Update tfserve and torchserve example to include metrics export (#1623)

Co-authored-by: Nim Jayawardena <[email protected]>

ai-ml/gke-online-serving-single-gpu/src/gke-config/deployment-tfserve.yaml

@@ -35,7 +35,7 @@ spec:
       containers:
       - name: tfserve-server
         image: tensorflow/serving:2.13.1-gpu
-        command: [ "tensorflow_model_server", "--model_name=$MODEL_NAME", "--model_base_path=/data/tfserve-model-repository/$MODEL_NAME", "--rest_api_port=8000"  ]
+        command: [ "tensorflow_model_server", "--model_name=$MODEL_NAME", "--model_base_path=/data/tfserve-model-repository/$MODEL_NAME", "--rest_api_port=8000", "--monitoring_config_file=/data/tfserve-model-repository/monitoring_config.txt" ]
         ports:
         - name: http
           containerPort: 8000

ai-ml/gke-online-serving-single-gpu/src/gke-config/service-tfserve.yaml

@@ -23,10 +23,10 @@ spec:
   ports:
     - port: 8500
       targetPort: 8500
-      name: http
+      name: grpc
     - port: 8000
       targetPort: 8000
-      name: grpc
+      name: http
   selector:
     app: tfserve
 

ai-ml/gke-online-serving-single-gpu/src/tfserve-model-repository/monitoring_config.txt

@@ -0,0 +1,4 @@
+prometheus_config {
+  enable: true,
+  path: "/monitoring/prometheus/metrics"
+}

ai-ml/t5-model-serving/model/config.properties

@@ -15,6 +15,7 @@
 inference_address=http://0.0.0.0:8080
 management_address=http://0.0.0.0:8081
 metrics_address=http://0.0.0.0:8082
+metrics_mode=prometheus
 number_of_netty_threads=32
 job_queue_size=1000
 install_py_dep_per_model=true