fix: add ingress to llm example (#1349)

examples/llm/README.md

@@ -232,10 +232,10 @@ Once the deployment is complete, you can test it using:
 export FRONTEND_POD=$(kubectl get pods -n ${KUBE_NS} | grep "${DEPLOYMENT_NAME}-frontend" | sort -k1 | tail -n1 | awk '{print $1}')
 
 # Forward the pod's port to localhost
-kubectl port-forward pod/$FRONTEND_POD 8000:8000 -n ${KUBE_NS}
+kubectl port-forward pod/$FRONTEND_POD 3000:3000 -n ${KUBE_NS}
 
 # Test the API endpoint
-curl localhost:8000/v1/chat/completions \
+curl localhost:3000/v1/chat/completions \
   -H "Content-Type: application/json" \
   -d '{
     "model": "deepseek-ai/DeepSeek-R1-Distill-Llama-8B",

examples/llm/components/frontend.py

@@ -14,6 +14,7 @@
 # limitations under the License.
 
 import logging
+import os
 import subprocess
 from pathlib import Path
 
@@ -23,12 +24,15 @@ from components.worker import VllmWorker
 from pydantic import BaseModel
 
 from dynamo import sdk
-from dynamo.sdk import async_on_shutdown, depends, service
+from dynamo.sdk import api, async_on_shutdown, depends, service
 from dynamo.sdk.lib.config import ServiceConfig
 from dynamo.sdk.lib.image import DYNAMO_IMAGE
 
 logger = logging.getLogger(__name__)
 
+# TODO: temp workaround to avoid port conflict with subprocess HTTP server; remove this once ingress is fixed
+os.environ["DYNAMO_PORT"] = "3999"
+
 
 def get_http_binary_path():
     """Find the HTTP binary path in SDK or fallback to 'http' command."""
@@ -105,6 +109,17 @@ class Frontend:
             stderr=None,
         )
 
+    @api()
+    def dummy_api(self) -> None:
+        """
+        Dummy API to enable the HTTP server for the Dynamo operator.
+        This API is not used by the model.
+
+        NOTE: this is a temporary solution to expose ingress
+        for the LLM examples. Will be fixed and removed in the future.
+        The resulting api_endpoints in dynamo.yaml will be incorrect.
+        """
+
     @async_on_shutdown
     def cleanup(self):
         """Clean up resources before shutdown."""