From 3eb32b8d66e851817d6ad3307b65fab8383a8b60 Mon Sep 17 00:00:00 2001
From: Michael Thomson <me@michaelthomson.dev>
Date: Tue, 12 May 2026 13:26:08 -0400
Subject: [PATCH] nvidia gpu operator

---
 .../generic-cdi-plugin/daemonset.yaml         | 47 -------------------
 .../controllers/gpu-operator/release.yaml     | 24 ++++++++++
 .../controllers/gpu-operator/repository.yaml  |  8 ++++
 .../gpu-operator/time-slicing-config.yaml     | 13 +++++
 .../namespace-generic-cdi-plugin.yaml         |  4 --
 .../namespaces/namespace-gpu-operator.yaml    |  8 ++++
 6 files changed, 53 insertions(+), 51 deletions(-)
 delete mode 100644 infrastructure/controllers/generic-cdi-plugin/daemonset.yaml
 create mode 100644 infrastructure/controllers/gpu-operator/release.yaml
 create mode 100644 infrastructure/controllers/gpu-operator/repository.yaml
 create mode 100644 infrastructure/controllers/gpu-operator/time-slicing-config.yaml
 delete mode 100644 infrastructure/namespaces/namespace-generic-cdi-plugin.yaml
 create mode 100644 infrastructure/namespaces/namespace-gpu-operator.yaml

diff --git a/infrastructure/controllers/generic-cdi-plugin/daemonset.yaml b/infrastructure/controllers/generic-cdi-plugin/daemonset.yaml
deleted file mode 100644
index 5c5994f..0000000
--- a/infrastructure/controllers/generic-cdi-plugin/daemonset.yaml
+++ /dev/null
@@ -1,47 +0,0 @@
-apiVersion: apps/v1
-kind: DaemonSet
-metadata:
-  name: generic-cdi-plugin-daemonset
-  namespace: generic-cdi-plugin
-spec:
-  selector:
-    matchLabels:
-      name: generic-cdi-plugin
-  template:
-    metadata:
-      labels:
-        name: generic-cdi-plugin
-        app.kubernetes.io/component: generic-cdi-plugin
-        app.kubernetes.io/name: generic-cdi-plugin
-    spec:
-      containers:
-      - image: ghcr.io/olfillasodikno/generic-cdi-plugin:main
-        name: generic-cdi-plugin
-        command:
-          - /generic-cdi-plugin
-          - /var/run/cdi/nvidia-container-toolkit.json
-        imagePullPolicy: Always
-        securityContext:
-          privileged: true
-        tty: true
-        volumeMounts:
-        - name: kubelet
-          mountPath: /var/lib/kubelet
-        - name: nvidia-container-toolkit
-          mountPath: /var/run/cdi/nvidia-container-toolkit.json
-      volumes:
-      - name: kubelet
-        hostPath:
-          path: /var/lib/kubelet
-      - name: nvidia-container-toolkit
-        hostPath:
-          path: /var/run/cdi/nvidia-container-toolkit.json
-      affinity:
-        nodeAffinity:
-          requiredDuringSchedulingIgnoredDuringExecution:
-            nodeSelectorTerms:
-            - matchExpressions:
-              - key: "nixos-nvidia-cdi"
-                operator: In
-                values:
-                - "enabled"
diff --git a/infrastructure/controllers/gpu-operator/release.yaml b/infrastructure/controllers/gpu-operator/release.yaml
new file mode 100644
index 0000000..80eee1b
--- /dev/null
+++ b/infrastructure/controllers/gpu-operator/release.yaml
@@ -0,0 +1,24 @@
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: gpu-operator
+  namespace: gpu-operator
+spec:
+  chart:
+    spec:
+      chart: gpu-operator
+      version: v26.3.x
+      sourceRef:
+        kind: HelmRepository
+        name: nvidia
+  interval: 15m
+  releaseName: gpu-operator
+  values:
+    driver:
+      enabled: false
+    toolkit:
+      enabled: false
+    devicePlugin:
+      config:
+        name: time-slicing-config
+        default: time-slicing
diff --git a/infrastructure/controllers/gpu-operator/repository.yaml b/infrastructure/controllers/gpu-operator/repository.yaml
new file mode 100644
index 0000000..bbe46b3
--- /dev/null
+++ b/infrastructure/controllers/gpu-operator/repository.yaml
@@ -0,0 +1,8 @@
+apiVersion: source.toolkit.fluxcd.io/v1
+kind: HelmRepository
+metadata:
+  name: nvidia
+  namespace: gpu-operator
+spec:
+  interval: 15m
+  url: https://helm.ngc.nvidia.com/nvidia
diff --git a/infrastructure/controllers/gpu-operator/time-slicing-config.yaml b/infrastructure/controllers/gpu-operator/time-slicing-config.yaml
new file mode 100644
index 0000000..d9a4a39
--- /dev/null
+++ b/infrastructure/controllers/gpu-operator/time-slicing-config.yaml
@@ -0,0 +1,13 @@
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: time-slicing-config
+  namespace: gpu-operator
+data:
+  time-slicing: |
+    version: v1
+    sharing:
+      timeSlicing:
+        resources:
+          - name: nvidia.com/gpu
+            replicas: 5
diff --git a/infrastructure/namespaces/namespace-generic-cdi-plugin.yaml b/infrastructure/namespaces/namespace-generic-cdi-plugin.yaml
deleted file mode 100644
index dc4e556..0000000
--- a/infrastructure/namespaces/namespace-generic-cdi-plugin.yaml
+++ /dev/null
@@ -1,4 +0,0 @@
-apiVersion: v1
-kind: Namespace
-metadata:
-  name: generic-cdi-plugin
diff --git a/infrastructure/namespaces/namespace-gpu-operator.yaml b/infrastructure/namespaces/namespace-gpu-operator.yaml
new file mode 100644
index 0000000..c4104d2
--- /dev/null
+++ b/infrastructure/namespaces/namespace-gpu-operator.yaml
@@ -0,0 +1,8 @@
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: gpu-operator
+  labels:
+    pod-security.kubernetes.io/enforce: privileged
+    pod-security.kubernetes.io/audit: privileged
+    pod-security.kubernetes.io/warn: privileged