no lb

clusters/k3s-dgx/apps/huihui-granite-inference.yaml

@@ -6,16 +6,15 @@ spec:
   model:
     uri: hf://huihui-ai/Huihui-granite-4.1-30b-abliterated
     name: huihui-ai/Huihui-granite-4.1-30b-abliterated
+  # Three replicas for load balancing
   replicas: 1
+  router:
+    scheduler: { }  # Default scheduler with default load balancing
+    route: { }
+    gateway: { }
   template:
     containers:
       - name: main
-        image: quay.io/pierdipi/vllm-cpu:latest
-        securityContext:
-          runAsNonRoot: false
-        env:
-          - name: VLLM_LOGGING_LEVEL
-            value: DEBUG
         resources:
           limits:
             cpu: "4"
@@ -26,11 +25,11 @@ spec:
             memory: 8Gi
             nvidia.com/gpu: "1"
         livenessProbe:
-          initialDelaySeconds: 30
+          httpGet:
+            path: /health
+            port: 8000
+            scheme: HTTPS
+          initialDelaySeconds: 120
           periodSeconds: 30
           timeoutSeconds: 30
           failureThreshold: 5
-  router:
-    gateway: {}
-    route: {}
-    scheduler: {}