chore: bump kubernetes resources

kubernetes/helm/values.yaml

@@ -5,11 +5,14 @@ ollama:
   image: ollama/ollama:latest
   servicePort: 11434
   resources:
-    limits:
+    requests:
       cpu: "2000m"
       memory: "2Gi"
+    limits:
+      cpu: "4000m"
+      memory: "4Gi"
       nvidia.com/gpu: "0"
-  volumeSize: 1Gi
+  volumeSize: 30Gi
   nodeSelector: {}
   tolerations: []
   service:
@@ -22,16 +25,19 @@ webui:
   image: ghcr.io/ollama-webui/ollama-webui:main
   servicePort: 8080
   resources:
-    limits:
+    requests:
       cpu: "500m"
       memory: "500Mi"
+    limits:
+      cpu: "1000m"
+      memory: "1Gi"
   ingress:
     enabled: true
     annotations:
     # Use appropriate annotations for your Ingress controller, e.g., for NGINX:
       # nginx.ingress.kubernetes.io/rewrite-target: /
     host: ollama.minikube.local
-  volumeSize: 1Gi
+  volumeSize: 2Gi
   nodeSelector: {}
   tolerations: []
   service:

kubernetes/manifest/base/ollama-statefulset.yaml

@@ -20,9 +20,13 @@ spec:
         ports:
         - containerPort: 11434
         resources:
-          limits:
+          requests:
             cpu: "2000m"
             memory: "2Gi"
+          limits:
+            cpu: "4000m"
+            memory: "4Gi"
+            nvidia.com/gpu: "0"
         volumeMounts:
         - name: ollama-volume
           mountPath: /root/.ollama
@@ -34,4 +38,4 @@ spec:
       accessModes: [ "ReadWriteOnce" ]
       resources:
         requests:
-          storage: 1Gi
+          storage: 30Gi

kubernetes/manifest/base/webui-deployment.yaml

@@ -19,9 +19,12 @@ spec:
         ports:
         - containerPort: 8080
         resources:
-          limits:
+          requests:
             cpu: "500m"
             memory: "500Mi"
+          limits:
+            cpu: "1000m"
+            memory: "1Gi"
         env:
         - name: OLLAMA_API_BASE_URL
           value: "http://ollama-service.ollama-namespace.svc.cluster.local:11434/api"

kubernetes/manifest/base/webui-pvc.yaml

@@ -9,4 +9,4 @@ spec:
   accessModes: ["ReadWriteOnce"]
   resources:
     requests:
-      storage: 1Gi
+      storage: 2Gi