fix(infra): share the afterhours VPC in prod (PLAT-215) (#203)

* fix(infra): share the afterhours VPC when existing_vpc_id is set

Prod is at the account VPC quota, so the v1.0.0 apply destroyed Lambda/API Gateway then failed on CreateVpc. Skip creating a sixth VPC when the workspace supplies afterhours subnets.

* style(test): format the VPC terraform assertion for ruff

terraform/data.tf

@@ -68,3 +68,26 @@ check "dev_has_no_custom_domain" {
     error_message = "attach_custom_domain must be false in non-prod; use the CloudFront distribution domain."
   }
 }
+
+check "existing_vpc_pair" {
+  assert {
+    condition     = (var.existing_vpc_id == "") == (length(var.existing_public_subnet_ids) == 0)
+    error_message = "existing_vpc_id and existing_public_subnet_ids must both be set or both be empty."
+  }
+}
+
+check "existing_vpc_two_az" {
+  assert {
+    condition     = var.existing_vpc_id == "" || length(var.existing_public_subnet_ids) >= 2
+    error_message = "existing_public_subnet_ids must include at least two subnets."
+  }
+}
+
+check "existing_subnets_in_vpc" {
+  assert {
+    condition = alltrue([
+      for subnet in data.aws_subnet.existing_public : subnet.vpc_id == var.existing_vpc_id
+    ])
+    error_message = "Every existing_public_subnet_ids value must belong to existing_vpc_id."
+  }
+}

terraform/ecs.tf

@@ -39,7 +39,7 @@ resource "aws_ecr_lifecycle_policy" "api" {
 resource "aws_security_group" "alb" {
   name        = "${local.project}-alb"
   description = "Public ALB for meal-order-manager"
-  vpc_id      = aws_vpc.this.id
+  vpc_id      = local.vpc_id
 
   ingress {
     # CloudFront prefix lists cannot cover GitHub-hosted weekly-menu HMAC
@@ -63,7 +63,7 @@ resource "aws_security_group" "alb" {
 resource "aws_security_group" "api" {
   name        = "${local.project}-api"
   description = "Fargate tasks for meal-order-manager"
-  vpc_id      = aws_vpc.this.id
+  vpc_id      = local.vpc_id
 
   ingress {
     description     = "From ALB"
@@ -86,7 +86,7 @@ resource "aws_lb" "api" {
   load_balancer_type = "application"
   idle_timeout       = 120
   security_groups    = [aws_security_group.alb.id]
-  subnets            = aws_subnet.public[*].id
+  subnets            = local.public_subnet_ids
 
   drop_invalid_header_fields = true
 }
@@ -95,7 +95,7 @@ resource "aws_lb_target_group" "api" {
   name        = "${local.project}-api"
   port        = 8080
   protocol    = "HTTP"
-  vpc_id      = aws_vpc.this.id
+  vpc_id      = local.vpc_id
   target_type = "ip"
 
   health_check {
@@ -202,7 +202,7 @@ resource "aws_ecs_service" "api" {
   launch_type     = "FARGATE"
 
   network_configuration {
-    subnets          = aws_subnet.public[*].id
+    subnets          = local.public_subnet_ids
     security_groups  = [aws_security_group.api.id]
     assign_public_ip = true
   }

terraform/locals.tf

@@ -7,8 +7,8 @@ locals {
 
   github_oidc_provider_arn = "arn:aws:iam::${local.account_id}:oidc-provider/token.actions.githubusercontent.com"
 
-  # Prod has no default VPC. 10.60 is unused in 011934824531
+  # Created only when existing_vpc_id is empty. 10.60 is unused in 011934824531.
-  # (10.0 proposal-system, 10.20 payments-dashboard, 10.40 syslog, 10.80 apm-wo).
+  manage_vpc          = var.existing_vpc_id == ""
   vpc_cidr            = "10.60.0.0/16"
   public_subnet_cidrs = ["10.60.0.0/24", "10.60.1.0/24"]
 

terraform/variables.tf

@@ -91,3 +91,15 @@ variable "checkcomponents_queue_arn" {
   type        = string
   default     = "arn:aws:sqs:us-east-1:011934824531:paychex-checkcomponents"
 }
+
+variable "existing_vpc_id" {
+  description = "When set, place the ALB and Fargate tasks in this VPC instead of creating one."
+  type        = string
+  default     = ""
+}
+
+variable "existing_public_subnet_ids" {
+  description = "Public subnet IDs in existing_vpc_id. Required with existing_vpc_id; ignored when that variable is empty."
+  type        = list(string)
+  default     = []
+}

terraform/vpc.tf

@@ -1,8 +1,21 @@
 data "aws_availability_zones" "available" {
+  count = local.manage_vpc ? 1 : 0
   state = "available"
 }
 
+data "aws_vpc" "existing" {
+  count = local.manage_vpc ? 0 : 1
+  id    = var.existing_vpc_id
+}
+
+data "aws_subnet" "existing_public" {
+  for_each = toset(var.existing_public_subnet_ids)
+  id       = each.value
+}
+
 resource "aws_vpc" "this" {
+  count = local.manage_vpc ? 1 : 0
+
   cidr_block           = local.vpc_cidr
   enable_dns_support   = true
   enable_dns_hostnames = true
@@ -19,7 +32,9 @@ resource "aws_vpc" "this" {
 }
 
 resource "aws_internet_gateway" "this" {
-  vpc_id = aws_vpc.this.id
+  count = local.manage_vpc ? 1 : 0
+
+  vpc_id = aws_vpc.this[0].id
 
   tags = {
     Name = "${local.project}-igw"
@@ -27,11 +42,11 @@ resource "aws_internet_gateway" "this" {
 }
 
 resource "aws_subnet" "public" {
-  count = length(local.public_subnet_cidrs)
+  count = local.manage_vpc ? length(local.public_subnet_cidrs) : 0
 
-  vpc_id                  = aws_vpc.this.id
+  vpc_id                  = aws_vpc.this[0].id
   cidr_block              = local.public_subnet_cidrs[count.index]
-  availability_zone       = data.aws_availability_zones.available.names[count.index]
+  availability_zone       = data.aws_availability_zones.available[0].names[count.index]
   map_public_ip_on_launch = true
 
   tags = {
@@ -40,7 +55,9 @@ resource "aws_subnet" "public" {
 }
 
 resource "aws_route_table" "public" {
-  vpc_id = aws_vpc.this.id
+  count = local.manage_vpc ? 1 : 0
+
+  vpc_id = aws_vpc.this[0].id
 
   tags = {
     Name = "${local.project}-public"
@@ -48,14 +65,46 @@ resource "aws_route_table" "public" {
 }
 
 resource "aws_route" "public_default" {
-  route_table_id         = aws_route_table.public.id
+  count = local.manage_vpc ? 1 : 0
+
+  route_table_id         = aws_route_table.public[0].id
   destination_cidr_block = "0.0.0.0/0"
-  gateway_id             = aws_internet_gateway.this.id
+  gateway_id             = aws_internet_gateway.this[0].id
 }
 
 resource "aws_route_table_association" "public" {
-  count = length(local.public_subnet_cidrs)
+  count = local.manage_vpc ? length(local.public_subnet_cidrs) : 0
 
   subnet_id      = aws_subnet.public[count.index].id
-  route_table_id = aws_route_table.public.id
+  route_table_id = aws_route_table.public[0].id
+}
+
+locals {
+  vpc_id            = local.manage_vpc ? aws_vpc.this[0].id : data.aws_vpc.existing[0].id
+  public_subnet_ids = local.manage_vpc ? aws_subnet.public[*].id : var.existing_public_subnet_ids
+}
+
+moved {
+  from = aws_vpc.this
+  to   = aws_vpc.this[0]
+}
+
+moved {
+  from = aws_internet_gateway.this
+  to   = aws_internet_gateway.this[0]
+}
+
+moved {
+  from = aws_route_table.public
+  to   = aws_route_table.public[0]
+}
+
+moved {
+  from = aws_route.public_default
+  to   = aws_route.public_default[0]
+}
+
+moved {
+  from = data.aws_availability_zones.available
+  to   = data.aws_availability_zones.available[0]
 }

tests/test_terraform_vpc.py

@@ -1,4 +1,4 @@
-"""Meals owns a dedicated VPC. Prod has no default VPC."""
+"""Meals creates a VPC unless existing_vpc_id is set (prod shares afterhours)."""
 
 from pathlib import Path
 
@@ -14,13 +14,22 @@ def test_meals_owns_a_vpc_instead_of_looking_up_default():
     vpc = _read("vpc.tf")
     ecs = _read("ecs.tf")
     locals_tf = _read("locals.tf")
+    variables = _read("variables.tf")
+    data = _read("data.tf")
 
     assert 'resource "aws_vpc" "this"' in vpc
-    assert "cidr_block           = local.vpc_cidr" in vpc
+    assert "count = local.manage_vpc ? 1 : 0" in vpc
-    assert 'vpc_cidr            = "10.60.0.0/16"' in locals_tf
+    assert 'variable "existing_vpc_id"' in variables
+    assert 'variable "existing_public_subnet_ids"' in variables
+    assert 'manage_vpc          = var.existing_vpc_id == ""' in locals_tf
     assert 'data "aws_vpc" "default"' not in ecs
     assert "data.aws_vpc.default" not in ecs
     assert "data.aws_subnets.default" not in ecs
-    assert "aws_vpc.this.id" in ecs
+    assert "vpc_id      = local.vpc_id" in ecs
-    assert "aws_subnet.public[*].id" in ecs
+    assert "subnets            = local.public_subnet_ids" in ecs
+    assert "subnets          = local.public_subnet_ids" in ecs
     assert "ec2:CreateVpc" in _read("hcp_iam.tf")
+    assert 'check "existing_vpc_pair"' in data
+    assert 'check "existing_subnets_in_vpc"' in data
+    assert "from = aws_vpc.this" in vpc
+    assert "to   = aws_vpc.this[0]" in vpc