├── gke ├── data.tf ├── versions.tf ├── providers.tf ├── outputs.tf ├── terraform.tfvars ├── variables.tf ├── main.tf └── README.md ├── .gitignore ├── eks ├── versions.tf ├── setup-kube-config.sh ├── provider.tf ├── data.tf ├── output.tf ├── terraform.tfvars ├── main.tf ├── variables.tf └── README.md ├── aks ├── versions.tf ├── outputs.tf ├── providers.tf ├── terraform.tfvars ├── variables.tf ├── main.tf └── README.md ├── CONTRIBUTING.md ├── README.md └── LICENSE /gke/data.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | data "google_client_config" "provider" {} -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | **/.terraform/* 4 | *.tfstate 5 | *.tfstate.* 6 | crash.log 7 | tfplan 8 | -------------------------------------------------------------------------------- /eks/versions.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | terraform { 5 | required_providers { 6 | aws = { 7 | source = "hashicorp/aws" 8 | version = "~>5.93.0" 9 | } 10 | kubernetes = { 11 | source = "hashicorp/kubernetes" 12 | version = "~>2.19.0" 13 | } 14 | } 15 | 16 | required_version = ">= 1.2.4" 17 | } 18 | -------------------------------------------------------------------------------- /aks/versions.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | terraform { 5 | required_providers { 6 | azurerm = { 7 | source = "hashicorp/azurerm" 8 | version = "~>4.25.0" 9 | } 10 | kubernetes = { 11 | source = "hashicorp/kubernetes" 12 | version = "~>2.19.0" 13 | } 14 | } 15 | 16 | required_version = ">= 1.3.4" 17 | } 18 | 19 | -------------------------------------------------------------------------------- /gke/versions.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | terraform { 4 | required_providers { 5 | google = { 6 | source = "hashicorp/google" 7 | version = "~>6.27.0" 8 | } 9 | google-beta = { 10 | source = "hashicorp/google-beta" 11 | version = "~>6.27.0" 12 | } 13 | kubernetes = { 14 | source = "hashicorp/kubernetes" 15 | version = "~>2.19.0" 16 | } 17 | } 18 | 19 | required_version = ">= 1.3.4" 20 | } 21 | -------------------------------------------------------------------------------- /eks/setup-kube-config.sh: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | #!/bin/bash 5 | 6 | eval "$(jq -r '@sh "_aws_profile=\(.aws_profile) _aws_region=\(.aws_region) _eks_cluster_name=\(.eks_cluster_name) _eks_cluster_arn=\(.eks_cluster_arn)"')" 7 | 8 | if aws --profile "${_aws_profile}" eks update-kubeconfig --region "${_aws_region}" --name "${_eks_cluster_name}" 1> /dev/null 2>&1; then 9 | jq -n --arg command_to_use "kubectl config use-context ${_eks_cluster_arn}" '{"status":"success","commandToUse":$command_to_use}' 10 | else 11 | jq -n '{"status":"failure"}' 12 | fi -------------------------------------------------------------------------------- /eks/provider.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | 5 | provider "aws" { 6 | region = var.aws_region # Ensure this variable is properly set 7 | } 8 | 9 | 10 | provider "helm" { 11 | kubernetes { 12 | host = data.aws_eks_cluster.eks.endpoint 13 | cluster_ca_certificate = base64decode(data.aws_eks_cluster.eks.certificate_authority.0.data) 14 | exec { 15 | api_version = "client.authentication.k8s.io/v1" 16 | args = ["eks", "get-token", "--cluster-name", data.aws_eks_cluster.eks.name] 17 | command = "aws" 18 | } 19 | } 20 | } 21 | -------------------------------------------------------------------------------- /gke/providers.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | provider "kubernetes" { 5 | host = "https://${google_container_cluster.gke.endpoint}" 6 | token = data.google_client_config.provider.access_token 7 | cluster_ca_certificate = base64decode( 8 | google_container_cluster.gke.master_auth.0.cluster_ca_certificate, 9 | ) 10 | } 11 | 12 | provider "helm" { 13 | kubernetes { 14 | token = data.google_client_config.provider.access_token 15 | host = "https://${google_container_cluster.gke.endpoint}" 16 | cluster_ca_certificate = base64decode( 17 | google_container_cluster.gke.master_auth.0.cluster_ca_certificate, 18 | ) 19 | } 20 | } 21 | -------------------------------------------------------------------------------- /aks/outputs.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | output "resource_group_name" { 5 | value = var.existing_resource_group_name == null ? azurerm_resource_group.aks[0].name : data.azurerm_resource_group.existing[0].name 6 | } 7 | 8 | output "kubernetes_cluster_name" { 9 | value = azurerm_kubernetes_cluster.aks.name 10 | } 11 | 12 | output "client_certificate" { 13 | sensitive = true 14 | value = azurerm_kubernetes_cluster.aks.kube_config.0.client_certificate 15 | } 16 | 17 | output "kube_config" { 18 | value = azurerm_kubernetes_cluster.aks.kube_config_raw 19 | sensitive = true 20 | } 21 | 22 | output "location" { 23 | value = azurerm_kubernetes_cluster.aks.location 24 | } 25 | -------------------------------------------------------------------------------- /eks/data.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | data "aws_availability_zones" "available" {} 5 | data "aws_region" "current" {} 6 | 7 | data "aws_ami" "lookup" { 8 | most_recent = true 9 | owners = local.ami_lookup.owners 10 | dynamic "filter" { 11 | for_each = local.ami_lookup.filters 12 | content { 13 | name = filter.value["name"] 14 | values = filter.value["values"] 15 | } 16 | } 17 | } 18 | 19 | data "aws_instances" "nodes" { 20 | filter { 21 | name = "tag:aws:autoscaling:groupName" 22 | values = module.eks.eks_managed_node_groups["gpu_node_pool"]["node_group_autoscaling_group_names"] 23 | } 24 | instance_state_names = ["running"] 25 | } 26 | 27 | data "aws_eks_cluster" "eks" { 28 | name = module.eks.cluster_id 29 | } 30 | -------------------------------------------------------------------------------- /aks/providers.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | provider "azurerm" { 5 | features {} 6 | 7 | subscription_id = var.subscription_id 8 | } 9 | 10 | provider "kubernetes" { 11 | host = azurerm_kubernetes_cluster.aks.kube_admin_config.0.host 12 | client_certificate = base64decode(azurerm_kubernetes_cluster.aks.kube_admin_config.0.client_certificate) 13 | client_key = base64decode(azurerm_kubernetes_cluster.aks.kube_admin_config.0.client_key) 14 | cluster_ca_certificate = base64decode(azurerm_kubernetes_cluster.aks.kube_admin_config.0.cluster_ca_certificate) 15 | } 16 | 17 | provider "helm" { 18 | kubernetes { 19 | host = azurerm_kubernetes_cluster.aks.kube_admin_config.0.host 20 | client_certificate = base64decode(azurerm_kubernetes_cluster.aks.kube_admin_config.0.client_certificate) 21 | client_key = base64decode(azurerm_kubernetes_cluster.aks.kube_admin_config.0.client_key) 22 | cluster_ca_certificate = base64decode(azurerm_kubernetes_cluster.aks.kube_admin_config.0.cluster_ca_certificate) 23 | } 24 | } 25 | -------------------------------------------------------------------------------- /eks/output.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | output "private_subnet_ids" { 5 | value = module.vpc.*.private_subnets 6 | } 7 | 8 | output "public_subnet_ids" { 9 | value = module.vpc.*.public_subnets 10 | } 11 | 12 | output "nodes" { 13 | value = data.aws_instances.nodes.public_ips 14 | } 15 | 16 | output "cluster_endpoint" { 17 | value = module.eks.cluster_endpoint 18 | } 19 | 20 | output "cpu_node_role_name" { 21 | description = "IAM Node Role Bane for CPU node pools" 22 | value = module.eks.eks_managed_node_groups.cpu_node_pool.iam_role_name 23 | } 24 | 25 | output "gpu_node_role_name" { 26 | description = "IAM Node Role Name for GPU node pools" 27 | value = module.eks.eks_managed_node_groups.gpu_node_pool.iam_role_name 28 | } 29 | 30 | output "oidc_endpoint" { 31 | value = module.eks.oidc_provider 32 | } 33 | output "cluster_ca_certificate" { 34 | value = module.eks.cluster_certificate_authority_data 35 | sensitive = true 36 | } 37 | 38 | output "kube_exec_api_version" { 39 | value = "client.authentication.k8s.io/v1beta1" 40 | } 41 | 42 | output "kube_exec_command" { 43 | value = "aws" 44 | } 45 | 46 | output "kube_exec_args" { 47 | value = [ 48 | # "--profile", 49 | # var.aws_profile, 50 | "eks", 51 | "get-token", 52 | "--region", 53 | data.aws_region.current.name, 54 | "--cluster-name", 55 | module.eks.cluster_id 56 | ] 57 | } 58 | -------------------------------------------------------------------------------- /gke/outputs.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | /*************************** 5 | Outputs 6 | ***************************/ 7 | 8 | output "region" { 9 | value = var.region 10 | description = "Region for Kubernetes Resources to be created in when using this module" 11 | } 12 | 13 | output "project_id" { 14 | value = var.project_id 15 | description = "GCloud Project ID" 16 | } 17 | 18 | /*************************** 19 | VPC Network Outputs 20 | ***************************/ 21 | 22 | output "vpc_project" { 23 | value = google_compute_network.gke-vpc[*].project 24 | description = "Project of the VPC network (can be different from the project launching Kubernetes resources)" 25 | } 26 | 27 | output "subnet_cidr_range" { 28 | value = google_compute_subnetwork.gke-subnet[*].ip_cidr_range 29 | description = "The IPs and CIDRs of the subnets" 30 | } 31 | 32 | output "subnet_region" { 33 | value = google_compute_subnetwork.gke-subnet[*].region 34 | description = "The region of the VPC subnet used in this module" 35 | } 36 | /*************************** 37 | GKE Outputs 38 | ***************************/ 39 | output "kubernetes_cluster_name" { 40 | value = google_container_cluster.gke.name 41 | description = "GKE Cluster Name" 42 | } 43 | 44 | output "kubernetes_cluster_endpoint_ip" { 45 | value = google_container_cluster.gke.endpoint 46 | description = "GKE Cluster IP Endpoint" 47 | } 48 | 49 | output "kubernetes_config_file" { 50 | value = google_container_cluster.gke.master_auth.0.cluster_ca_certificate 51 | description = "GKE Cluster IP Endpoint" 52 | sensitive = true 53 | } 54 | -------------------------------------------------------------------------------- /aks/terraform.tfvars: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | # Sample tfvars file. Uncomment out values to use 5 | # Do not commit this file to Git with sensitive values 6 | 7 | 8 | # MANDATORY PARAMETER: 9 | admin_group_object_ids = ["xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"] 10 | subscription_id = "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx" 11 | 12 | # Uncomment and Modify below parameters if needed: 13 | # cluster_name = "aks-cluster-tf" 14 | # kubernetes_version = "1.32" 15 | # location = "eastus" 16 | # cpu_machine_type = "Standard_D16_v5" 17 | # cpu_node_pool_count = 1 18 | # cpu_node_pool_disk_size = 100 19 | # cpu_node_pool_max_count = 5 20 | # cpu_node_pool_min_count = 1 21 | # cpu_os_sku = "Ubuntu" 22 | # existing_resource_group_name = "" 23 | # gpu_machine_type = "Standard_NC4as_T4_v3" 24 | # gpu_node_pool_count = 2 25 | # gpu_node_pool_disk_size = 100 26 | # gpu_node_pool_max_count = 5 27 | # gpu_node_pool_min_count = 1 28 | 29 | 30 | 31 | ######################## 32 | # GPU OPERATOR # 33 | ######################## 34 | 35 | # install_gpu_operator = "true" 36 | # gpu_operator_driver_version = "570.124.06" 37 | # gpu_operator_namespace = "gpu-operator" 38 | # gpu_operator_version = "v25.3.0" 39 | 40 | 41 | ######################## 42 | # NIM OPERATOR # 43 | ######################## 44 | 45 | # install_nim_operator = "false" 46 | # nim_operator_version = "v1.0.1" 47 | # nim_operator_namespace = "nim-operator" 48 | -------------------------------------------------------------------------------- /gke/terraform.tfvars: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | # Sample tfvars file. Uncomment out values to use 5 | # Do not commit this file to Git with sensitive values 6 | 7 | 8 | ######################## 9 | # CLUSTER PARAMETERS # 10 | ######################## 11 | # Useful link: https://cloud.google.com/kubernetes-engine/docs/how-to/gpus#limitations 12 | 13 | # MANDATORY PARAMETER: 14 | project_id = "xx-xxxx-xxxx" 15 | 16 | 17 | # Uncomment and Modify below parameters if needed: 18 | # cluster_name = "gke-cluster-tf" 19 | # min_master_version = "1.32" 20 | # cpu_instance_type = "n1-standard-4" 21 | # cpu_max_node_count = "5" 22 | # cpu_min_node_count = "1" 23 | # disk_size_gb = "512" 24 | # gpu_count = "1" 25 | # gpu_instance_tags = [] 26 | # gpu_instance_type = "n1-standard-4" 27 | # gpu_max_node_count = "5" 28 | # gpu_min_node_count = "2" 29 | # gpu_type = "nvidia-tesla-t4" 30 | # network = "" 31 | # num_cpu_nodes = 1 32 | # num_gpu_nodes = 2 33 | # region = "us-west1" 34 | # node_zones = ["us-west1-b"] 35 | # release_channel = "REGULAR" 36 | # subnetwork = "" 37 | # use_cpu_spot_instances = false 38 | # use_gpu_spot_instances = false 39 | # vpc_enabled = true 40 | 41 | 42 | ######################## 43 | # GPU OPERATOR # 44 | ######################## 45 | 46 | # install_gpu_operator = "true" 47 | # gpu_operator_driver_version = "570.124.06" 48 | # gpu_operator_namespace = "gpu-operator" 49 | # gpu_operator_version = "v25.3.0" 50 | 51 | 52 | ######################## 53 | # NIM OPERATOR # 54 | ######################## 55 | 56 | # install_nim_operator = "false" 57 | # nim_operator_version = "v1.0.1" 58 | # nim_operator_namespace = "nim-operator" 59 | 60 | -------------------------------------------------------------------------------- /eks/terraform.tfvars: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | # Sample tfvars file. Uncomment out values to use 5 | # Do not commit this file to Git with sensitive values 6 | 7 | 8 | ######################## 9 | # CLUSTER PARAMETERS # 10 | ######################## 11 | # For Instances refer https://docs.aws.amazon.com/dlami/latest/devguide/gpu.html 12 | 13 | 14 | # cluster_name = "eks-cluster-tf" 15 | # cluster_version = "1.32" 16 | # aws_region = "us-west-2" 17 | 18 | 19 | # additional_node_security_groups_rules = {} 20 | # additional_security_group_ids = [] 21 | # additional_user_data = "" 22 | # aws_profile = "development" 23 | # cidr_block = "10.0.0.0/16" 24 | # cpu_instance_type = "t2.xlarge" 25 | # cpu_node_pool_additional_user_data = "" 26 | # cpu_node_pool_delete_on_termination = true 27 | # cpu_node_pool_root_disk_size_gb = 512 28 | # cpu_node_pool_root_volume_type = "gp2" 29 | # desired_count_cpu_nodes = "1" 30 | # desired_count_gpu_nodes = "2" 31 | # enable_dns_hostnames = true 32 | # enable_dns_support = true 33 | # enable_nat_gateway = true 34 | # existing_vpc_details = "" 35 | # gpu_ami_id = "" 36 | # gpu_instance_type = "g4dn.2xlarge" 37 | # gpu_node_pool_additional_user_data = "" 38 | # gpu_node_pool_delete_on_termination = true 39 | # gpu_node_pool_root_disk_size_gb = 512 40 | # gpu_node_pool_root_volume_type = "gp2" 41 | # max_cpu_nodes = "2" 42 | # max_gpu_nodes = "5" 43 | # min_cpu_nodes = "0" 44 | # min_gpu_nodes = "1" 45 | # private_subnets = [ 46 | # "10.0.0.0/19", 47 | # "10.0.32.0/19", 48 | # "10.0.64.0/19" 49 | # ] 50 | # public_subnets = [ 51 | # "10.0.96.0/19", 52 | # "10.0.128.0/19", 53 | # "10.0.160.0/19" 54 | # ] 55 | # single_nat_gateway = false 56 | # ssh_key = "" 57 | 58 | 59 | ######################## 60 | # GPU OPERATOR # 61 | ######################## 62 | 63 | #install_gpu_operator = "true" 64 | #gpu_operator_driver_version = "570.124.06" 65 | #gpu_operator_namespace = "gpu-operator" 66 | #gpu_operator_version = "v25.3.0" 67 | 68 | 69 | ######################## 70 | # NIM OPERATOR # 71 | ######################## 72 | 73 | # install_nim_operator = "false" 74 | # nim_operator_version = "v1.0.0" 75 | # nim_operator_namespace = "nim-operator" 76 | 77 | -------------------------------------------------------------------------------- /aks/variables.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | /**************************** 5 | Azure Resource Group Variables 6 | ****************************/ 7 | 8 | variable "existing_resource_group_name" { 9 | description = "The name of an existing resource group the Kubernetes cluster should be deployed into. Defaults to the name of the cluster + `-rg` if none is specified" 10 | default = null 11 | type = string 12 | } 13 | 14 | variable "location" { 15 | description = "The region to create resources in" 16 | default = "eastus" 17 | } 18 | 19 | /**************************** 20 | AKS Variables 21 | ****************************/ 22 | 23 | variable "cluster_name" { 24 | default = "aks-cluster-tf" 25 | description = "The name of the AKS Cluster to be created" 26 | } 27 | 28 | variable "kubernetes_version" { 29 | default = "1.32" 30 | description = "Version of Kubernetes to turn on. Run 'az aks get-versions --location --output table' to view all available versions " 31 | } 32 | 33 | variable "cpu_node_pool_disk_size" { 34 | description = "Disk size in GB of nodes in the Default GPU pool" 35 | default = 100 36 | } 37 | 38 | variable "cpu_node_pool_count" { 39 | description = "Count of nodes in Default GPU pool" 40 | default = 1 41 | } 42 | 43 | variable "cpu_node_pool_min_count" { 44 | description = "Min ount of number of nodes in Default CPU pool" 45 | default = 1 46 | } 47 | variable "cpu_node_pool_max_count" { 48 | description = "Max count of nodes in Default CPU pool" 49 | default = 5 50 | } 51 | variable "cpu_machine_type" { 52 | default = "Standard_D16_v5" 53 | description = "Machine instance type of the AKS CPU node pool" 54 | } 55 | variable "cpu_os_sku" { 56 | description = "Specifies the OS SKU used by the agent pool. Possible values include: Ubuntu, CBLMariner, Mariner, Windows2019, Windows2022" 57 | default = "Ubuntu" 58 | } 59 | 60 | /**************************** 61 | GPU Node Pool Variables 62 | ****************************/ 63 | variable "gpu_node_pool_disk_size" { 64 | description = "Disk size in GB of nodes in the Default GPU pool" 65 | default = 100 66 | } 67 | variable "gpu_node_pool_count" { 68 | description = "Count of nodes in Default GPU pool" 69 | default = 2 70 | } 71 | variable "gpu_node_pool_min_count" { 72 | description = "Min count of number of nodes in Default GPU pool" 73 | default = 2 74 | } 75 | variable "gpu_node_pool_max_count" { 76 | description = "Max count of nodes in Default GPU pool" 77 | default = 5 78 | } 79 | variable "gpu_machine_type" { 80 | default = "Standard_NC4as_T4_v3" 81 | description = "Machine instance type of the AKS GPU node pool" 82 | } 83 | variable "gpu_os_sku" { 84 | description = "Specifies the OS SKU used by the agent pool. Possible values include: Ubuntu, CBLMariner, Mariner, Windows2019, Windows2022" 85 | default = "Ubuntu" 86 | } 87 | /**************************** 88 | GPU Operator Variables 89 | ****************************/ 90 | variable "install_gpu_operator" { 91 | default = "true" 92 | description = "Whether to Install GPU Operator. Defaults to false available." 93 | } 94 | 95 | variable "gpu_operator_version" { 96 | default = "v25.3.0" 97 | description = "Version of the GPU operator to be installed" 98 | } 99 | 100 | variable "gpu_operator_namespace" { 101 | type = string 102 | default = "gpu-operator" 103 | description = "The namespace to deploy the NVIDIA GPU operator into" 104 | } 105 | 106 | variable "gpu_operator_driver_version" { 107 | type = string 108 | default = "570.124.06" 109 | description = "The NVIDIA Driver version deployed with GPU Operator. Defaults to latest available." 110 | } 111 | 112 | /************************ 113 | NIM Operator Variables 114 | *************************/ 115 | variable "install_nim_operator" { 116 | default = "false" 117 | description = "Whether to Install NIM Operator. Defaults to false available." 118 | } 119 | 120 | variable "nim_operator_version" { 121 | default = "v1.0.1" 122 | description = "Version of the GPU Operator to deploy. Defaults to latest available." 123 | } 124 | 125 | variable "nim_operator_namespace" { 126 | type = string 127 | default = "nim-operator" 128 | description = "The namespace for the GPU operator deployment" 129 | } 130 | 131 | /**************************** 132 | Active Directory Variables 133 | ****************************/ 134 | variable "admin_group_object_ids" { 135 | type = list(any) 136 | description = < - 28 | 29 | 30 | ``` 31 | 32 | - Make sure that you can contribute your work to open source (no license and/or patent conflict is introduced by your code). You will need to [`sign`](#signing-your-work) your commit. 33 | 34 | - Thanks in advance for your patience as we review your contributions; we do appreciate them! 35 | 36 | 37 | #### Pull Requests 38 | Developer workflow for code contributions is as follows: 39 | 40 | 1. Developers must first [fork](https://help.github.com/en/articles/fork-a-repo) the [upstream](https://github.com/nvidia/nvidia-terraform-modules) NVIDIA Terraform Modules repository. 41 | 42 | 2. Git clone the forked repository and push changes to the personal fork. 43 | 44 | ```bash 45 | git clone https://github.com/YOUR_USERNAME/YOUR_FORK.git NVIDIAK8s 46 | # Checkout the targeted branch and commit changes 47 | # Push the commits to a branch on the fork (remote). 48 | git push -u origin : 49 | ``` 50 | 51 | 3. Once the code changes are staged on the fork and ready for review, a [Pull Request](https://help.github.com/en/articles/about-pull-requests) (PR) can be [requested](https://help.github.com/en/articles/creating-a-pull-request) to merge the changes from a branch of the fork into a selected branch of upstream. 52 | * Exercise caution when selecting the source and target branches for the PR. 53 | Note that versioned releases of TensorRT OSS are posted to `release/` branches of the upstream repo. 54 | * Creation of a PR creation kicks off the code review process. 55 | * Atleast one TensorRT engineer will be assigned for the review. 56 | * While under review, mark your PRs as work-in-progress by prefixing the PR title with [WIP]. 57 | 58 | 4. Since there is no CI/CD process in place yet, the PR will be accepted and the corresponding issue closed only after adequate testing has been completed, manually, by the developer and/or NVIDIA engineer reviewing the code. 59 | 60 | 61 | #### Signing Your Work 62 | 63 | * We require that all contributors "sign-off" on their commits. This certifies that the contribution is your original work, or you have rights to submit it under the same license, or a compatible license. 64 | 65 | * Any contribution which contains commits that are not Signed-Off will not be accepted. 66 | 67 | * To sign off on a commit you simply use the `--signoff` (or `-s`) option when committing your changes: 68 | ```bash 69 | $ git commit -s -m "Add cool feature." 70 | ``` 71 | This will append the following to your commit message: 72 | ``` 73 | Signed-off-by: Your Name 74 | ``` 75 | 76 | * Full text of the DCO: 77 | 78 | ``` 79 | Developer Certificate of Origin 80 | Version 1.1 81 | 82 | Copyright (C) 2004, 2006 The Linux Foundation and its contributors. 83 | 1 Letterman Drive 84 | Suite D4700 85 | San Francisco, CA, 94129 86 | 87 | Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. 88 | ``` 89 | 90 | ``` 91 | Developer's Certificate of Origin 1.1 92 | 93 | By making a contribution to this project, I certify that: 94 | 95 | (a) The contribution was created in whole or in part by me and I have the right to submit it under the open source license indicated in the file; or 96 | 97 | (b) The contribution is based upon previous work that, to the best of my knowledge, is covered under an appropriate open source license and I have the right under that license to submit that work with modifications, whether created in whole or in part by me, under the same open source license (unless I am permitted to submit under a different license), as indicated in the file; or 98 | 99 | (c) The contribution was provided directly to me by some other person who certified (a), (b) or (c) and I have not modified it. 100 | 101 | (d) I understand and agree that this project and the contribution are public and that a record of the contribution (including all personal information I submit with it, including my sign-off) is maintained indefinitely and may be redistributed consistent with this project or the open source license(s) involved. 102 | ``` -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # NVIDIA Terraform Kubernetes Modules 2 | 3 | ## Objective 4 | 5 | NVIDIA Terraform Modules provide a reference architecture for deploying CSP managed Kubernetes clusters equipped with NVIDIA softwares: 6 | 7 | - NVIDIA GPU Operator. 8 | - NVIDIA NIM Operator. 9 | 10 | All the components listed below have been tested successfully together. 11 | 12 | 13 | ## Life Cycle 14 | 15 | When NVIDIA Terraform Modules is released, the previous release enters maintenance support and only receives patch release updates. All prior batches enter end-of-life (EOL) and are no longer supported and do not receive patch updates. 16 | 17 | 18 | | Release | Status | 19 | | :-----: | :--------------:| 20 | | [25.4.0](https://github.com/NVIDIA/cloud-native-stack/releases/tag/v25.4.0) | Generally Available | 21 | | [24.11.0](https://github.com/NVIDIA/cloud-native-stack/releases/tag/v24.11.0) | Maintenance | 22 | 23 | 24 | ## Support Matrix 25 | 26 | The Kubernetes clusters provisioned by the modules in this repository provide tested and certified versions of Kubernetes, the NVIDIA GPU operator, and NVIDIA NIM Operator. 27 | 28 | If your application does not require a specific version of Kubernetes, we recommend using the latest available version. We also recommend you plan to upgrade your version of Kubernetes at least every 6 months. 29 | 30 | 31 | NVIDIA Terraform Modules 25.4.0 Release: 32 | 33 | | TF Modules | K8s 1.32 | K8s 1.31 | K8s 1.30 | 34 | | :--------- | :-------- | :------- | :------- | 35 | | Platforms | Amazon EKS
Azure AKS
Google GKE | Amazon EKS
Azure AKS
Google GKE | Amazon EKS
Azure AKS
Google GKE | 36 | | Supported OS | Ubuntu 24.04 LTS (GKE)
Ubuntu 22.04 LTS (AKS/EKS) | Ubuntu 24.04 LTS (GKE)
Ubuntu 22.04 LTS (AKS/EKS) | Ubuntu 24.04 LTS (GKE)
Ubuntu 22.04 LTS (AKS/EKS) | 37 | | Kernel | EKS: 6.8.0-1024-aws
AKS: 5.15.0-1082-azure
GKE: 6.8.0-1017-gke | EKS: 6.8.0-1024-aws
AKS: 5.15.0-1082-azure
GKE: 6.8.0-1017-gke | EKS: 6.8.0-1024-aws
AKS: 5.15.0-1082-azure
GKE: 6.8.0-1017-gke | 38 | | Containerd | EKS: 1.7.24
AKS: 1.7.26-1
GKE: 1.7.24 | EKS: 1.7.24
AKS: 1.7.26-1
GKE: 1.7.24 | EKS: 1.7.24
AKS: 1.7.26-1
GKE: 1.7.24 | 39 | | CNI | CSP dependent | CSP dependent | CSP dependent | 40 | | CSI | CSP dependent | CSP dependent | CSP dependent | 41 | | NVIDIA GPU Operator | 25.3.0 | 25.3.0 | 25.3.0 | 42 | | NVIDIA GPU Operator Operands | NVIDIA Container Toolkit: 1.17.5
NVIDIA Device Plugin: 0.17.1
NVIDIA MIG Manager: 0.12.1
NVIDIA DCGM Exporter: 4.1.1-4.0.4 | NVIDIA Container Toolkit: 1.17.5
NVIDIA Device Plugin: 0.17.1
NVIDIA MIG Manager: 0.12.1
NVIDIA DCGM Exporter: 4.1.1-4.0.4 | NVIDIA Container Toolkit: 1.17.5
NVIDIA Device Plugin: 0.17.1
NVIDIA MIG Manager: 0.12.1
NVIDIA DCGM Exporter: 4.1.1-4.0.4 | 43 | | NVIDIA DataCenter Driver | 570.124.06 | 570.124.06 | 570.124.06 | 44 | | NVIDIA NIM Operator | 1.0.1 | 1.0.1 | 1.0.1 | 45 | | Helm | 3.17.2 | 3.17.2 | 3.17.2 | 46 | 47 | 48 | ### CSP Managed K8s Services Life Cycle 49 | 50 | Each CSP has its own end of life date for the versions of Kubernetes they support. For more information see: 51 | 52 | - [Amazon EKS release calendar](https://docs.aws.amazon.com/eks/latest/userguide/kubernetes-versions.html#kubernetes-release-calendar) 53 | - [Azure AKS release calendar](https://learn.microsoft.com/en-us/azure/aks/supported-kubernetes-versions?tabs=azure-cli#aks-kubernetes-release-calendar) 54 | - [GCP GKE release calendar](https://cloud.google.com/kubernetes-engine/docs/release-schedule#schedule_for_static_no-channel_versions). 55 | 56 | 57 | ## Getting Started 58 | 59 | Infrastructure as code for GPU accelerated managed Kubernetes clusters. These scripts automate the deployment of GPU-Enabled Kubernetes clusters on various cloud service platforms. 60 | 61 | Terraform is an open-source infrastructure as code software tool that we will use to automate the deployment of Kubernetes clusters with the required add-ons to enable NVIDIA GPUs. This repository contains Terraform [modules](https://developer.hashicorp.com/terraform/tutorials/modules/module), which are sets of Terraform configuration files ready for deployment. The modules in this repository can be incorporated into existing Terraform-managed infrastructure, or used to set up new infrastructure from scratch. You can learn more about Terraform [here](https://developer.hashicorp.com/terraform/tutorials/aws-get-started/infrastructure-as-code). 62 | 63 | You can download Terraform (CLI) [here](https://developer.hashicorp.com/terraform/downloads). 64 | 65 | ### Usage 66 | 67 | Clone the repo 68 | 69 | ``` 70 | git clone https://github.com/NVIDIA/nvidia-terraform-modules.git 71 | ``` 72 | 73 | #### Provision a GPU enabled Kubernetes Cluster 74 | 75 | Select the CSP managed K8s cluster and follow steps indicated in the corresponding page: 76 | 77 | - Create an [EKS Cluster](./eks/README.md) 78 | - Create an [AKS Cluster](./aks/README.md) 79 | - Create a [GKE Cluster](./gke/README.md) 80 | 81 | 82 | ### State Management 83 | These modules do not set up state management for the generated Terraform state file, deleting the statefile (`terraform.tfstate`) generated by Terraform could result in cloud resources needing to be manually deleted. We strongly encourage you [configure remote state](https://developer.hashicorp.com/terraform/language/state/remote). 84 | 85 | Please see the [Terraform Documentation](https://developer.hashicorp.com/terraform/language/state) for more information. 86 | 87 | ## Contributing 88 | 89 | Pull requests are welcome! Please see our [contribution guidelines](./CONTRIBUTING.md). 90 | 91 | ## Getting help or Providing feedback 92 | 93 | Please open an [issue](https://github.com/NVIDIA/nvidia-terraform-modules/issues) on the GitHub project for any questions. Your feedback is appreciated. 94 | 95 | 96 | ## Useful Links 97 | - [NVIDIA GPU Operator](https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/overview.html) 98 | - [NVIDIA NIM Operator](https://docs.nvidia.com/nim-operator/latest/index.html) 99 | - [NVIDIA GPU Cloud (NGC)](https://catalog.ngc.nvidia.com/) 100 | -------------------------------------------------------------------------------- /eks/main.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | /******************************************** 5 | Network Config 6 | ********************************************/ 7 | module "vpc" { 8 | source = "terraform-aws-modules/vpc/aws" 9 | version = "4.0.2" 10 | count = var.existing_vpc_details == null ? 1 : 0 11 | name = "tf-${var.cluster_name}-vpc" 12 | cidr = var.cidr_block 13 | azs = data.aws_availability_zones.available.names 14 | # FUTURE: Make configurable, or set statically for the max number of pods a cluster can handle 15 | private_subnets = var.private_subnets 16 | public_subnets = var.public_subnets 17 | enable_nat_gateway = var.enable_nat_gateway 18 | single_nat_gateway = var.single_nat_gateway # Future: Revisit the VPC defaults 19 | enable_dns_hostnames = var.enable_dns_hostnames 20 | map_public_ip_on_launch = true 21 | } 22 | 23 | 24 | /******************************************** 25 | Kubernetes Cluster Configuration 26 | ********************************************/ 27 | 28 | locals { 29 | node_security_group_additional_rules = { 30 | ingress_self_all = { 31 | description = "Node to node ingress, no external ingress" 32 | protocol = "-1" 33 | from_port = 0 34 | to_port = 0 35 | type = "ingress" 36 | self = true 37 | } 38 | 39 | egress_all = { 40 | description = "Node egress to open internet" 41 | protocol = "-1" 42 | from_port = 0 43 | to_port = 0 44 | type = "egress" 45 | cidr_blocks = ["0.0.0.0/0"] 46 | ipv6_cidr_blocks = ["::/0"] 47 | } 48 | } 49 | } 50 | 51 | module "eks" { 52 | source = "terraform-aws-modules/eks/aws" 53 | version = "18.29.0" 54 | cluster_name = "tf-${var.cluster_name}" 55 | cluster_version = var.cluster_version 56 | cluster_endpoint_private_access = true 57 | cluster_endpoint_public_access = true 58 | create_cloudwatch_log_group = false 59 | vpc_id = var.existing_vpc_details == null ? module.vpc[0].vpc_id : var.existing_vpc_details.vpc_id 60 | enable_irsa = true 61 | subnet_ids = var.existing_vpc_details == null ? module.vpc[0].private_subnets : var.existing_vpc_details.subnet_ids 62 | control_plane_subnet_ids = var.existing_vpc_details == null ? module.vpc[0].private_subnets : var.existing_vpc_details.subnet_ids 63 | # KMS Config 64 | create_kms_key = true 65 | enable_kms_key_rotation = true 66 | kms_key_deletion_window_in_days = 7 67 | kms_key_enable_default_policy = true 68 | cluster_encryption_config = [ 69 | { 70 | resources : ["secrets"] 71 | } 72 | ] 73 | # Cluster Security Group 74 | cluster_security_group_additional_rules = { 75 | egress_nodes_ephemeral_ports_tcp = { 76 | description = "Control plane egress to nodes on TCP Ports 1025-65535" 77 | protocol = "tcp" 78 | from_port = 1025 79 | to_port = 65535 80 | type = "egress" 81 | source_node_security_group = true 82 | } 83 | } 84 | # NodeGroup Config 85 | node_security_group_additional_rules = merge(local.node_security_group_additional_rules, var.additional_node_security_groups_rules) 86 | 87 | eks_managed_node_groups = { 88 | gpu_node_pool = { 89 | name = "tf-gpu" 90 | instance_types = [var.gpu_instance_type] 91 | min_size = var.min_gpu_nodes 92 | max_size = var.max_gpu_nodes 93 | desired_size = var.desired_count_gpu_nodes 94 | ami_id = data.aws_ami.lookup.id 95 | ami_type = "CUSTOM" 96 | enable_bootstrap_user_data = true 97 | post_bootstrap_user_data = var.gpu_node_pool_additional_user_data 98 | vpc_security_group_ids = var.existing_vpc_details == null ? [] : var.additional_security_group_ids 99 | block_device_mappings = { 100 | root = { 101 | device_name = data.aws_ami.lookup.root_device_name 102 | ebs = { 103 | volume_size = var.gpu_node_pool_root_disk_size_gb 104 | volume_type = var.gpu_node_pool_root_volume_type 105 | delete_on_termination = var.gpu_node_pool_delete_on_termination 106 | } 107 | } 108 | } 109 | ssh_key = var.ssh_key 110 | 111 | }, 112 | cpu_node_pool = { 113 | name = "tf-cpu" 114 | instance_types = [var.cpu_instance_type] 115 | min_size = var.min_cpu_nodes 116 | max_size = var.max_cpu_nodes 117 | desired_size = var.desired_count_cpu_nodes 118 | vpc_security_group_ids = var.existing_vpc_details == null ? [] : var.additional_security_group_ids 119 | ssh_key = var.ssh_key 120 | block_device_mappings = { 121 | root = { 122 | device_name = "/dev/xvda" 123 | ebs = { 124 | volume_size = var.cpu_node_pool_root_disk_size_gb 125 | volume_type = var.cpu_node_pool_root_volume_type 126 | delete_on_termination = var.cpu_node_pool_delete_on_termination 127 | } 128 | } 129 | } 130 | } 131 | } 132 | # Cluster Add-on Config 133 | cluster_addons = { 134 | aws-ebs-csi-driver = { 135 | service_account_role_arn = module.ebs_csi_irsa_role.iam_role_arn 136 | most_recent = true 137 | } 138 | } 139 | } 140 | 141 | /******************************************** 142 | Configure AWS Role for Service Account 143 | for EKS CSI Driver 144 | ********************************************/ 145 | module "ebs_csi_irsa_role" { 146 | source = "terraform-aws-modules/iam/aws//modules/iam-role-for-service-accounts-eks" 147 | role_name = "${var.cluster_name}-ebs-csi" 148 | attach_ebs_csi_policy = true 149 | oidc_providers = { 150 | cluster = { 151 | provider_arn = module.eks.oidc_provider_arn 152 | namespace_service_accounts = ["kube-system:ebs-csi-controller-sa"] 153 | } 154 | } 155 | } 156 | 157 | /******************************************** 158 | Custom AMI Lookup 159 | ********************************************/ 160 | locals { 161 | ubuntu_ami_lookup = { 162 | owners = ["099720109477"] # Canonical 163 | filters = [ 164 | { 165 | name = "name" 166 | values = ["ubuntu-eks/k8s_${var.cluster_version}/images/hvm-ssd/ubuntu-jammy-22.04-amd64-server-*"] 167 | }, 168 | { 169 | name = "virtualization-type" 170 | values = ["hvm"] 171 | } 172 | ] 173 | } 174 | no_ami_lookup = { 175 | owners = [] 176 | filters = [] 177 | } 178 | ami_lookup = var.gpu_ami_id == "" ? local.ubuntu_ami_lookup : local.no_ami_lookup 179 | ami_id = var.gpu_ami_id == "" ? data.aws_ami.lookup.id : var.gpu_ami_id 180 | } 181 | 182 | 183 | /******************************************** 184 | GPU Operator Configuration 185 | ********************************************/ 186 | resource "helm_release" "gpu_operator" { 187 | count = var.install_gpu_operator ? 1 : 0 188 | name = "gpu-operator" 189 | repository = "https://helm.ngc.nvidia.com/nvidia" 190 | chart = "gpu-operator" 191 | version = var.gpu_operator_version 192 | namespace = var.gpu_operator_namespace 193 | create_namespace = true 194 | atomic = true 195 | cleanup_on_fail = true 196 | reset_values = true 197 | replace = true 198 | 199 | set { 200 | name = "driver.version" 201 | value = var.gpu_operator_driver_version 202 | } 203 | 204 | } 205 | 206 | /******************************************** 207 | NIM Operator Configuration 208 | ********************************************/ 209 | resource "helm_release" "nim_operator" { 210 | count = var.install_nim_operator ? 1 : 0 211 | name = "nim-operator" 212 | repository = "https://helm.ngc.nvidia.com/nvidia" 213 | chart = "k8s-nim-operator" 214 | version = var.nim_operator_version 215 | namespace = var.nim_operator_namespace 216 | create_namespace = true 217 | atomic = true 218 | cleanup_on_fail = true 219 | reset_values = true 220 | replace = true 221 | } 222 | -------------------------------------------------------------------------------- /eks/variables.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | /************************ 5 | AWS Variables 6 | *************************/ 7 | 8 | variable "aws_profile" { 9 | type = string 10 | default = "development" 11 | description = "" 12 | } 13 | 14 | variable "aws_region" { 15 | default = "us-west-2" 16 | description = "AWS region to provision the Kubernetes Cluster" 17 | } 18 | 19 | 20 | variable "cluster_name" { 21 | default = "eks-cluster-tf" 22 | type = string 23 | } 24 | 25 | variable "cluster_version" { 26 | type = string 27 | default = "1.32" 28 | description = "Version of EKS to install on the control plane (Major and Minor version only, do not include the patch)" 29 | } 30 | /************************ 31 | GPU Operator Variables 32 | *************************/ 33 | variable "install_gpu_operator" { 34 | default = "true" 35 | description = "Whether to Install GPU Operator. Defaults to false available." 36 | } 37 | 38 | variable "gpu_operator_version" { 39 | default = "v25.3.0" 40 | description = "Version of the GPU Operator to deploy. Defaults to latest available. " 41 | } 42 | 43 | variable "gpu_operator_driver_version" { 44 | type = string 45 | default = "570.124.06" 46 | description = "The NVIDIA Driver version deployed with GPU Operator. Defaults to latest available." 47 | } 48 | 49 | variable "gpu_operator_namespace" { 50 | type = string 51 | default = "gpu-operator" 52 | description = "The namespace for the GPU operator deployment" 53 | } 54 | 55 | /************************ 56 | NIM Operator Variables 57 | *************************/ 58 | variable "install_nim_operator" { 59 | default = "false" 60 | description = "Whether to Install NIM Operator. Defaults to false available." 61 | } 62 | 63 | variable "nim_operator_version" { 64 | default = "v1.0.1" 65 | description = "Version of the GPU Operator to deploy. Defaults to latest available." 66 | } 67 | 68 | variable "nim_operator_namespace" { 69 | type = string 70 | default = "nim-operator" 71 | description = "The namespace for the GPU operator deployment" 72 | } 73 | 74 | /***************************** 75 | Managed Node Pool Variables 76 | ******************************/ 77 | 78 | /****************************** 79 | GPU-only Node Pool Variables 80 | *******************************/ 81 | variable "gpu_ami_id" { 82 | type = string 83 | description = "AMI ID of the EKS Ubuntu Image cooresponding to the region and version of the cluser. Not required as we do a lookup for this image" 84 | default = "" 85 | } 86 | 87 | variable "gpu_instance_type" { 88 | type = string 89 | default = "g4dn.2xlarge" 90 | description = "GPU EC2 worker node instance type" 91 | } 92 | 93 | variable "max_gpu_nodes" { 94 | type = string 95 | default = "5" 96 | description = "Maximum number of GPU nodes in the Autoscaling Group" 97 | } 98 | 99 | variable "min_gpu_nodes" { 100 | type = string 101 | default = "2" 102 | description = "Minimum number of GPU nodes in the Autoscaling Group" 103 | } 104 | 105 | variable "desired_count_gpu_nodes" { 106 | type = string 107 | default = "2" 108 | description = "Minimum number of GPU nodes in the Autoscaling Group" 109 | } 110 | 111 | variable "gpu_node_pool_root_disk_size_gb" { 112 | type = number 113 | default = 512 114 | description = "The size of the root disk on all GPU nodes in the EKS-managed GPU-only Node Pool. This is primarily for container image storage on the node" 115 | } 116 | 117 | variable "gpu_node_pool_root_volume_type" { 118 | type = string 119 | default = "gp2" 120 | description = "The type of disk to use for the GPU node pool root disk (eg. gp2, gp3). Note, this is different from the type of disk used by applications via EKS Storage classes/PVs & PVCs" 121 | } 122 | 123 | variable "gpu_node_pool_delete_on_termination" { 124 | type = bool 125 | default = true 126 | description = "Delete the VM nodes root filesystem on each node of the instance type. This is set to true by default, but can be changed when desired when using the 'local-storage provisioner' and are keeping important application data on the nodes" 127 | } 128 | 129 | variable "gpu_node_pool_additional_user_data" { 130 | type = string 131 | default = "" 132 | description = "User data that is appended to the user data script after of the EKS bootstrap script on EKS-managed GPU node pool." 133 | } 134 | 135 | /************************ 136 | CPU-only Node Pool Variables 137 | *************************/ 138 | 139 | variable "cpu_instance_type" { 140 | type = string 141 | default = "t2.xlarge" 142 | description = "CPU EC2 worker node instance type" 143 | } 144 | 145 | variable "cpu_node_pool_root_disk_size_gb" { 146 | type = number 147 | default = 512 148 | description = "The size of the root disk on all GPU nodes in the EKS-managed GPU-only Node Pool. This is primarily for container image storage on the node" 149 | } 150 | 151 | variable "cpu_node_pool_root_volume_type" { 152 | type = string 153 | default = "gp2" 154 | description = "The type of disk to use for the GPU node pool root disk (eg. gp2, gp3). Note, this is different from the type of disk used by applications via EKS Storage classes/PVs & PVCs" 155 | } 156 | 157 | variable "cpu_node_pool_delete_on_termination" { 158 | type = bool 159 | default = true 160 | description = "Delete the VM nodes root filesystem on each node of the instance type. This is set to true by default, but can be changed when desired when using the 'local-storage provisioner' and are keeping important application data on the nodes" 161 | } 162 | 163 | variable "cpu_node_pool_additional_user_data" { 164 | type = string 165 | default = "" 166 | description = "User data that is appended to the user data script after of the EKS bootstrap script on EKS-managed GPU node pool." 167 | } 168 | 169 | variable "max_cpu_nodes" { 170 | type = string 171 | default = "2" 172 | description = "Maximum number of CPU nodes in the Autoscaling Group" 173 | } 174 | 175 | variable "min_cpu_nodes" { 176 | type = string 177 | default = "0" 178 | description = "Minimum number of CPU nodes in the Autoscaling Group" 179 | } 180 | 181 | variable "desired_count_cpu_nodes" { 182 | type = string 183 | default = "1" 184 | description = "Minimum number of CPU nodes in the Autoscaling Group" 185 | } 186 | 187 | 188 | variable "existing_vpc_details" { 189 | type = object({ 190 | vpc_id = string 191 | subnet_ids = list(string) 192 | }) 193 | default = null 194 | description = "Variables used for re-using existing VPC for vpc_id & subnet_id" 195 | } 196 | 197 | variable "cidr_block" { 198 | type = string 199 | default = "10.0.0.0/16" 200 | description = "CIDR for VPC" 201 | } 202 | 203 | variable "additional_user_data" { 204 | type = string 205 | default = "" 206 | description = "User data that is appended to the user data script after of the EKS bootstrap script." 207 | } 208 | 209 | variable "private_subnets" { 210 | type = list(any) 211 | description = "List of subnet ranges for the Private VPC" 212 | default = ["10.0.0.0/19", "10.0.32.0/19", "10.0.64.0/19"] 213 | } 214 | 215 | variable "public_subnets" { 216 | type = list(any) 217 | description = "List of subnet ranges for the Private VPC" 218 | default = ["10.0.96.0/19", "10.0.128.0/19", "10.0.160.0/19"] 219 | } 220 | 221 | variable "ssh_key" { 222 | type = string 223 | default = "" 224 | } 225 | 226 | variable "enable_nat_gateway" { 227 | description = "Should be true if you want to provision NAT Gateways for each of your private networks" 228 | default = true 229 | type = bool 230 | } 231 | 232 | variable "single_nat_gateway" { 233 | type = bool 234 | description = "Should be true if you want to provision a single shared NAT Gateway across all of your private networks" 235 | default = false 236 | } 237 | 238 | variable "enable_dns_support" { 239 | type = bool 240 | default = true 241 | description = "Whether or not the Default VPC has DNS support" 242 | } 243 | 244 | variable "enable_dns_hostnames" { 245 | description = "Whether or not the Default VPC has DNS hostname support" 246 | default = true 247 | type = bool 248 | } 249 | 250 | variable "additional_security_group_ids" { 251 | type = list(any) 252 | default = [] 253 | description = "list of additional security groups to add to nodes" 254 | } 255 | 256 | variable "additional_node_security_groups_rules" { 257 | description = "List of additional security group rules to add to the node security group created" 258 | type = any 259 | default = {} 260 | } 261 | -------------------------------------------------------------------------------- /gke/main.tf: -------------------------------------------------------------------------------- 1 | # SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved. 2 | # SPDX-License-Identifier: Apache-2.0 3 | 4 | /*************************** 5 | VPC Network Configuration 6 | ***************************/ 7 | resource "google_compute_network" "gke-vpc" { 8 | count = var.vpc_enabled ? 1 : 0 9 | name = "${var.cluster_name}-vpc" 10 | auto_create_subnetworks = "false" 11 | project = var.project_id 12 | } 13 | 14 | /*************************** 15 | Subnet Configuration 16 | ***************************/ 17 | resource "google_compute_subnetwork" "gke-subnet" { 18 | name = "${var.cluster_name}-subnet" 19 | count = var.vpc_enabled ? 1 : 0 20 | region = var.region 21 | network = google_compute_network.gke-vpc[0].name 22 | ip_cidr_range = "10.150.0.0/24" 23 | project = var.project_id 24 | } 25 | 26 | /*************************** 27 | GKE Configuration 28 | ***************************/ 29 | 30 | # Add data block to provide latest k8s version as an output 31 | data "google_container_engine_versions" "latest" { 32 | provider = google-beta 33 | location = var.region 34 | project = var.project_id 35 | } 36 | 37 | resource "google_container_cluster" "gke" { 38 | name = var.cluster_name 39 | project = var.project_id 40 | location = length(var.node_zones) == 1 ? one(var.node_zones) : var.region 41 | release_channel { 42 | channel = var.release_channel 43 | } 44 | min_master_version = var.min_master_version 45 | # Default Node Pool is required, to create a cluster, but we need a custom one instead 46 | # So we delete the default 47 | remove_default_node_pool = true 48 | initial_node_count = 1 49 | 50 | network = var.vpc_enabled ? google_compute_network.gke-vpc[0].name : var.network 51 | subnetwork = var.vpc_enabled ? google_compute_subnetwork.gke-subnet[0].name : var.subnetwork 52 | 53 | deletion_protection = false 54 | 55 | // Workload Identity Configuration 56 | workload_identity_config { 57 | workload_pool = "${var.project_id}.svc.id.goog" 58 | } 59 | } 60 | /*************************** 61 | GKE CPU Node Pool Config 62 | ***************************/ 63 | resource "google_container_node_pool" "cpu_nodes" { 64 | name = "tf-${var.cluster_name}-cpu-pool" 65 | project = var.project_id 66 | location = length(var.node_zones) == 1 ? one(var.node_zones) : var.region 67 | node_locations = length(var.node_zones) > 1 ? var.node_zones : null 68 | cluster = google_container_cluster.gke.name 69 | node_count = var.num_cpu_nodes 70 | autoscaling { 71 | min_node_count = var.cpu_min_node_count 72 | max_node_count = var.cpu_max_node_count 73 | } 74 | node_config { 75 | image_type = "UBUNTU_CONTAINERD" 76 | oauth_scopes = [ 77 | "https://www.googleapis.com/auth/logging.write", 78 | "https://www.googleapis.com/auth/monitoring", 79 | "https://www.googleapis.com/auth/devstorage.read_only", 80 | "https://www.googleapis.com/auth/compute" 81 | ] 82 | 83 | preemptible = var.use_cpu_spot_instances 84 | machine_type = var.cpu_instance_type 85 | disk_size_gb = var.disk_size_gb 86 | tags = concat(["tf-managed", "${var.cluster_name}"], var.gpu_instance_tags) 87 | metadata = { 88 | disable-legacy-endpoints = "true" 89 | } 90 | 91 | labels = { 92 | part_of = var.cluster_name 93 | env = var.project_id 94 | managed_by = "terraform" 95 | } 96 | workload_metadata_config { 97 | mode = "GKE_METADATA" 98 | } 99 | } 100 | timeouts { 101 | create = "30m" 102 | update = "20m" 103 | } 104 | } 105 | 106 | /*************************** 107 | GKE GPU Node Pool Config 108 | ***************************/ 109 | resource "google_container_node_pool" "gpu_nodes" { 110 | name = "tf-${var.cluster_name}-gpu-pool" 111 | project = var.project_id 112 | location = length(var.node_zones) == 1 ? one(var.node_zones) : var.region 113 | node_locations = length(var.node_zones) > 1 ? var.node_zones : null 114 | cluster = google_container_cluster.gke.name 115 | node_count = var.num_gpu_nodes 116 | autoscaling { 117 | min_node_count = var.gpu_min_node_count 118 | max_node_count = var.gpu_max_node_count 119 | } 120 | node_config { 121 | image_type = "UBUNTU_CONTAINERD" 122 | oauth_scopes = [ 123 | "https://www.googleapis.com/auth/logging.write", 124 | "https://www.googleapis.com/auth/monitoring", 125 | "https://www.googleapis.com/auth/devstorage.read_only", 126 | "https://www.googleapis.com/auth/compute" 127 | ] 128 | guest_accelerator { 129 | type = var.gpu_type 130 | count = var.gpu_count 131 | gpu_driver_installation_config { 132 | gpu_driver_version = "INSTALLATION_DISABLED" 133 | } 134 | } 135 | 136 | preemptible = var.use_gpu_spot_instances 137 | machine_type = var.gpu_instance_type 138 | disk_size_gb = var.disk_size_gb 139 | tags = concat(["tf-managed", "${var.cluster_name}"], var.gpu_instance_tags) 140 | metadata = { 141 | disable-legacy-endpoints = "true" 142 | } 143 | 144 | labels = { 145 | part_of = var.cluster_name 146 | env = var.project_id 147 | managed_by = "terraform" 148 | "gke-no-default-nvidia-gpu-device-plugin" = "true" 149 | } 150 | workload_metadata_config { 151 | mode = "GKE_METADATA" 152 | } 153 | } 154 | timeouts { 155 | create = "30m" 156 | update = "20m" 157 | } 158 | } 159 | 160 | /*************************** 161 | Create GPU Operator Namespace 162 | ***************************/ 163 | resource "kubernetes_namespace_v1" "gpu-operator" { 164 | metadata { 165 | annotations = { 166 | name = "gpu-operator" 167 | } 168 | 169 | labels = { 170 | cluster = var.cluster_name 171 | managed_by = "Terraform" 172 | } 173 | 174 | name = var.gpu_operator_namespace 175 | } 176 | } 177 | /*************************** 178 | K8s Resource Quota Config 179 | ***************************/ 180 | resource "kubernetes_resource_quota_v1" "gpu-operator-quota" { 181 | depends_on = [google_container_node_pool.gpu_nodes, kubernetes_namespace_v1.gpu-operator] 182 | metadata { 183 | name = "gpu-operator-quota" 184 | namespace = var.gpu_operator_namespace 185 | } 186 | spec { 187 | hard = { 188 | pods = 100 189 | } 190 | scope_selector { 191 | match_expression { 192 | operator = "In" 193 | scope_name = "PriorityClass" 194 | values = ["system-node-critical", "system-cluster-critical"] 195 | } 196 | } 197 | } 198 | } 199 | /*************************** 200 | GPU Operator Configuration 201 | ***************************/ 202 | resource "helm_release" "gpu-operator" { 203 | depends_on = [google_container_node_pool.gpu_nodes, kubernetes_resource_quota_v1.gpu-operator-quota, kubernetes_namespace_v1.gpu-operator] 204 | count = var.install_gpu_operator ? 1 : 0 205 | name = "gpu-operator" 206 | repository = "https://helm.ngc.nvidia.com/nvidia" 207 | chart = "gpu-operator" 208 | version = var.gpu_operator_version 209 | namespace = var.gpu_operator_namespace 210 | create_namespace = false 211 | atomic = true 212 | cleanup_on_fail = true 213 | reset_values = true 214 | replace = true 215 | 216 | set { 217 | name = "driver.version" 218 | value = var.gpu_operator_driver_version 219 | } 220 | 221 | } 222 | 223 | /*************************** 224 | Create NIM Operator Namespace 225 | ***************************/ 226 | resource "kubernetes_namespace_v1" "nim-operator" { 227 | metadata { 228 | annotations = { 229 | name = "nim-operator" 230 | } 231 | 232 | labels = { 233 | cluster = var.cluster_name 234 | managed_by = "Terraform" 235 | } 236 | 237 | name = var.nim_operator_namespace 238 | } 239 | } 240 | /*************************** 241 | K8s Resource Quota Config 242 | ***************************/ 243 | resource "kubernetes_resource_quota_v1" "nim-operator-quota" { 244 | depends_on = [google_container_node_pool.gpu_nodes, kubernetes_namespace_v1.nim-operator] 245 | metadata { 246 | name = "gpu-operator-quota" 247 | namespace = var.nim_operator_namespace 248 | } 249 | spec { 250 | hard = { 251 | pods = 100 252 | } 253 | scope_selector { 254 | match_expression { 255 | operator = "In" 256 | scope_name = "PriorityClass" 257 | values = ["system-node-critical", "system-cluster-critical"] 258 | } 259 | } 260 | } 261 | } 262 | 263 | /******************************************** 264 | NIM Operator Configuration 265 | ********************************************/ 266 | resource "helm_release" "nim_operator" { 267 | depends_on = [google_container_node_pool.gpu_nodes, kubernetes_resource_quota_v1.nim-operator-quota, kubernetes_namespace_v1.nim-operator] 268 | count = var.install_nim_operator ? 1 : 0 269 | name = "nim-operator" 270 | repository = "https://helm.ngc.nvidia.com/nvidia" 271 | chart = "k8s-nim-operator" 272 | version = var.nim_operator_version 273 | namespace = var.nim_operator_namespace 274 | create_namespace = true 275 | atomic = true 276 | cleanup_on_fail = true 277 | reset_values = true 278 | replace = true 279 | } 280 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "[]" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright 2022-2023 NVIDIA CORPORATION & AFFILIATES. 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. -------------------------------------------------------------------------------- /aks/README.md: -------------------------------------------------------------------------------- 1 | # NVIDIA AKS cluster 2 | 3 | ## Tested on 4 | This module was created and tested on Linux and MacOS. 5 | 6 | ## Resources Created 7 | - Azure Resource Group 8 | - AKS Cluster 9 | - 1x CPU nodepool (defaults to 1x CPU node -- Standard_D16_v5) 10 | - 1x GPU nodepool (defaults to 2x GPU node -- Standard_NC4as_T4_v3 with T4) 11 | - Installs Latest version of GPU Operator 12 | 13 | ## Prerequisites 14 | 1. Kubectl 15 | 2. [Azure CLI](https://learn.microsoft.com/en-us/cli/azure/install-azure-cli) 16 | 3. Azure Account & Subscription where you are permitted to create cloud resources 17 | 4. Terraform (CLI) 18 | 5. [Azure Kubelogin](https://github.com/Azure/kubelogin#setup) 19 | 20 | ## Usage 21 | 22 | This module assumes that you have a working `terraform` binary and active Azure credentials. 23 | 24 | No Terraform Provider is setup for remote state management but can be added. 25 | We strongly encourage you [configure remote state](https://developer.hashicorp.com/terraform/language/state/remote) before running in production. 26 | 27 | 1. Clone the repo 28 | 29 | ``` 30 | git clone https://github.com/NVIDIA/nvidia-terraform-modules.git 31 | 32 | cd nvidia-terraform-modules/aks 33 | ``` 34 | 35 | 2. Logging in to Azure via the CLI 36 | - Run the below command , this will authenticate you to your Azure account 37 | 38 | ``` 39 | az login 40 | ``` 41 | 42 | 3. Update `terraform.tfvars` file to customize a parameter from its default value, please uncomment the line and change the content 43 | 44 | Mandatory: provide your admin_group_object_ids 45 | 46 | 47 | Add the IDs of the members or groups who should have cluster access to the variable `admin_group_object_ids`. 48 | The GUID input can be retrieved in the Azure portal by searching for the desired user or group, for more info please refer [Find Object Id](https://learn.microsoft.com/en-us/partner-center/marketplace/find-tenant-object-id). 49 | 50 | 51 | 52 | ``` 53 | admin_group_object_ids = ["xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"] 54 | ``` 55 | 56 | 57 | Mandatory: provide your subscription_id 58 | 59 | 60 | ``` 61 | subscription_id = "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx" 62 | ``` 63 | 64 | 65 | 4. Initialize the module with the command: 66 | 67 | ``` 68 | terraform init 69 | ``` 70 | 71 | 5. Run this command to view the proposed changes: 72 | 73 | ``` 74 | terraform plan -out tfplan 75 | ``` 76 | 77 | 6. Run the below command to apply the configuration: 78 | 79 | ``` 80 | terraform apply tfplan 81 | ``` 82 | 83 | 7. Once cluster is created run the below command with aks cluster name and resource group name to get kubeconfig so you are able to run `kubectl` commands: 84 | 85 | ``` 86 | az aks get-credentials --resource-group aks-cluster-tf-rg --name aks-cluster-tf 87 | ``` 88 | 89 | #### Cleaning up / Deleting resources 90 | 91 | 1. Run the beloe commands to delete all remaining Azure resources created by this module. You should see `Destroy complete!` message after a few minutes. 92 | 93 | ``` 94 | terraform state rm kubernetes_namespace_v1.gpu-operator 95 | 96 | terraform state rm kubernetes_namespace_v1.nim-operator 97 | ``` 98 | 99 | ``` 100 | terraform destroy --auto-approve 101 | ``` 102 | 103 | ## Running as a module 104 | 105 | Call the AKS module by adding this to an existing Terraform file: 106 | 107 | ```hcl 108 | module "nvidia-aks" { 109 | source = "git::github.com/NVIDIA/nvidia-terraform-modules/aks" 110 | cluster_name = "nvidia-aks" 111 | admin_group_object_ids = [] # See below for the value of this variable 112 | } 113 | ``` 114 | All configurable options for this module are listed below. 115 | If you need additional values added, please open a pull request. ``` 116 | 117 | ## Issues 118 | - None. If you do encounter an issue, please file a GitHub issue. 119 | 120 | ## Troubleshooting 121 | - ### Quota Errors 122 | New Azure accounts which have not turned on VMs or GPU VMs in any region will need to request quota in that region. 123 | During installation, if you see a quota-related error, click the link in the error message to be redirected to the Azure console with a prepopulated quota request. Re-run `terraform apply` once the quota request is complete. This will take ~5m per quota request 124 | 125 | - ### Azure Cloudshell Errors 126 | When using Azure Cloudshell during installation, if you see a `MSI` or `Bad Request(400)` error, it is [known issue](https://github.com/Azure/azure-cli/issues/11749) with Azure Cloushell. 127 | There are 2 workarounds: 128 | - Use Azure CLI on a local machine 129 | - In Cloud Shell, run `az login` and re-run `terraform apply` 130 | 131 | 132 | ## Requirements 133 | 134 | | Name | Version | 135 | |------|---------| 136 | | [terraform](#requirement\_terraform) | >= 1.3.4 | 137 | | [azurerm](#requirement\_azurerm) | ~>4.24.0 | 138 | | [kubernetes](#requirement\_kubernetes) | ~>2.19.0 | 139 | 140 | ## Providers 141 | 142 | | Name | Version | 143 | |------|---------| 144 | | [azurerm](#provider\_azurerm) | ~>4.24.0 | 145 | | [helm](#provider\_helm) | n/a | 146 | | [kubernetes](#provider\_kubernetes) | ~>2.19.0 | 147 | 148 | ## Modules 149 | 150 | No modules. 151 | 152 | ## Resources 153 | 154 | | Name | Type | 155 | |------|------| 156 | | [azurerm_kubernetes_cluster.aks](https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/resources/kubernetes_cluster) | resource | 157 | | [azurerm_kubernetes_cluster_node_pool.aks](https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/resources/kubernetes_cluster_node_pool) | resource | 158 | | [azurerm_resource_group.aks](https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/resources/resource_group) | resource | 159 | | [helm_release.gpu-operator](https://registry.terraform.io/providers/hashicorp/helm/latest/docs/resources/release) | resource | 160 | | [helm_release.nim_operator](https://registry.terraform.io/providers/hashicorp/helm/latest/docs/resources/release) | resource | 161 | | [kubernetes_namespace_v1.gpu-operator](https://registry.terraform.io/providers/hashicorp/kubernetes/latest/docs/resources/namespace_v1) | resource | 162 | | [kubernetes_namespace_v1.nim-operator](https://registry.terraform.io/providers/hashicorp/kubernetes/latest/docs/resources/namespace_v1) | resource | 163 | | [azurerm_kubernetes_cluster.akscluster](https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/data-sources/kubernetes_cluster) | data source | 164 | | [azurerm_resource_group.existing](https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/data-sources/resource_group) | data source | 165 | 166 | ## Inputs 167 | 168 | | Name | Description | Type | Default | Required | 169 | |------|-------------|------|---------|:--------:| 170 | | [admin\_group\_object\_ids](#input\_admin\_group\_object\_ids) | (Required) A list of Object IDs (GUIDs) of Azure Active Directory Groups which should have Owner Role on the Cluster.
This is not the email address of the group, the GUID can be found in the Azure panel by searching for the AD Group
NOTE: You will need Azure "Owner" role (not "Contributor") to attach an AD role to the Kubernetes cluster. | `list(any)` | n/a | yes | 171 | | [cluster\_name](#input\_cluster\_name) | The name of the AKS Cluster to be created | `string` | `"aks-cluster"` | no | 172 | | [cpu\_machine\_type](#input\_cpu\_machine\_type) | Machine instance type of the AKS CPU node pool | `string` | `"Standard_D16_v5"` | no | 173 | | [cpu\_node\_pool\_count](#input\_cpu\_node\_pool\_count) | Count of nodes in Default GPU pool | `number` | `1` | no | 174 | | [cpu\_node\_pool\_disk\_size](#input\_cpu\_node\_pool\_disk\_size) | Disk size in GB of nodes in the Default GPU pool | `number` | `100` | no | 175 | | [cpu\_node\_pool\_max\_count](#input\_cpu\_node\_pool\_max\_count) | Max count of nodes in Default CPU pool | `number` | `5` | no | 176 | | [cpu\_node\_pool\_min\_count](#input\_cpu\_node\_pool\_min\_count) | Min ount of number of nodes in Default CPU pool | `number` | `1` | no | 177 | | [cpu\_os\_sku](#input\_cpu\_os\_sku) | Specifies the OS SKU used by the agent pool. Possible values include: Ubuntu, CBLMariner, Mariner, Windows2019, Windows2022 | `string` | `"Ubuntu"` | no | 178 | | [existing\_resource\_group\_name](#input\_existing\_resource\_group\_name) | The name of an existing resource group the Kubernetes cluster should be deployed into. Defaults to the name of the cluster + `-rg` if none is specified | `string` | `null` | no | 179 | | [gpu\_machine\_type](#input\_gpu\_machine\_type) | Machine instance type of the AKS GPU node pool | `string` | `"Standard_NC6s_v3"` | no | 180 | | [gpu\_node\_pool\_count](#input\_gpu\_node\_pool\_count) | Count of nodes in Default GPU pool | `number` | `2` | no | 181 | | [gpu\_node\_pool\_disk\_size](#input\_gpu\_node\_pool\_disk\_size) | Disk size in GB of nodes in the Default GPU pool | `number` | `100` | no | 182 | | [gpu\_node\_pool\_max\_count](#input\_gpu\_node\_pool\_max\_count) | Max count of nodes in Default GPU pool | `number` | `5` | no | 183 | | [gpu\_node\_pool\_min\_count](#input\_gpu\_node\_pool\_min\_count) | Min count of number of nodes in Default GPU pool | `number` | `2` | no | 184 | | [gpu\_operator\_driver\_version](#input\_gpu\_operator\_driver\_version) | The NVIDIA Driver version deployed with GPU Operator. Defaults to latest available. | `string` | `"570.124.06"` | no | 185 | | [gpu\_operator\_namespace](#input\_gpu\_operator\_namespace) | The namespace to deploy the NVIDIA GPU operator into | `string` | `"gpu-operator"` | no | 186 | | [gpu\_operator\_version](#input\_gpu\_operator\_version) | Version of the GPU operator to be installed | `string` | `"v25.3.0"` | no | 187 | | [gpu\_os\_sku](#input\_gpu\_os\_sku) | Specifies the OS SKU used by the agent pool. Possible values include: Ubuntu, CBLMariner, Mariner, Windows2019, Windows2022 | `string` | `"Ubuntu"` | no | 188 | | [install\_gpu\_operator](#input\_install\_gpu\_operator) | Whether to Install GPU Operator. Defaults to false available. | `string` | `"true"` | no | 189 | | [install\_nim\_operator](#input\_install\_nim\_operator) | Whether to Install NIM Operator. Defaults to false available. | `string` | `"false"` | no | 190 | | [kubernetes\_version](#input\_kubernetes\_version) | Version of Kubernetes to turn on. Run 'az aks get-versions --location --output table' to view all available versions | `string` | `"1.30"` | no | 191 | | [location](#input\_location) | The region to create resources in | `any` | n/a | yes | 192 | | [nim\_operator\_namespace](#input\_nim\_operator\_namespace) | The namespace for the GPU operator deployment | `string` | `"nim-operator"` | no | 193 | | [nim\_operator\_version](#input\_nim\_operator\_version) | Version of the GPU Operator to deploy. Defaults to latest available. | `string` | `"v1.0.0"` | no | 194 | 195 | ## Outputs 196 | 197 | | Name | Description | 198 | |------|-------------| 199 | | [client\_certificate](#output\_client\_certificate) | n/a | 200 | | [kube\_config](#output\_kube\_config) | n/a | 201 | | [kubernetes\_cluster\_name](#output\_kubernetes\_cluster\_name) | n/a | 202 | | [location](#output\_location) | n/a | 203 | | [resource\_group\_name](#output\_resource\_group\_name) | n/a | 204 | -------------------------------------------------------------------------------- /gke/README.md: -------------------------------------------------------------------------------- 1 | # NVIDIA GKE Cluster 2 | 3 | This repo provides Terraform configuration to bring up a GKE Kubernetes Cluster with the GPU operator and GPU nodes from scratch. 4 | 5 | 6 | ## Tested on 7 | This module was created with and tested on Linux using Bash, it may or may not work on Windows or when using Powershell. 8 | 9 | ## Resources Created 10 | - VPC Network for GKE Cluster 11 | - Subnet in VPC 12 | - GKE Cluster 13 | - 1 CPU nodepool (defaults to 1x CPU node -- n1-standard-4) 14 | - 1 GPU nodepool (defaults to 2x GPU nodes -- n1-standard-4 with 1x Tesla T4) 15 | - Installs latest version of GPU Operator via Helm 16 | 17 | ## Prerequisites 18 | 1. Kubectl 19 | 2. Google Cloud ((glcoud)[https://cloud.google.com/sdk/docs/install]) CLI 20 | 3. GCP Account & Project where you are permitted to create cloud resources 21 | 4. Terraform (CLI) 22 | 23 | 24 | #### Setup 25 | 26 | 1. Requires the `gcloud` SDK binary -- [Download here](https://cloud.google.com/sdk/docs/install) 27 | 28 | 2. Requires the Terraform cli @ Version 1.3.4 or higher -- [Download here](https://developer.hashicorp.com/terraform/downloads) 29 | 30 | 3. To run this module assumes elevated permissions (Kubernetes Engine Admin) in your GCP account, specifically permissions to create VPC networks, GKE clusters, and Compute nodes. This will not work on accounts using the "free plan" as you cannot use GPU nodes until a billing account is attached and activated. 31 | 32 | 4. You will need to enable both the Kubernetes API and the Compute Engine APIs enabled. Click [the GKE tab in the GCP panel](https://console.cloud.google.com/kubernetes) for your project and enable the GKE API, which will also enable the Compute engine API at the same time 33 | 34 | 5. Ensure you have [GPU Quota](https://cloud.google.com/compute/quotas#gpu_quota) in your desired region/zone. You can [request](GPU Quota) if it is not enabled in a new account. You will need quota for both `GPUS_ALL_REGIONS` and for the specific SKU in the desired region. 35 | 36 | ## Usage 37 | 38 | 0. Authenticate to GCP: 39 | 40 | ``` 41 | gcloud auth login 42 | ``` 43 | 44 | 45 | 1. Run this command to clone the repo: 46 | 47 | ``` 48 | git clone https://github.com/NVIDIA/nvidia-terraform-modules.git 49 | 50 | cd nvidia-terraform-modules/gke 51 | ``` 52 | 53 | 2. Update `terraform.tfvars` to customize a parameter from its default value, please uncomment the line and change the content 54 | 55 | Mandatory: provide your project ID by updating the parameter `project_id`. 56 | 57 | You can get the `project_id` from your [GCP console](#https://cloud.google.com/resource-manager/docs/creating-managing-projects#identifying_projects). 58 | 59 | ``` 60 | project_id = "xx-xxxx-xxxx" 61 | 62 | ``` 63 | 64 | 3. Run this command to make your Google Credentials available to the `terraform` executable: 65 | 66 | ``` 67 | gcloud auth application-default login 68 | ``` 69 | 70 | 4. Run this command to fetch the required Terraform provider plugins: 71 | 72 | ``` 73 | terraform init 74 | ``` 75 | 76 | 5. If your credentials are setup correctly, you should see the proposed changes in GCP by running `terraform plan -out tfplan`: 77 | 78 | ``` 79 | terraform plan -out tfplan 80 | ``` 81 | 82 | ** Note on IAM Permissions:** you need either `Admin` permissions or `Compute Instance Admin (v1)`, `Kubernetes Engine Admin` and `Compute Network Admin (v1)` to run this module. 83 | 84 | 6. If this configuration looks approproate run this command: 85 | 86 | ``` 87 | terraform apply tfplan 88 | ``` 89 | 90 | 7. It will take ~5 minutes after the `terraform apply` successful completion message for the GPU operator to get to a running state 91 | 92 | 8. Connect to the cluster with `kubectl` by running the following two commands after the cluster is created: 93 | 94 | ``` 95 | gcloud components install gke-gcloud-auth-plugin 96 | 97 | gcloud container clusters get-credentials --region= 98 | ``` 99 | 100 | #### Cleaning up / Deleting resources 101 | 102 | 1. Run these commands to delete all remaining GCP resources created by this module. You should see `Destroy complete!` message after a few minutes. 103 | 104 | ``` 105 | terraform state rm kubernetes_namespace_v1.gpu-operator 106 | 107 | terraform state rm kubernetes_namespace_v1.nim-operator 108 | ``` 109 | 110 | ``` 111 | sed -i '' 's/\"deletion_protection\": true\,/\"deletion_protection\": false\,/g' terraform.tfstate 112 | ``` 113 | 114 | ``` 115 | terraform destroy --auto-approve 116 | ``` 117 | 118 | # Terraform Module Information 119 | ## Running as a module 120 | 121 | Call the GKE module by adding this to an existing Terraform file: 122 | 123 | ```hcl 124 | module "nvidia-gke" { 125 | source = "git::github.com/NVIDIA/nvidia-terraform-modules/gke" 126 | project_id = "" 127 | region = "us-west1" # Can be any region 128 | node_zones = ["us-west1-b"] # Can be any region but ensure your desired machine types/gpus exist 129 | } 130 | ``` 131 | 132 | All configurable options for this module are listed below. 133 | If you need additional values added, please open a merge request. 134 | 135 | ## Requirements 136 | 137 | | Name | Version | 138 | |------|---------| 139 | | [terraform](#requirement\_terraform) | >= 1.3.4 | 140 | | [google](#requirement\_google) | 6.27.0 | 141 | | [google-beta](#requirement\_google-beta) | 6.27.0 | 142 | 143 | ## Providers 144 | 145 | | Name | Version | 146 | |------|---------| 147 | | [google](#provider\_google) | 6.27.0 | 148 | | [google-beta](#provider\_google-beta) | 6.27.0 | 149 | | [helm](#provider\_helm) | n/a | 150 | | [kubernetes](#provider\_kubernetes) | 2.19.0 | 151 | 152 | ## Modules 153 | 154 | No modules. 155 | 156 | ## Resources 157 | 158 | | Name | Type | 159 | |------|------| 160 | | [google_compute_network.gke-vpc](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/resources/compute_network) | resource | 161 | | [google_compute_subnetwork.gke-subnet](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/resources/compute_subnetwork) | resource | 162 | | [google_container_cluster.gke](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/resources/container_cluster) | resource | 163 | | [google_container_node_pool.cpu_nodes](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/resources/container_node_pool) | resource | 164 | | [google_container_node_pool.gpu_nodes](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/resources/container_node_pool) | resource | 165 | | [helm_release.gpu-operator](https://registry.terraform.io/providers/hashicorp/helm/latest/docs/resources/release) | resource | 166 | | [helm_release.nim_operator](https://registry.terraform.io/providers/hashicorp/helm/latest/docs/resources/release) | resource | 167 | | [kubernetes_namespace_v1.gpu-operator](https://registry.terraform.io/providers/hashicorp/kubernetes/latest/docs/resources/namespace_v1) | resource | 168 | | [kubernetes_namespace_v1.nim-operator](https://registry.terraform.io/providers/hashicorp/kubernetes/latest/docs/resources/namespace_v1) | resource | 169 | | [kubernetes_resource_quota_v1.gpu-operator-quota](https://registry.terraform.io/providers/hashicorp/kubernetes/latest/docs/resources/resource_quota_v1) | resource | 170 | | [kubernetes_resource_quota_v1.nim-operator-quota](https://registry.terraform.io/providers/hashicorp/kubernetes/latest/docs/resources/resource_quota_v1) | resource | 171 | | [google-beta_google_container_engine_versions.latest](https://registry.terraform.io/providers/hashicorp/google-beta/4.57.0/docs/data-sources/google_container_engine_versions) | data source | 172 | | [google_client_config.provider](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/data-sources/client_config) | data source | 173 | | [google_container_cluster.gke-cluster](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/data-sources/container_cluster) | data source | 174 | | [google_project.cluster](https://registry.terraform.io/providers/hashicorp/google/4.27.0/docs/data-sources/project) | data source | 175 | 176 | ## Inputs 177 | 178 | | Name | Description | Type | Default | Required | 179 | |------|-------------|------|---------|:--------:| 180 | | [cluster\_name](#input\_cluster\_name) | Name of the Kubernetes Cluster to provision | `string` | n/a | yes | 181 | | [cpu\_instance\_type](#input\_cpu\_instance\_type) | Machine Type for CPU node pool | `string` | `"n1-standard-4"` | no | 182 | | [cpu\_max\_node\_count](#input\_cpu\_max\_node\_count) | Max Number of CPU nodes in CPU nodepool | `string` | `"5"` | no | 183 | | [cpu\_min\_node\_count](#input\_cpu\_min\_node\_count) | Number of CPU nodes in CPU nodepool | `string` | `"1"` | no | 184 | | [disk\_size\_gb](#input\_disk\_size\_gb) | n/a | `string` | `"512"` | no | 185 | | [gpu\_count](#input\_gpu\_count) | Number of GPUs to attach to each node in GPU pool | `string` | `"1"` | no | 186 | | [gpu\_instance\_tags](#input\_gpu\_instance\_tags) | GPU instance nodes tags | `list(string)` | `[]` | no | 187 | | [gpu\_instance\_type](#input\_gpu\_instance\_type) | Machine Type for GPU node pool | `string` | `"n1-standard-4"` | no | 188 | | [gpu\_max\_node\_count](#input\_gpu\_max\_node\_count) | Max Number of GPU nodes in GPU nodepool | `string` | `"5"` | no | 189 | | [gpu\_min\_node\_count](#input\_gpu\_min\_node\_count) | Min number of GPU nodes in GPU nodepool | `string` | `"2"` | no | 190 | | [gpu\_operator\_driver\_version](#input\_gpu\_operator\_driver\_version) | The NVIDIA Driver version deployed with GPU Operator. Defaults to latest available | `string` | `"570.124.06"` | no | 191 | | [gpu\_operator\_namespace](#input\_gpu\_operator\_namespace) | The namespace to deploy the NVIDIA GPU operator into | `string` | `"gpu-operator"` | no | 192 | | [gpu\_operator\_version](#input\_gpu\_operator\_version) | Version of the GPU Operator to deploy. Defaults to latest available | `string` | `"v25.3.0"` | no | 193 | | [gpu\_type](#input\_gpu\_type) | GPU SKU To attach to NVIDIA GPU Node (eg. nvidia-tesla-k80) | `string` | `"nvidia-tesla-t4"` | no | 194 | | [install\_gpu\_operator](#input\_install\_gpu\_operator) | Whether to Install GPU Operator. Defaults to false available. | `string` | `"true"` | no | 195 | | [install\_nim\_operator](#input\_install\_nim\_operator) | Whether to Install NIM Operator. Defaults to false available. | `string` | `"false"` | no | 196 | | [min\_master\_version](#input\_min\_master\_version) | The minimum cluster version of the master. | `string` | `"1.32"` | no | 197 | | [network](#input\_network) | Network CIDR for VPC | `string` | `""` | no | 198 | | [nim\_operator\_namespace](#input\_nim\_operator\_namespace) | The namespace for the GPU operator deployment | `string` | `"nim-operator"` | no | 199 | | [nim\_operator\_version](#input\_nim\_operator\_version) | Version of the GPU Operator to deploy. Defaults to latest available | `string` | `"v1.0.1"` | no | 200 | | [node\_zones](#input\_node\_zones) | Specify zones to put nodes in (must be in same region defined above) | `list(any)` | n/a | yes | 201 | | [num\_cpu\_nodes](#input\_num\_cpu\_nodes) | Number of CPU nodes when pool is created | `number` | `1` | no | 202 | | [num\_gpu\_nodes](#input\_num\_gpu\_nodes) | Number of GPU nodes when pool is created | `number` | `2` | no | 203 | | [project\_id](#input\_project\_id) | GCP Project ID for the VPC and K8s Cluster. This module currently does not support projects with a SharedVPC | `any` | n/a | yes | 204 | | [region](#input\_region) | The Region resources (VPC, GKE, Compute Nodes) will be created in | `any` | n/a | yes | 205 | | [release\_channel](#input\_release\_channel) | Configuration options for the Release channel feature, which provide more control over automatic upgrades of your GKE clusters. When updating this field, GKE imposes specific version requirements | `string` | `"REGULAR"` | no | 206 | | [subnetwork](#input\_subnetwork) | Subnet name used for k8s cluster nodes | `string` | `""` | no | 207 | | [use\_cpu\_spot\_instances](#input\_use\_cpu\_spot\_instances) | Use Spot instance for CPU pool | `bool` | `false` | no | 208 | | [use\_gpu\_spot\_instances](#input\_use\_gpu\_spot\_instances) | Use Spot instance for GPU pool | `bool` | `false` | no | 209 | | [vpc\_enabled](#input\_vpc\_enabled) | Variable to control nvidia-kubernetes GKE module VPC creation | `bool` | `true` | no | 210 | 211 | ## Outputs 212 | 213 | | Name | Description | 214 | |------|-------------| 215 | | [kubernetes\_cluster\_endpoint\_ip](#output\_kubernetes\_cluster\_endpoint\_ip) | GKE Cluster IP Endpoint | 216 | | [kubernetes\_cluster\_name](#output\_kubernetes\_cluster\_name) | GKE Cluster Name | 217 | | [kubernetes\_config\_file](#output\_kubernetes\_config\_file) | GKE Cluster IP Endpoint | 218 | | [project\_id](#output\_project\_id) | GCloud Project ID | 219 | | [region](#output\_region) | Region for Kubernetes Resources to be created in when using this module | 220 | | [subnet\_cidr\_range](#output\_subnet\_cidr\_range) | The IPs and CIDRs of the subnets | 221 | | [subnet\_region](#output\_subnet\_region) | The region of the VPC subnet used in this module | 222 | | [vpc\_project](#output\_vpc\_project) | Project of the VPC network (can be different from the project launching Kubernetes resources) | 223 | -------------------------------------------------------------------------------- /eks/README.md: -------------------------------------------------------------------------------- 1 | # EKS Ready Kubernetes (EKS) 2 | This repo provides Terraform configuration to bring up a EKS Kubernetes Cluster with the GPU operator and GPU nodes from scratch. 3 | 4 | ## Tested on 5 | This module was created and tested on Linux and MacOS. 6 | 7 | ## Resources Created 8 | - VPC Network for EKS Cluster 9 | - Subnets in VPC for EKS CLuster 10 | - EKS Cluster 11 | - 1x CPU nodepool 12 | - 1x GPU nodepool 13 | - Installs latest version of GPU Operator via Helm 14 | - 1x KMS Key to encrypt cluster secrets 15 | 16 | For more details on resources created and their default values, please see the [Terraform Module Inputs](#inputs) section. 17 | 18 | ## Prerequisites 19 | 1. Kubectl 20 | 2. [AWS CLI](https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html) 21 | - You must run `aws-configure` once on your machine to populate the default region present in `~/.aws/config` 22 | 3. AWS Account where you have permissions to create a cluster, IAM roles and networking 23 | 4. Terraform (CLI)[https://developer.hashicorp.com/terraform/downloads] 24 | 5. [JQ](https://stedolan.github.io/jq/download/) 25 | - The provisioning will fail without this step, as it is used to set up your Kubernetes configuration file file after the cluster is provisioned 26 | 27 | #### Issues 28 | - None. If you encounter any, please file a GitHub issue 29 | 30 | ## Usage 31 | 32 | This module assumes that you have a working `terraform` binary and active AWS credentials (admin access or finely scoped permissions with basic EC2, EKS, VPC and IAM creation permissions). 33 | 34 | No Terraform Provider is setup for remote state management but can be added. 35 | We strongly encourage you [configure remote state](https://developer.hashicorp.com/terraform/language/state/remote) before running in production. 36 | 37 | 1. Clone the repo 38 | 39 | 40 | ``` 41 | git clone https://github.com/NVIDIA/nvidia-terraform-modules.git 42 | 43 | cd nvidia-terraform-modules/eks 44 | ``` 45 | 46 | 2. Ensure you have active credentials set with the AWS CLI. 47 | 48 | ``` 49 | aws configure 50 | ``` 51 | 52 | 3. Update `terraform.tfvars` to customize a parameter from its default value, please uncomment the line and change the content 53 | 54 | 55 | 4. Run the below command to initialize the configured 56 | 57 | ``` 58 | terraform init 59 | ``` 60 | 61 | 5. Run the below command to see what will be applied 62 | 63 | ``` 64 | terraform plan -out tfplan 65 | ``` 66 | 67 | 6. Run the below command to apply the code against your AWS environment 68 | 69 | ``` 70 | terraform apply tfplan 71 | ``` 72 | 73 | 7. Connect to the cluster with `kubectl` by running below command with updated cluster name and region after the cluster is created 74 | 75 | ``` 76 | aws eks update-kubeconfig --name --region 77 | ``` 78 | 79 | #### Cleaning up / Deleting resources 80 | 81 | 1. Run the beloe commands to delete all remaining AWS resources created by this module. You should see `Destroy complete!` message after a few minutes. 82 | 83 | ``` 84 | terraform destroy --auto-approve 85 | ``` 86 | 87 | ## Running as a module 88 | 89 | Call the EKS module by adding this to an existing Terraform file: 90 | 91 | ```hcl 92 | module "nvidia-eks" { 93 | source = "git::github.com/nvidia/nvidia-terraform-modules/eks" 94 | cluster_name = "nvidia-eks" 95 | } 96 | ``` 97 | In a production environment, we suggest pinning to a known tag of this Terraform module 98 | All configurable options for this module are listed below. 99 | If you need additional values added, please open a pull request. 100 | 101 | ## Requirements 102 | 103 | | Name | Version | 104 | |------|---------| 105 | | [terraform](#requirement\_terraform) | >= 1.3.4 | 106 | | [aws](#requirement\_aws) | ~>5.93.0 | 107 | | [kubernetes](#requirement\_kubernetes) | ~>2.19.0 | 108 | 109 | ## Providers 110 | 111 | | Name | Version | 112 | |------|---------| 113 | | [aws](#provider\_aws) | ~>5.93.0 | 114 | | [helm](#provider\_helm) | n/a | 115 | 116 | ## Modules 117 | 118 | | Name | Source | Version | 119 | |------|--------|---------| 120 | | [ebs\_csi\_irsa\_role](#module\_ebs\_csi\_irsa\_role) | terraform-aws-modules/iam/aws//modules/iam-role-for-service-accounts-eks | n/a | 121 | | [eks](#module\_eks) | terraform-aws-modules/eks/aws | 18.29.0 | 122 | | [vpc](#module\_vpc) | terraform-aws-modules/vpc/aws | 4.0.2 | 123 | 124 | ## Resources 125 | 126 | | Name | Type | 127 | |------|------| 128 | | [helm_release.gpu_operator](https://registry.terraform.io/providers/hashicorp/helm/latest/docs/resources/release) | resource | 129 | | [helm_release.nim_operator](https://registry.terraform.io/providers/hashicorp/helm/latest/docs/resources/release) | resource | 130 | | [aws_ami.lookup](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/data-sources/ami) | data source | 131 | | [aws_availability_zones.available](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/data-sources/availability_zones) | data source | 132 | | [aws_eks_cluster.eks](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/data-sources/eks_cluster) | data source | 133 | | [aws_instances.nodes](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/data-sources/instances) | data source | 134 | | [aws_region.current](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/data-sources/region) | data source | 135 | 136 | ## Inputs 137 | 138 | | Name | Description | Type | Default | Required | 139 | |------|-------------|------|---------|:--------:| 140 | | [additional\_node\_security\_groups\_rules](#input\_additional\_node\_security\_groups\_rules) | List of additional security group rules to add to the node security group created | `any` | `{}` | no | 141 | | [additional\_security\_group\_ids](#input\_additional\_security\_group\_ids) | list of additional security groups to add to nodes | `list(any)` | `[]` | no | 142 | | [additional\_user\_data](#input\_additional\_user\_data) | User data that is appended to the user data script after of the EKS bootstrap script. | `string` | `""` | no | 143 | | [aws\_profile](#input\_aws\_profile) | n/a | `string` | `"development"` | no | 144 | | [cidr\_block](#input\_cidr\_block) | CIDR for VPC | `string` | `"10.0.0.0/16"` | no | 145 | | [cluster\_name](#input\_cluster\_name) | n/a | `string` | n/a | yes | 146 | | [cluster\_version](#input\_cluster\_version) | Version of EKS to install on the control plane (Major and Minor version only, do not include the patch) | `string` | `"1.30"` | no | 147 | | [cpu\_instance\_type](#input\_cpu\_instance\_type) | CPU EC2 worker node instance type | `string` | `"t2.xlarge"` | no | 148 | | [cpu\_node\_pool\_additional\_user\_data](#input\_cpu\_node\_pool\_additional\_user\_data) | User data that is appended to the user data script after of the EKS bootstrap script on EKS-managed GPU node pool. | `string` | `""` | no | 149 | | [cpu\_node\_pool\_delete\_on\_termination](#input\_cpu\_node\_pool\_delete\_on\_termination) | Delete the VM nodes root filesystem on each node of the instance type. This is set to true by default, but can be changed when desired when using the 'local-storage provisioner' and are keeping important application data on the nodes | `bool` | `true` | no | 150 | | [cpu\_node\_pool\_root\_disk\_size\_gb](#input\_cpu\_node\_pool\_root\_disk\_size\_gb) | The size of the root disk on all GPU nodes in the EKS-managed GPU-only Node Pool. This is primarily for container image storage on the node | `number` | `512` | no | 151 | | [cpu\_node\_pool\_root\_volume\_type](#input\_cpu\_node\_pool\_root\_volume\_type) | The type of disk to use for the GPU node pool root disk (eg. gp2, gp3). Note, this is different from the type of disk used by applications via EKS Storage classes/PVs & PVCs | `string` | `"gp2"` | no | 152 | | [desired\_count\_cpu\_nodes](#input\_desired\_count\_cpu\_nodes) | Minimum number of CPU nodes in the Autoscaling Group | `string` | `"1"` | no | 153 | | [desired\_count\_gpu\_nodes](#input\_desired\_count\_gpu\_nodes) | Minimum number of GPU nodes in the Autoscaling Group | `string` | `"2"` | no | 154 | | [enable\_dns\_hostnames](#input\_enable\_dns\_hostnames) | Whether or not the Default VPC has DNS hostname support | `bool` | `true` | no | 155 | | [enable\_dns\_support](#input\_enable\_dns\_support) | Whether or not the Default VPC has DNS support | `bool` | `true` | no | 156 | | [enable\_nat\_gateway](#input\_enable\_nat\_gateway) | Should be true if you want to provision NAT Gateways for each of your private networks | `bool` | `true` | no | 157 | | [existing\_vpc\_details](#input\_existing\_vpc\_details) | Variables used for re-using existing VPC for vpc\_id & subnet\_id | 
object({
    vpc_id     = string
    subnet_ids = list(string)
  })
| `null` | no | 158 | | [gpu\_ami\_id](#input\_gpu\_ami\_id) | AMI ID of the EKS Ubuntu Image cooresponding to the region and version of the cluser. Not required as we do a lookup for this image | `string` | `""` | no | 159 | | [gpu\_instance\_type](#input\_gpu\_instance\_type) | GPU EC2 worker node instance type | `string` | `"g6e.12xlarge"` | no | 160 | | [gpu\_node\_pool\_additional\_user\_data](#input\_gpu\_node\_pool\_additional\_user\_data) | User data that is appended to the user data script after of the EKS bootstrap script on EKS-managed GPU node pool. | `string` | `""` | no | 161 | | [gpu\_node\_pool\_delete\_on\_termination](#input\_gpu\_node\_pool\_delete\_on\_termination) | Delete the VM nodes root filesystem on each node of the instance type. This is set to true by default, but can be changed when desired when using the 'local-storage provisioner' and are keeping important application data on the nodes | `bool` | `true` | no | 162 | | [gpu\_node\_pool\_root\_disk\_size\_gb](#input\_gpu\_node\_pool\_root\_disk\_size\_gb) | The size of the root disk on all GPU nodes in the EKS-managed GPU-only Node Pool. This is primarily for container image storage on the node | `number` | `512` | no | 163 | | [gpu\_node\_pool\_root\_volume\_type](#input\_gpu\_node\_pool\_root\_volume\_type) | The type of disk to use for the GPU node pool root disk (eg. gp2, gp3). Note, this is different from the type of disk used by applications via EKS Storage classes/PVs & PVCs | `string` | `"gp2"` | no | 164 | | [gpu\_operator\_driver\_version](#input\_gpu\_operator\_driver\_version) | The NVIDIA Driver version deployed with GPU Operator. Defaults to latest available. | `string` | `"570.124.06"` | no | 165 | | [gpu\_operator\_namespace](#input\_gpu\_operator\_namespace) | The namespace for the GPU operator deployment | `string` | `"gpu-operator"` | no | 166 | | [gpu\_operator\_version](#input\_gpu\_operator\_version) | Version of the GPU Operator to deploy. Defaults to latest available. | `string` | `"v25.3.0"` | no | 167 | | [install\_gpu\_operator](#input\_install\_gpu\_operator) | Whether to Install GPU Operator. Defaults to false available. | `string` | `"true"` | no | 168 | | [install\_nim\_operator](#input\_install\_nim\_operator) | Whether to Install NIM Operator. Defaults to false available. | `string` | `"false"` | no | 169 | | [max\_cpu\_nodes](#input\_max\_cpu\_nodes) | Maximum number of CPU nodes in the Autoscaling Group | `string` | `"2"` | no | 170 | | [max\_gpu\_nodes](#input\_max\_gpu\_nodes) | Maximum number of GPU nodes in the Autoscaling Group | `string` | `"5"` | no | 171 | | [min\_cpu\_nodes](#input\_min\_cpu\_nodes) | Minimum number of CPU nodes in the Autoscaling Group | `string` | `"0"` | no | 172 | | [min\_gpu\_nodes](#input\_min\_gpu\_nodes) | Minimum number of GPU nodes in the Autoscaling Group | `string` | `"2"` | no | 173 | | [nim\_operator\_namespace](#input\_nim\_operator\_namespace) | The namespace for the GPU operator deployment | `string` | `"nim-operator"` | no | 174 | | [nim\_operator\_version](#input\_nim\_operator\_version) | Version of the GPU Operator to deploy. Defaults to latest available. | `string` | `"v1.0.1"` | no | 175 | | [private\_subnets](#input\_private\_subnets) | List of subnet ranges for the Private VPC | `list(any)` | 
[
  "10.0.0.0/19",
  "10.0.32.0/19",
  "10.0.64.0/19"
]
| no | 176 | | [public\_subnets](#input\_public\_subnets) | List of subnet ranges for the Private VPC | `list(any)` | 
[
  "10.0.96.0/19",
  "10.0.128.0/19",
  "10.0.160.0/19"
]
| no | 177 | | [region](#input\_region) | AWS region to provision the Kubernetes Cluster | `string` | `"us-west-2"` | no | 178 | | [single\_nat\_gateway](#input\_single\_nat\_gateway) | Should be true if you want to provision a single shared NAT Gateway across all of your private networks | `bool` | `false` | no | 179 | | [ssh\_key](#input\_ssh\_key) | n/a | `string` | `""` | no | 180 | 181 | ## Outputs 182 | 183 | | Name | Description | 184 | |------|-------------| 185 | | [cluster\_ca\_certificate](#output\_cluster\_ca\_certificate) | n/a | 186 | | [cluster\_endpoint](#output\_cluster\_endpoint) | n/a | 187 | | [cpu\_node\_role\_name](#output\_cpu\_node\_role\_name) | IAM Node Role Bane for CPU node pools | 188 | | [gpu\_node\_role\_name](#output\_gpu\_node\_role\_name) | IAM Node Role Name for GPU node pools | 189 | | [kube\_exec\_api\_version](#output\_kube\_exec\_api\_version) | n/a | 190 | | [kube\_exec\_args](#output\_kube\_exec\_args) | n/a | 191 | | [kube\_exec\_command](#output\_kube\_exec\_command) | n/a | 192 | | [nodes](#output\_nodes) | n/a | 193 | | [oidc\_endpoint](#output\_oidc\_endpoint) | n/a | 194 | | [private\_subnet\_ids](#output\_private\_subnet\_ids) | n/a | 195 | | [public\_subnet\_ids](#output\_public\_subnet\_ids) | n/a | 196 | --------------------------------------------------------------------------------