quick fix

docs/source/distributions/k8s/port-foward.sh

@@ -1,4 +1,48 @@
-kubectl port-forward svc/llama-stack-ui-service 8322:8322 &
-kubectl port-forward svc/llama-stack-service 8321:8321 & 
-kubectl port-forward svc/jaeger-dev-query 16686:16686 -n observability & 
-kubectl port-forward svc/kube-prometheus-stack-1754270486-grafana 3000:3000 -n prometheus
+#!/bin/bash
+
+# Function to port-forward to a pod with fallback to service
+port_forward_with_fallback() {
+  local namespace=$1
+  local label_selector=$2
+  local service_name=$3
+  local local_port=$4
+  local pod_port=$5
+
+  echo "Attempting to port-forward to pod with label $label_selector in namespace $namespace..."
+
+  # Try to get pod name using the label selector
+  POD_NAME=$(kubectl get pods -n $namespace -l $label_selector -o jsonpath="{.items[0].metadata.name}" 2>/dev/null)
+
+  if [ -n "$POD_NAME" ]; then
+    echo "Found pod: $POD_NAME. Port-forwarding $local_port:$pod_port..."
+    kubectl port-forward -n $namespace $POD_NAME $local_port:$pod_port &
+  else
+    echo "No pods found with label $label_selector in namespace $namespace."
+    echo "Falling back to service port-forwarding for $service_name..."
+    kubectl port-forward -n $namespace svc/$service_name $local_port:$pod_port &
+  fi
+}
+
+# Port-forward to llama-stack-ui pod or service
+port_forward_with_fallback "default" "app.kubernetes.io/name=llama-stack-ui" "llama-stack-ui-service" 8322 8322
+
+# Port-forward to llama-stack server pod or service
+port_forward_with_fallback "default" "app.kubernetes.io/name=llama-stack,app.kubernetes.io/component=server" "llama-stack-service" 8321 8321
+
+# Port-forward to jaeger query pod or service in observability namespace
+port_forward_with_fallback "observability" "app.kubernetes.io/component=query,app.kubernetes.io/instance=jaeger-dev" "jaeger-dev-query" 16686 16686
+
+# Port-forward to grafana pod or service in prometheus namespace
+kubectl port-forward svc/kube-prometheus-stack-1754270486-grafana 3000:80 -n prometheus
+
+echo "Port-forwarding started for all components."
+echo "Access the services at:"
+echo "  - Llama Stack UI: http://localhost:8322"
+echo "  - Llama Stack API: http://localhost:8321"
+echo "  - Jaeger UI: http://localhost:16686"
+echo "  - Grafana: http://localhost:3000"
+echo ""
+echo "Press Ctrl+C to stop all port-forwarding processes."
+
+# Wait for all background processes to complete
+wait

docs/source/distributions/k8s/stack-k8s.yaml.template

@@ -59,12 +59,12 @@ spec:
         image: llamastack/distribution-starter:0.2.15
         resources:
           requests:
-            memory: "512Mi"
-            cpu: "500m"
+            memory: "2Gi"
+            cpu: "8000m"
             ephemeral-storage: "6Gi"
           limits:
-            memory: "1Gi"
-            cpu: "1000m"
+            memory: "2Gi"
+            cpu: "8000m"
             ephemeral-storage: "6Gi"
         env:
         - name: ENABLE_CHROMADB
@@ -74,7 +74,7 @@ spec:
         - name: VLLM_URL
           value: http://vllm-server.default.svc.cluster.local:8001/v1
         - name: VLLM_MAX_TOKENS
-          value: "3072"
+          value: "80000"
         - name: OTEL_EXPORTER_OTLP_ENDPOINT
           value: http://jaeger-dev-collector.observability:4318
         - name: OTEL_SERVICE_NAME

llama_stack/providers/utils/inference/stream_utils.py

@@ -111,7 +111,8 @@ async def stream_and_store_openai_completion(
                 final_logprobs = OpenAIChoiceLogprobs(content=logprobs_content) if logprobs_content else None
 
                 # Ensure finish_reason is a string as required by OpenAIChoice
-                finish_reason = choice_data["finish_reason"] or "unknown"
+                # Using explicit None check to guarantee a string value
+                finish_reason = choice_data["finish_reason"] if choice_data["finish_reason"] is not None else "unknown"
                 assembled_choices.append(
                     OpenAIChoice(
                         finish_reason=finish_reason,