successfully created but I am facing issue in writing subnet creation as mentioned in variables.tf
Please can you help me with that or suggest a better way of implementing this
Output.tf of module
output "vnet_names" {
description = "The name of the virtual networks"
value = {for k, v in azurerm_virtual_network.vnets: k => v.name}
}
output "vnet_addresses" {
description = "The name of the virtual networks"
value = {for k, v in azurerm_virtual_network.vnets: k => v.address_space}
}
output "subnet_names" {
description = "The name of the subnets"
value = {for k, v in azurerm_subnet.subnets: k => v.name}
}
output "subnet_addresses" {
description = "The name of the subnet addresses"
value = {for k, v in azurerm_subnet.subnets: k => v.address_prefixes}
}
output "subnet_ids" {
description = "The name of the subnet addresses"
value = {for k, v in azurerm_subnet.subnets: k => v.id}
}
When I am taking the subnet value same for two vnets as per the updated variables.tf
vnets = {
"mel-dev-identity-vnet01" = {
address_space = "10.0.0.0/16",
subnets = [
{
subnet_name = "subnet-mel-AD-dev"
subnet_address = "10.0.2.0/24"
service_endpoints = []
},
{
subnet_name = "subnet-mel-okt-dev"
subnet_address = "10.0.0.0/24"
service_endpoints = []
},
{
subnet_name = "GatewaySubnet"
subnet_address = "10.0.0.0/26"
service_endpoints = []
},
]
},
"mel-dev-identity-vnet02" = {
address_space = "10.80.0.0/16"
subnets = [
{
subnet_name = "subnet-syd-AD-dev"
subnet_address = "10.80.2.0/24"
service_endpoints = []
},
{
subnet_name = "subnet-syd-okt-dev"
subnet_address = "10.80.1.0/24"
service_endpoints = []
},
{
subnet_name = "GatewaySubnet"
subnet_address = "10.80.0.0/26"
service_endpoints = []
},
]
}
}
I am getting the below error:
│ Error: Duplicate object key
│
│ on network_resources\locals.tf line 11, in locals:
│ 11: subnets = { for subnet in local.subnets_flatlist : subnet.subnet_name => subnet }
│ ├────────────────
│ │ subnet.subnet_name is "GatewaySubnet"
│
│ Two different items produced the key "GatewaySubnet" in this 'for' expression. If duplicates are expected, use the
│ ellipsis (...) after the value expression to enable grouping by key.
I see two issues in your code:
You store attributes (e.g. subnet_name, subnet_addressed) of the same resources in different variables of type list or nested list.
This way you have to ensure consistency across the different variables manually which can become tedious and error prone. Imagine you want to add a third subnet to vnet bupavnet2. You have to make sure that you insert the new name and subnet at the corresponding positions in two nested lists.
BTW: Is it intended that the subnets in vnet bupavnet2 are named subnet#-bupavnet1 ? See what I mean ??? :)
In order to create multiple subnets in multiple vnets dynamically, you would need a nested loop which iterates over the vnets in the outer loop and across the subnets in the inner loop. Terraform, however, does not support nested for_each loops on resource level.
One way to overcome the first issue is to use complex types like objects or maps in order to make the relationship between the attributes of a resource explicit. This way it would be easy to spot the naming issue of the subnets in bupavnet2. If you want to add an additional subnet, you just need to add an additional object to the subnets list.
variable "vnets" {
type = map(object({
address_space = string
subnets = list(object({
subnet_name = string
subnet_address = string
}))
}))
default = {
"bupavnet1" = {
address_space = "192.168.0.0/16",
subnets = []
},
"bupavnet2" = {
address_space = "10.0.0.0/16",
subnets = [
{
subnet_name = "subnet1_bupavnet1"
subnet_address = "10.0.2.0/24"
},
{
subnet_name = "subnet2_bupavnet1"
subnet_address = "10.0.0.0/24"
}
]
},
"bupavnet3" = {
address_space = "10.80.0.0/16"
subnets = [
{
subnet_name = "subnet1_bupavnet3"
subnet_address = "10.80.2.0/24"
},
{
subnet_name = "subnet2_bupavnet3"
subnet_address = "10.80.1.0/24"
},
{
subnet_name = "subnet3_bupavnet3"
subnet_address = "10.80.0.0/24"
},
]
}
}
}
In succession, the creation of the vnets would change to
resource "azurerm_virtual_network" "vnets" {
for_each = var.vnets
name = each.key
resource_group_name = data.azurerm_resource_group.network.name
location = var.vnet_location != null ? var.vnet_location : data.azurerm_resource_group.network.location
address_space = [each.value.address_space]
}
Now let's have a look how to tackle the nested for loop issue. In Terraform you would address this by flattening the nested structure which we defined above. In the first step we create a flat list of objects representing the subnets to be created. So our variable subnets_flatlist has the type tuple of objects.
Unfortunately the for_each argument in Terraform requires the type map or set of strings. So we need a second step to create a map out of our flat list.
locals {
subnets_flatlist = flatten([for key, val in var.vnets : [
for subnet in val.subnets : {
vnet_name = key
subnet_name = subnet.subnet_name
subnet_address = subnet.subnet_address
}
]
])
subnets = { for subnet in local.subnets_flatlist : subnet.subnet_name => subnet }
}
Once we created our helper structures we can iterate over the subnets map and create the subnet resources:
resource "azurerm_subnet" "subnets" {
for_each = local.subnets
name = each.value.subnet_name
resource_group_name = data.azurerm_resource_group.network.name
virtual_network_name = azurerm_virtual_network.vnets[each.value.vnet_name].name
address_prefixes = [each.value.subnet_address]
}

Referencing a created subnet and associate it with a given EIP for a NLB (aws provider)

I'm trying to parametrize the creation of a NLB, and provision in the same plan the necessary public subnets.
The subnets are specified as a variable of the plan:
variable "nlb_public_subnets" {
type = list(object({
name = string
network_number = number
availability_zone = string
elastic_ip = string
}))
default = [
{
name = "sftp_sub_A"
network_number = 1
availability_zone = "eu-west-1a"
elastic_ip = "X.Y.Z.T"
},
{
name = "sftp_sub_B"
network_number = 2
availability_zone = "eu-west-1b"
elastic_ip = "XX.YY.ZZ.TT"
}
]
}
variable "common_tags" {
description = "A map containing the common tags to apply to all resources"
type = map(string)
default = {}
}
locals {
vpc_id = "dummy"
base_cidr = "10.85.23.0/24"
publicSubnets = { for s in var.nlb_public_subnets :
s.name => {
name = s.name
cidr_block = cidrsubnet(var.base_public_subnet_cidr_block, 6,
s.network_number )
availability_zone = s.availability_zone
elastic_ip = s.elastic_ip
}
}
}
I'm specifying a name, a network number (to compute the cidr block), an availability zone, and an elastic IP to map to when creating the NLB.
Here I'm creating the subnets:
#Comment added after solution was given
#This will result in a Map indexed by subnet.name provided in var.nlb_public_subnets
resource "aws_subnet" "sftp_nlb_subnets" {
for_each = { for subnet in local.publicSubnets :
subnet.name => subnet
}
cidr_block = each.value.cidr_block
vpc_id = local.vpc_id
availability_zone = each.value.availability_zone
tags = {
Name = each.key
Visibility = "public"
Purpose = "NLB"
}
}
Now I need to create my NLB, and this is where I'm struggling on how to associate the freshly created subnets with the Elastic IP provided in the configuration:
resource "aws_lb" "sftp" {
name = var.name
internal = false
load_balancer_type = "network"
subnets = [for subnet in aws_subnet.sftp_nlb_subnets: subnet.id]
enable_deletion_protection = true
tags = merge(var.common_tags,{
Name=var.name
})
dynamic "subnet_mapping" {
for_each = aws_subnet.sftp_nlb_subnets
content {
subnet_id = subnet_mapping.value.id
allocation_id = ????Help???
}
}
}
Could I somehow look up the configuration object with the help of the subnet name in the tags?
UPDATE1
Updated the dynamic block, as it had a typo.
UPDATE2
#tmatilai nailed the answer!
Here's the modified aws_lb block:
#
#This will result in a Map indexed by subnet.name provided in var.nlb_public_subnets
data "aws_eip" "nlb" {
for_each = local.publicSubnets
public_ip = each.value.elastic_ip
}
resource "aws_lb" "sftp" {
name = var.name
internal = false
load_balancer_type = "network"
subnets = [for subnet in aws_subnet.sftp_nlb_subnets : subnet.id]
enable_deletion_protection = true
tags = merge(var.common_tags, {
Name = var.name
})
dynamic "subnet_mapping" {
#subnet_mapping.key will contain subnet.name, so we can use it to access the Map data.aws_eip.nlb (also indexed by subnet.name) to get the eip allocation_id
for_each = aws_subnet.sftp_nlb_subnets
content {
subnet_id = subnet_mapping.value.id
allocation_id = data.aws_eip.nlb[subnet_mapping.key].id
}
}
}
The trick is to realize that both aws_subnet.sftp_nlb_subnets and data.aws_eip.nlb are a Map, indexed by the key of local.publicSubnets. This allows us to use this common key (the subnet name) in the map aws_subnet.sftp to look up information in the data (data.aws_eip.nlb) obtained from the original input, local.publicSubnets.
Thanks. This is a neat trick.
Passing the IP address of the elastic IPs sounds strange. If you create the EIPs elsewhere, why not pass the (allocation) ID of them instead?
But with this setup, you can get the allocation ID with the aws_eip data source:
data "aws_eip" "nlb" {
for_each = local.publicSubnets
public_ip = each.value.elastic_ip
}
resource "aws_lb" "sftp" {
# ...
dynamic "subnet_mapping" {
for_each = aws_subnet.sftp_nlb_subnets
content {
subnet_id = subnet_mapping.value.id
allocation_id = data.aws_eip.nlb[subnet_mapping.key].id
}
}
}
But maybe it would make more sense to create the EIPs also here. For example something like this:
resource "aws_eip" "nlb" {
for_each = local.publicSubnets
vpc = true
}
resource "aws_lb" "sftp" {
# ...
dynamic "subnet_mapping" {
for_each = aws_subnet.sftp_nlb_subnets
content {
subnet_id = subnet_mapping.value.id
allocation_id = aws_eip.nlb[subnet_mapping.key].id
}
}
}

Dynamic interpolation of all components of cidrsubnet function in Terraform

Is there a way in Terraform where I can interpolate all three components of cidrsubnet function ? The reason I'm asking this is because I want to create VPCs with different CIDR prefix and accordingly the cidrsubnet function takes care of creating subnets for me.
Supposingly, VPC "prod" has CIDR 10.10.0.0/16 and VPC "dev" has CIDR 10.20.0.0/24. I'm looking for a solution wherein the same code will create subnets for me by interpolating values of cidrsubnet function something like below:
# map for different CIDR
variable "VPC_CIDR" {
type = "map"
default = {
"dev" = "10.10.0.0/24"
"prod" = "10.10.0.0/16"
}
}
variable "PRI_SUBNET_COUNT" {
default = "1"
}
# intended logic which interpolates netnum (impractical code)
if var.VPC_CIDR = "prod"
netnum = 4
elif var.VPC_CIDR = "dev"
netnum = 3
else
netnum = 2
resource "aws_subnet" "sub-node-private" {
count = var.PRI_SUBNET_COUNT
cidr_block = cidrsubnet(var.VPC_CIDR, var.netnum, count.index + 2) #all three components interpolated
The above code will create subnet CIDR 10.10.0.0/20 for prod and 10.10.0.0/28 for dev. This way my code remains same, only the variables get interpolated.
N.B: This code is for demonstration purpose. It is known that this is not the practical code for if/else in Terraform.
If the keys of var.VPC_CIDR are selectable by the caller then it may be best to combine the CIDR prefix and the number of new bits to use for its subnets together in the variable. Since the conventional way to name Terraform variables is in lowercase, I'm going to also rename it to vpc_cidr in the following examples.
variable "vpc_cidr" {
type = map(object({
cidr_block = string
subnet_bits = number
}))
default = {
dev = {
cidr_block = "10.10.0.0/24"
subnet_bits = 3
}
prod = {
cidr_block = "10.10.0.0/16"
subnet_bits = 4
}
}
}
variable "pri_subnet_count" {
type = number
default = 1
}
locals {
vpc_subnets = flatten([
for name, vpc in var.vpc_cidr : [
for i in count(var.pri_subnet_count) : {
name = "${name}-${i}"
vpc_name = name
cidr_block = vpc.cidr_block
subnet_bits = vpc.subnet_bits
network_num = i + 2
}
]
])
}
resource "aws_vpc" "example" {
for_each = var.vpc_cidr
cidr_block = each.value.cidr_block
}
resource "aws_subnet" "private" {
for_each = { for s in local.vpc_subnets : s.name => s }
vpc_id = aws_vpc.example[each.value.vpc_name].id
cidr_block = cidrsubnet(each.value.cidr_block, each.value.subnet_bits, each.value.network_num)
# ...
}
If the names "prod" and "dev" are fixed and thus your module will assume they will always be specified, you can derive the subnet_bits values automatically in a way similar to what you described, like this:
variable "vpc_cidr" {
type = object({
# Force caller to provide "dev" and "prod" values, so
# that it will match up with the attributes in
# local.subnet_bits defined below.
dev = string
prod = string
})
value = {
dev = "10.10.0.0/24"
prod = "10.10.0.0/16"
}
}
variable "pri_subnet_count" {
type = number
default = 1
}
locals {
subnet_bits = {
dev = 3
prod = 4
}
vpcs = {
for name, cidr_block in var.vpc_cidr : name => {
cidr_block = cidr_block
subnet_bits = local.subnet_bits[name]
}
}
vpc_subnets = flatten([
for name, vpc in local.vpcs : [
for i in count(var.pri_subnet_count) : {
name = "${name}-${i}"
vpc_name = name
cidr_block = vpc.cidr_block
subnet_bits = vpc.subnet_bits
network_num = i + 2
}
]
])
}
resource "aws_vpc" "example" {
for_each = local.vpcs
cidr_block = each.value.cidr_block
}
resource "aws_subnet" "private" {
for_each = { for s in local.vpc_subnets : s.name => s }
vpc_id = aws_vpc.example[each.value.vpc_name].id
cidr_block = cidrsubnet(each.value.cidr_block, each.value.subnet_bits, each.value.network_num + 2)
# ...
}
The general pattern illustrated above is building the data structure local.vpc_subnets which contains one element for each subnet you want to create. That then allows repeating aws_subnet.private for each element, and gathers together all of the values required to populate the vpc_id and cidr_block arguments on the subnet.

Resources