From bff6b83d1189153ad190f5912bc6260940ffed2b Mon Sep 17 00:00:00 2001
From: Brad Stein <Brad.Stein@gmail.com>
Date: Thu, 9 Oct 2025 18:28:20 -0500
Subject: [PATCH] gpu(titan-24): add RuntimeClass + NVIDIA device-plugin DS;
 enable containerd nvidia runtime

---
 .../gpu/device-plugin-tethys/daemonset.yaml   | 48 +++++++++++++++++++
 .../device-plugin-tethys/kustomization.yaml   |  4 ++
 .../profiles/tethys-only/kustomization.yaml   |  5 ++
 .../gpu/runtimeclass/nvidia-runtimeclass.yaml |  5 ++
 infrastructure/core/kustomization.yaml        |  3 +-
 5 files changed, 64 insertions(+), 1 deletion(-)
 create mode 100644 infrastructure/core/gpu/device-plugin-tethys/daemonset.yaml
 create mode 100644 infrastructure/core/gpu/device-plugin-tethys/kustomization.yaml
 create mode 100644 infrastructure/core/gpu/profiles/tethys-only/kustomization.yaml
 create mode 100644 infrastructure/core/gpu/runtimeclass/nvidia-runtimeclass.yaml

diff --git a/infrastructure/core/gpu/device-plugin-tethys/daemonset.yaml b/infrastructure/core/gpu/device-plugin-tethys/daemonset.yaml
new file mode 100644
index 0000000..44a47d9
--- /dev/null
+++ b/infrastructure/core/gpu/device-plugin-tethys/daemonset.yaml
@@ -0,0 +1,48 @@
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  name: nvidia-device-plugin-titan24
+  namespace: kube-system
+  labels:
+    app.kubernetes.io/name: nvidia-device-plugin
+    app.kubernetes.io/instance: titan24
+spec:
+  selector:
+    matchLabels:
+      app.kubernetes.io/name: nvidia-device-plugin
+      app.kubernetes.io/instance: titan24
+  template:
+    metadata:
+      labels:
+        app.kubernetes.io/name: nvidia-device-plugin
+        app.kubernetes.io/instance: titan24
+    spec:
+      nodeSelector:
+        kubernetes.io/hostname: titan-24
+        kubernetes.io/arch: amd64
+      tolerations:
+        - operator: Exists
+      priorityClassName: system-node-critical
+      runtimeClassName: nvidia
+      containers:
+        - name: nvidia-device-plugin-ctr
+          image: nvcr.io/nvidia/k8s-device-plugin:v0.16.2
+          imagePullPolicy: IfNotPresent
+          args:
+            - "--fail-on-init-error=false"
+            - "--device-list-strategy=envvar"
+            - "--mig-strategy=none"
+          securityContext:
+            privileged: true
+          env:
+            - name: NVIDIA_VISIBLE_DEVICES
+              value: "all"
+            - name: NVIDIA_DRIVER_CAPABILITIES
+              value: "compute,video,utility"
+          volumeMounts:
+            - name: device-plugin
+              mountPath: /var/lib/kubelet/device-plugins
+      volumes:
+        - name: device-plugin
+          hostPath:
+            path: /var/lib/kubelet/device-plugins
diff --git a/infrastructure/core/gpu/device-plugin-tethys/kustomization.yaml b/infrastructure/core/gpu/device-plugin-tethys/kustomization.yaml
new file mode 100644
index 0000000..48f1d3d
--- /dev/null
+++ b/infrastructure/core/gpu/device-plugin-tethys/kustomization.yaml
@@ -0,0 +1,4 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - daemonset.yaml
diff --git a/infrastructure/core/gpu/profiles/tethys-only/kustomization.yaml b/infrastructure/core/gpu/profiles/tethys-only/kustomization.yaml
new file mode 100644
index 0000000..c78c41c
--- /dev/null
+++ b/infrastructure/core/gpu/profiles/tethys-only/kustomization.yaml
@@ -0,0 +1,5 @@
+# infrastructure/core/gpu/daemonsets/profiles/tethys-only/kustomization.yaml
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - ../../device-plugin-tethys
diff --git a/infrastructure/core/gpu/runtimeclass/nvidia-runtimeclass.yaml b/infrastructure/core/gpu/runtimeclass/nvidia-runtimeclass.yaml
new file mode 100644
index 0000000..635f2b6
--- /dev/null
+++ b/infrastructure/core/gpu/runtimeclass/nvidia-runtimeclass.yaml
@@ -0,0 +1,5 @@
+apiVersion: node.k8s.io/v1
+kind: RuntimeClass
+metadata:
+name: nvidia
+handler: nvidia
diff --git a/infrastructure/core/kustomization.yaml b/infrastructure/core/kustomization.yaml
index 939a93d..f752d47 100644
--- a/infrastructure/core/kustomization.yaml
+++ b/infrastructure/core/kustomization.yaml
@@ -5,4 +5,5 @@ resources:
   - base
   # - gpu/profiles/jetson-only 
   # - gpu/profiles/minipc-and-jetson
-  - gpu/profiles/minipc-only 
+  # - gpu/profiles/minipc-only
+  - gpu/profiles/tethys-only
-- 
2.47.2