add: nim

components/app-configs/nim/README.md

@@ -0,0 +1,19 @@
+# Notes
+
+## Quickstart
+
+Query NIM Open AI API
+
+```sh
+URL=https://$(oc get route -o go-template='{{.spec.host}}' nim)/v1/completions
+
+curl -s -X 'POST' \
+  "${URL}" \
+  -H 'accept: application/json' \
+  -H 'Content-Type: application/json' \
+  -d '{
+      "model": "meta/llama3-8b-instruct",
+      "prompt": "Once upon a time",
+      "max_tokens": 64
+    }' | jq .choices[0].text
+```

components/app-configs/nim/base/deployment.yaml

@@ -0,0 +1,69 @@
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: llama3
+spec:
+  replicas: 1
+  strategy:
+    type: Recreate
+  template:
+    spec:
+      containers:
+      - name: nim
+        image: 'nvcr.io/nim/meta/llama3-8b-instruct:1.0.0'
+        env:
+          - name: NGC_API_KEY
+            value: your-key-here
+        ports:
+        - containerPort: 8000
+          protocol: TCP
+          name: api
+        resources:
+          limits:
+            cpu: '2'
+            memory: 8Gi
+          requests:
+            cpu: 500m
+            memory: 8Gi
+        # livenessProbe:
+        #   httpGet:
+        #     path: /
+        #     port: http
+        #     scheme: HTTP
+        #   timeoutSeconds: 5
+        #   periodSeconds: 5
+        #   successThreshold: 1
+        #   failureThreshold: 3
+        # readinessProbe:
+        #   httpGet:
+        #     path: /
+        #     port: notebook
+        #     scheme: HTTP
+        #   timeoutSeconds: 1
+        #   periodSeconds: 10
+        #   successThreshold: 1
+        #   failureThreshold: 3
+        # startupProbe:
+        #   httpGet:
+        #     path: /
+        #     port: notebook
+        #     scheme: HTTP
+        #   timeoutSeconds: 1
+        #   periodSeconds: 5
+        #   successThreshold: 1
+        #   failureThreshold: 10
+        volumeMounts:
+          - name: shm
+            mountPath: /dev/shm
+          - name: nim-cache
+            mountPath: /opt/nim/.cache
+            subPath: data
+      volumes:
+        - name: shm
+          emptyDir:
+            medium: Memory
+            sizeLimit: 512Mi
+        - name: nim-cache
+          persistentVolumeClaim:
+            claimName: nim

components/app-configs/nim/base/kustomization.yaml

@@ -0,0 +1,13 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+labels:
+- includeSelectors: true
+  pairs:
+    component: nim
+
+resources:
+- deployment.yaml
+- pvc.yaml
+- route.yaml
+- service.yaml

components/app-configs/nim/base/pvc.yaml

@@ -0,0 +1,11 @@
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: nim
+spec:
+  accessModes:
+  - ReadWriteOnce
+  resources:
+    requests:
+      storage: 10Gi

components/app-configs/nim/base/route.yaml

@@ -0,0 +1,18 @@
+---
+apiVersion: route.openshift.io/v1
+kind: Route
+metadata:
+  name: nim
+  annotations:
+    argocd.argoproj.io/sync-wave: "1"
+spec:
+  port:
+    targetPort: 8000
+  tls:
+    insecureEdgeTerminationPolicy: Redirect
+    termination: edge
+  to:
+    kind: Service
+    name: api
+    weight: 100
+  wildcardPolicy: None

components/app-configs/nim/base/service.yaml

@@ -0,0 +1,12 @@
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: nim
+spec:
+  ports:
+  - name: api
+    port: 8000
+    targetPort: 8000
+  sessionAffinity: None
+  type: ClusterIP

components/app-configs/nim/components/gpu/kustomization.yaml

@@ -0,0 +1,7 @@
+apiVersion: kustomize.config.k8s.io/v1alpha1
+kind: Component
+
+patches:
+  - path: patch-deployment.yaml
+    target:
+      kind: Deployment

components/app-configs/nim/components/gpu/patch-deployment.yaml

@@ -0,0 +1,12 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: nim
+spec:
+  template:
+    spec:
+      containers:
+        - name: nim
+          resources:
+            limits:
+              nvidia.com/gpu: 1

components/app-configs/nim/components/namespace/kustomization.yaml

@@ -0,0 +1,5 @@
+apiVersion: kustomize.config.k8s.io/v1alpha1
+kind: Component
+
+resources:
+- namespace.yaml

components/app-configs/nim/components/namespace/namespace.yaml

@@ -0,0 +1,7 @@
+---
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: demo-nim
+  annotations:
+    openshift.io/display-name: "Demo - NIM"

components/app-configs/nim/overlays/default/kustomization.yaml

@@ -0,0 +1,5 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+resources:
+- ../../base

components/app-configs/nim/overlays/demo/kustomization.yaml

@@ -0,0 +1,11 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+namespace: demo-nim
+
+components:
+- ../../components/gpu
+- ../../components/namespace
+
+resources:
+- ../../base

components/app-configs/nim/overlays/gpu/kustomization.yaml

@@ -0,0 +1,8 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+
+components:
+- ../../components/gpu
+
+resources:
+- ../../base

components/app-configs/notebook-standalone/components/nemo/patch-deployment.yaml

@@ -17,7 +17,8 @@ spec:
             - |
               #!/bin/sh
               jupyter lab \
-                --ServerApp.ip=0.0.0.0:8888 \
+                --ServerApp.ip=0.0.0.0 \
+                --ServerApp.port=8888 \
                 --ServerApp.allow_origin="*" \
                 --ServerApp.open_browser=False \
                 --ServerApp.token=''