cleanup

.gitignore

@@ -0,0 +1,40 @@
+# FluxCD secrets
+*.yaml.enc
+*.yaml.dec
+flux-gitea-key
+flux-gitea-key.pub
+
+# Kubernetes secrets
+*-secret.yaml
+secrets/
+
+# Temporary files
+*.tmp
+*.bak
+*.swp
+*~
+
+# IDE files
+.vscode/
+.idea/
+*.iml
+
+# OS files
+.DS_Store
+Thumbs.db
+
+# Logs
+*.log
+logs/
+
+# Build artifacts
+dist/
+build/
+*.tar.gz
+
+# Environment files
+.env
+.env.local
+
+# Helm
+charts/*.tgz

clusters/k3s-dgx/flux-system/gotk-components.yaml

@@ -47,4 +47,4 @@ metadata:
   namespace: flux-system
 spec:
   interval: 30m
-  url: https://nvidia.github.io/k8s-device-plugin
+  url: https://nvidia.github.io/k8s-device-plugin

clusters/k3s-dgx/gpu-support/nvidia-device-plugin.yaml

@@ -1,39 +0,0 @@
-apiVersion: v1
-kind: Namespace
-metadata:
-  name: gpu-operator
----
-apiVersion: apps/v1
-kind: DaemonSet
-metadata:
-  name: nvidia-device-plugin-daemonset
-  namespace: gpu-operator
-spec:
-  selector:
-    matchLabels:
-      name: nvidia-device-plugin-ds
-  template:
-    metadata:
-      labels:
-        name: nvidia-device-plugin-ds
-    spec:
-      tolerations:
-        - key: nvidia.com/gpu
-          operator: Exists
-          effect: NoSchedule
-      containers:
-        - image: nvcr.io/nvidia/k8s-device-plugin:v0.14.0
-          name: nvidia-device-plugin-ctr
-          args: ["--fail-on-init-error=false"]
-          env:
-            - name: NVIDIA_VISIBLE_DEVICES
-              value: "all"
-            - name: MIG_STRATEGY
-              value: "single"
-          volumeMounts:
-            - name: device-plugin
-              mountPath: /var/lib/kubelet/device-plugins
-      volumes:
-        - name: device-plugin
-          hostPath:
-            path: /var/lib/kubelet/device-plugins

clusters/k3s-dgx/gpu-support/runtime-class.yaml

@@ -1,16 +0,0 @@
-apiVersion: node.k8s.io/v1
-kind: RuntimeClass
-metadata:
-  name: nvidia
-handler: nvidia
-overhead:
-  podFixed:
-    memory: "1Gi"
-    cpu: "500m"
-scheduling:
-  nodeSelector:
-    nvidia.com/gpu.present: "true"
-  tolerations:
-    - key: nvidia.com/gpu
-      operator: Exists
-      effect: NoSchedule

clusters/k3s-dgx/kserve/gpu-serving-runtime.yaml

@@ -1,42 +0,0 @@
-apiVersion: serving.kserve.io/v1beta1
-kind: ServingRuntime
-metadata:
-  name: kserve-gpu-runtime
-  namespace: kserve
-spec:
-  supportedModelFormats:
-    - name: tensorflow
-      version: "2"
-      autoSelect: true
-    - name: pytorch
-      version: "2"
-      autoSelect: true
-    - name: sklearn
-      version: "1"
-      autoSelect: true
-    - name: xgboost
-      version: "1"
-      autoSelect: true
-  protocol: v1
-  protocolVersions:
-    - v1
-    - v2
-  containers:
-    - name: kserve-container
-      image: kserve/sklearnserver-gpu:latest
-      resources:
-        requests:
-          cpu: "1"
-          memory: "2Gi"
-          nvidia.com/gpu: "1"
-        limits:
-          cpu: "4"
-          memory: "8Gi"
-          nvidia.com/gpu: "1"
-      env:
-        - name: NVIDIA_VISIBLE_DEVICES
-          value: "all"
-        - name: NVIDIA_DRIVER_CAPABILITIES
-          value: "compute,utility"
-  multiModel: false
-  disabled: false

clusters/k3s-dgx/kserve/kserve-controller.yaml

@@ -34,9 +34,6 @@ spec:
     config:
       ingress:
         className: istio
-      storage:
-        initialCapacity: 10Gi
-        storageClassName: local-path
     knative:
       enabled: true
     istio:

clusters/k3s-dgx/kserve/kustomization.yaml

@@ -5,6 +5,3 @@ resources:
   - kserve-namespace.yaml
   - kserve-controller.yaml
   - istio-gateway.yaml
-  # - gpu-serving-runtime.yaml
-  # - model-storage-pvc.yaml
-  # - storage-config.yaml

clusters/k3s-dgx/kserve/model-storage-pvc.yaml

@@ -1,12 +0,0 @@
-apiVersion: v1
-kind: PersistentVolumeClaim
-metadata:
-  name: model-storage-pvc
-  namespace: kserve
-spec:
-  accessModes:
-    - ReadWriteOnce
-  resources:
-    requests:
-      storage: 50Gi
-  storageClassName: local-path

clusters/k3s-dgx/kserve/storage-config.yaml

@@ -1,20 +0,0 @@
-apiVersion: v1
-kind: ConfigMap
-metadata:
-  name: storage-config
-  namespace: kserve
-data:
-  storageConfig.yaml: |
-    defaultStorageUri: "pvc://model-storage"
-    storageSpec:
-      - name: model-storage
-        type: pvc
-        pvcSpec:
-          accessModes:
-            - ReadWriteOnce
-          resources:
-            requests:
-              storage: 50Gi
-          storageClassName: local-path
-    modelCacheSize: 10Gi
-    modelCacheMemory: 2Gi