opendatahub/vllm: include adapter smoke test for main (openshift#52662)

* opendatahub-io/vllm: use Dockerfile.ubi for main branch

* opendatahub-io/vllm: add grpc test to main branch smoke test

ci-operator/config/opendatahub-io/vllm/opendatahub-io-vllm-main.yaml

@@ -56,13 +56,15 @@ tests:
 
     # we will need to download test models off HF hub
     unset HF_HUB_OFFLINE
-    # spin up the server and run it in the background, allowing for images download
+
+    # spin up the OpenAPI server in the background
     python -m vllm.entrypoints.openai.api_server &
     server_pid=$!
 
     # wait for the server to be up
     sleep 60
 
+    # OpenAI API tests
     curl -v --no-progress-meter --fail-with-body \
       localhost:8000/v1/models | python -m json.tool || \
       (kill -9 $server_pid && exit 1)
@@ -76,7 +78,35 @@ tests:
       localhost:8000/v1/completions | python -m json.tool || \
         (kill -9 $server_pid && exit 1)
 
-    echo "success"
+    echo "OpenAI API success"
+    kill -9 $server_pid
+
+    # spin up the grpc server in the background
+    python -m vllm_tgis_adapter &
+    server_pid=$!
+
+    # wait for the server to be up
+    sleep 60
+
+
+    # get grpcurl
+    curl --no-progress-meter --location --output grpcurl.tar.gz \
+      https://github.com/fullstorydev/grpcurl/releases/download/v1.9.1/grpcurl_1.9.1_linux_x86_64.tar.gz
+    tar -xf grpcurl.tar.gz
+
+    # get grpc proto
+    curl --no-progress-meter --location --remote-name \
+      https://github.com/opendatahub-io/text-generation-inference/raw/main/proto/generation.proto
+
+    # GRPC API test
+    ./grpcurl -v \
+        -plaintext \
+        -proto generation.proto \
+        -d '{ "requests": [{"text": "A red fedora symbolizes "}]}' \
+        localhost:8033 \
+        fmaas.GenerationService/Generate || (kill -9 $server_pid && exit 1)
+
+    echo "GRPC API success"
     kill -9 $server_pid
   container:
     clone: false