Set open webui to local-path same as ollama

manifests/ollama/values.yaml

@@ -1,15 +1,35 @@
 service:
   type: ClusterIP
 
+# Enable GPU support
+ollama:
+  gpu:
+    enabled: true
+    type: nvidia
+    number: 1
+
 resources:
   requests:
-    cpu: 250m
-    memory: 1Gi
-  limits:
     cpu: 2
     memory: 4Gi
+  limits:
+    cpu: 4
+    memory: 8Gi
+
+# Schedule on GPU worker node
+nodeSelector:
+  kubernetes.io/hostname: k3s-worker-5-gpu
+
+# Tolerate GPU node taint
+tolerations:
+  - key: gpu
+    operator: Exists
+    effect: NoSchedule
+  - key: nvidia.com/gpu
+    operator: Exists
+    effect: NoSchedule
 
 persistentVolume:
   enabled: true
-  size: 30Gi
-  storageClass: nfs-client
+  size: 50Gi
+  storageClass: local-path

manifests/open-webui/values.yaml

@@ -1,16 +1,17 @@
 service:
   type: ClusterIP
 
-env:
-  - name: OLLAMA_BASE_URL
-    value: http://ollama.ai-stack.svc.cluster.local:11434
+# IMPORTANT: disable the embedded Ollama that the chart can deploy
+ollama:
+  enabled: false
 
+# IMPORTANT: set BOTH vars to your existing Ollama service
 extraEnvVars:
   - name: OLLAMA_BASE_URL
     value: http://ollama:11434
   - name: OLLAMA_BASE_URLS
     value: http://ollama:11434
-    
+
 resources:
   requests:
     cpu: 100m
@@ -18,3 +19,21 @@ resources:
   limits:
     cpu: 500m
     memory: 1Gi
+
+persistence:
+  enabled: true
+  storageClass: local-path
+  size: 2Gi
+
+# Schedule on GPU worker node (same as Ollama for low latency)
+nodeSelector:
+  kubernetes.io/hostname: k3s-worker-5-gpu
+
+# Tolerate GPU node taint
+tolerations:
+  - key: gpu
+    operator: Exists
+    effect: NoSchedule
+  - key: nvidia.com/gpu
+    operator: Exists
+    effect: NoSchedule