Terraform Updates for HyperPod Inference Operator v2 Official EKS addon

1.architectures/7.sagemaker-hyperpod-eks/terraform-modules/README.md

@@ -500,6 +500,14 @@ Set the following parameters to `true` in your `custom.tfvars` file to enable op
 | `create_hyperpod_inference_operator_module`  | Installs the [HyperPod inference operator addon](https://docs.aws.amazon.com/sagemaker/latest/dg/sagemaker-hyperpod-model-deployment-setup.html) for deployment and management of machine learning inference endpoints |
 | `create_observability_module` | Installs the [HyperPod Observability addon](https://docs.aws.amazon.com/sagemaker/latest/dg/hyperpod-observability-addon-setup.html) to publish key metrics to Amazon Managed Service for Prometheus and displays them in Amazon Managed Grafana dashboards | 
 
+The HyperPod training and inference operators both require the [cert-manager](https://cert-manager.io/) EKS addon to be installed as a prerequisite. The variable `enable_cert_manager` is set to `true` by default, so that when `create_hyperpod_training_operator_module` or `create_hyperpod_inference_operator_module` are also set to `true`, cert-manager will be installed as a dependency of the operators. In other words, this stack will not install cert-manager as a standalone component, but it can be disabled if you already have it installed on an existing EKS cluster and wish to use one of the HyperPod operators. 
+
+The HyperPod inference operator also has the following additional dependencies: 
+  - The [Amazon FSx for Lustre CSI driver](https://github.com/kubernetes-sigs/aws-fsx-csi-driver): This EKS addon is installed by default as part of the FSx for Lustre module. Set `create_fsx_module = false` if you already have it installed on an existing EKS cluster. 
+  - The [Mountpoint for Amazon S3 CSI Driver](https://github.com/awslabs/mountpoint-s3-csi-driver): This EKS addon is bundled with the Hyperpod inference operator module and is enabled by default. Set `enable_s3_csi_driver = false` if you already have it installed on an existing EKS cluster.
+  - The [AWS Load Balancer Controller](https://github.com/kubernetes-sigs/aws-load-balancer-controller): This is bundled with the HyperPod inference operator EKS addon and is enabled by default. Set `enable_alb_controller = false` if you already have it installed on an existing EKS cluster.
+  - The [KEDA (Kubernetes Event-driven Autoscaling) Operator](https://keda.sh/): This is bundled with the HyperPod inference operator EKS addon and is enabled by default. Set `enable_keda = false` if you already have it installed on an existing EKS cluster.
+
 ---
 ### Advanced Observability Metrics Configuration
 In addition to enabling the [HyperPod Observability addon](https://docs.aws.amazon.com/sagemaker/latest/dg/hyperpod-observability-addon-setup.html) by setting `create_observability_module = true`, you can also configure the following metrics that you wish to collect on your cluster: 

1.architectures/7.sagemaker-hyperpod-eks/terraform-modules/hyperpod-eks-tf/custom.tfvars

@@ -4,17 +4,17 @@ hyperpod_cluster_name = "tf-hp-cluster"
 resource_name_prefix  = "tf-eks-test"
 aws_region            = "us-east-1"
 instance_groups = [
-  {
-    name                      = "accelerated-instance-group-1"
-    instance_type             = "ml.g5.8xlarge",
-    instance_count            = 2,
-    availability_zone_id      = "use1-az2",
-    ebs_volume_size_in_gb     = 100,
-    threads_per_core          = 1,
-    enable_stress_check       = false,
-    enable_connectivity_check = false,
-    lifecycle_script          = "on_create.sh"
-  }
+    {
+        name                      = "accelerated-instance-group-1"
+        instance_type             = "ml.g5.8xlarge",
+        instance_count            = 2,
+        availability_zone_id      = "use1-az2",
+        ebs_volume_size_in_gb     = 100,
+        threads_per_core          = 1,
+        enable_stress_check       = false,
+        enable_connectivity_check = false,
+        lifecycle_script          = "on_create.sh"
+    }
 ]
 create_observability_module               = true
 network_metric_level                      = "ADVANCED"
@@ -23,4 +23,4 @@ create_task_governance_module             = true
 create_hyperpod_training_operator_module  = true
 create_hyperpod_inference_operator_module = true
 enable_guardduty_cleanup                  = true
-create_new_fsx_filesystem                 = true
+create_new_fsx_filesystem                 = true

1.architectures/7.sagemaker-hyperpod-eks/terraform-modules/hyperpod-eks-tf/main.tf

@@ -57,7 +57,7 @@ locals {
   create_s3_bucket_module                   = !local.rig_mode && var.create_s3_bucket_module
   s3_bucket_name                            = !local.rig_mode ? (var.create_s3_bucket_module ? module.s3_bucket[0].s3_bucket_name : var.existing_s3_bucket_name) : null
   create_lifecycle_script_module            = !local.rig_mode && var.create_lifecycle_script_module
-  enable_cert_manager                       = !local.rig_mode && (var.create_hyperpod_training_operator_module || var.create_hyperpod_inference_operator_module) 
+  enable_cert_manager                       = !local.rig_mode && var.enable_cert_manager && (var.create_hyperpod_training_operator_module || var.create_hyperpod_inference_operator_module) 
   wait_for_nodes                            = !local.rig_mode && anytrue(local.features_requiring_nodes)
   create_fsx_module                         = !local.rig_mode ? var.create_fsx_module : false
   create_task_governance_module             = !local.rig_mode && var.create_task_governance_module
@@ -295,14 +295,13 @@ module "hyperpod_inference_operator" {
   source = "./modules/hyperpod_inference_operator"
 
   resource_name_prefix    = var.resource_name_prefix
-  helm_repo_path          = var.helm_repo_path_hpio
-  helm_release_name       = var.helm_release_name_hpio
-  helm_repo_revision      = var.helm_repo_revision_hpio
-  namespace               = var.namespace
   eks_cluster_name        = local.eks_cluster_name
-  vpc_id                  = local.vpc_id
   hyperpod_cluster_arn    = module.hyperpod_cluster[0].hyperpod_cluster_arn
   access_logs_bucket_name = module.s3_bucket[0].s3_logs_bucket_name
+  enable_s3_csi_driver    = var.enable_s3_csi_driver
+  enable_alb_controller   = var.enable_alb_controller
+  enable_keda             = var.enable_keda
+  enable_metrics_server   = var.enable_metrics_server
 
   depends_on = [
     module.hyperpod_cluster,

1.architectures/7.sagemaker-hyperpod-eks/terraform-modules/hyperpod-eks-tf/modules/eks_cluster/main.tf

@@ -1,9 +1,5 @@
 data "aws_region" "current" {}
 
-data "aws_vpc" "selected" {
-  id = var.vpc_id
-}
-
 data "aws_availability_zones" "available" {
   state = "available"
   filter {
@@ -19,7 +15,7 @@ locals {
 
 resource "aws_subnet" "private" {
   count             = var.create_eks_subnets ? length(var.private_subnet_cidrs) : 0
-  vpc_id            = data.aws_vpc.selected.id
+  vpc_id            = var.vpc_id
   cidr_block        = var.private_subnet_cidrs[count.index]
   availability_zone = data.aws_availability_zones.available.names[count.index]
 
@@ -32,7 +28,7 @@ resource "aws_subnet" "private" {
 
 resource "aws_route_table" "eks_private" {
   count  = var.create_eks_subnets ? length(var.private_subnet_cidrs) : 0
-  vpc_id = data.aws_vpc.selected.id
+  vpc_id = var.vpc_id
 
   tags = {
     Name = "${var.resource_name_prefix}-EKS-Private-RT-${count.index + 1}"

1.architectures/7.sagemaker-hyperpod-eks/terraform-modules/hyperpod-eks-tf/modules/fsx_lustre/main.tf

@@ -43,7 +43,7 @@ resource "aws_eks_addon" "fsx_lustre_csi_driver" {
   }
 }
 
-# Wait for FSx CSI driver to be available (required for the HPIO and dynamic provisioning)
+# Wait for FSx CSI driver to be available (required for the HPIO and PVC binding)
 resource "null_resource" "wait_for_fsx_csi_driver" {
   count = local.wait_for_fsx_csi_driver ? 1 : 0