gpu to nvidia

aws-observability · Jan 19, 2024 · e5ee517 · e5ee517
1 parent 5526f14
commit e5ee517
Show file tree

Hide file tree

Showing 4 changed files with 17 additions and 17 deletions.
diff --git a/modules/eks-monitoring/dashboards.tf b/modules/eks-monitoring/dashboards.tf
@@ -95,23 +95,23 @@ YAML
   depends_on = [module.external_secrets]
 }
 
-# gpu dashboards
-resource "kubectl_manifest" "gpu_monitoring_dashboards" {
+# nvidia dashboards
+resource "kubectl_manifest" "nvidia_monitoring_dashboards" {
   yaml_body  = <<YAML
 apiVersion: kustomize.toolkit.fluxcd.io/v1beta2
 kind: Kustomization
 metadata:
-  name: ${local.gpu_monitoring_config.flux_kustomization_name}
+  name: ${local.nvidia_monitoring_config.flux_kustomization_name}
   namespace: flux-system
 spec:
   interval: 1m0s
-  path: ${local.gpu_monitoring_config.flux_kustomization_path}
+  path: ${local.nvidia_monitoring_config.flux_kustomization_path}
   prune: true
   sourceRef:
     kind: GitRepository
-    name: ${local.gpu_monitoring_config.flux_gitrepository_name}
+    name: ${local.nvidia_monitoring_config.flux_gitrepository_name}
 YAML
-  count      = var.enable_gpu_monitoring ? 1 : 0
+  count      = var.enable_nvidia_monitoring ? 1 : 0
   depends_on = [module.external_secrets]
 }
 

diff --git a/modules/eks-monitoring/locals.tf b/modules/eks-monitoring/locals.tf
@@ -132,13 +132,13 @@ locals {
     }
   }
 
-  gpu_monitoring_config = {
+  nvidia_monitoring_config = {
     # can be overriden by providing a config
-    flux_gitrepository_name   = try(var.gpu_monitoring_config.flux_gitrepository_name, var.flux_gitrepository_name)
-    flux_gitrepository_url    = try(var.gpu_monitoring_config.flux_gitrepository_url, var.flux_gitrepository_url)
-    flux_gitrepository_branch = try(var.gpu_monitoring_config.flux_gitrepository_branch, var.flux_gitrepository_branch)
-    flux_kustomization_name   = try(var.gpu_monitoring_config.flux_kustomization_name, "grafana-dashboards-adothealth")
-    flux_kustomization_path   = try(var.gpu_monitoring_config.flux_kustomization_path, "./artifacts/grafana-operator-manifests/eks/gpu")
+    flux_gitrepository_name   = try(var.nvidia_monitoring_config.flux_gitrepository_name, var.flux_gitrepository_name)
+    flux_gitrepository_url    = try(var.nvidia_monitoring_config.flux_gitrepository_url, var.flux_gitrepository_url)
+    flux_gitrepository_branch = try(var.nvidia_monitoring_config.flux_gitrepository_branch, var.flux_gitrepository_branch)
+    flux_kustomization_name   = try(var.nvidia_monitoring_config.flux_kustomization_name, "grafana-dashboards-adothealth")
+    flux_kustomization_path   = try(var.nvidia_monitoring_config.flux_kustomization_path, "./artifacts/grafana-operator-manifests/eks/gpu")
   }
 
   kubeproxy_monitoring_config = {

diff --git a/modules/eks-monitoring/main.tf b/modules/eks-monitoring/main.tf
@@ -198,7 +198,7 @@ module "helm_addon" {
     },
     {
       name  = "enableGpuMonitoring"
-      value = var.enable_gpu_monitoring
+      value = var.enable_nvidia_monitoring
     },
     {
       name  = "serviceAccount"

diff --git a/modules/eks-monitoring/variables.tf b/modules/eks-monitoring/variables.tf
@@ -552,14 +552,14 @@ variable "enable_adotcollector_metrics" {
   default     = true
 }
 
-variable "enable_gpu_monitoring" {
-  description = "Enables monitoring of GPU metrics"
+variable "enable_nvidia_monitoring" {
+  description = "Enables monitoring of nvidia metrics"
   type        = bool
   default     = true
 }
 
-variable "gpu_monitoring_config" {
-  description = "Config object for GPU monitoring"
+variable "nvidia_monitoring_config" {
+  description = "Config object for nvidia monitoring"
   type = object({
     flux_gitrepository_name   = string
     flux_gitrepository_url    = string