Added gpu and healthcheck to vllm server, added vllm test server

Author: adsarver

docker-compose.components.test.yml

@@ -40,6 +40,23 @@ services:
   gemini-video-summarization:
     <<: *test-args
 
+  gemini-video-summarization-vllm-tests:
+    image: ${REGISTRY}openmpf_gemini_video_summarization_tests:${TAG}
+    build:
+      context: ${OPENMPF_PROJECTS_PATH}/openmpf-components/python/GeminiVideoSummarization
+      target: vllm-tests
+      args:
+        BUILD_REGISTRY: ${REGISTRY}
+        BUILD_TAG: ${TAG}
+        RUN_TESTS: 'true'
+    environment:
+      RUN_VLLM_TESTS: 'true'
+      GEMINI_TEST_OPENAI_BASE_URL: http://gemini-video-summarization-server:8000/v1
+      GEMINI_TEST_MODEL_NAME: ${GEMINI_MODEL_NAME:-google/gemma-4-12B-it}
+    depends_on:
+      gemini-video-summarization-server:
+        condition: service_healthy
+
   mog-motion-detection:
     <<: *test-args
 

docker-compose.components.yml

@@ -151,8 +151,10 @@ services:
       MPF_PROP_PROCESS_FPS: ${GEMINI_PROCESS_FPS:-1.0}
       MPF_PROP_OPENAI_REQUEST_TIMEOUT_SECONDS: ${GEMINI_OPENAI_REQUEST_TIMEOUT_SECONDS:-600}
     depends_on:
-      - workflow-manager
-      - gemini-video-summarization-server
+      workflow-manager:
+        condition: service_started
+      gemini-video-summarization-server:
+        condition: service_healthy
 
   gemini-video-summarization-server:
     image: ${REGISTRY}openmpf_gemini_video_summarization_server:${TAG}
@@ -162,14 +164,20 @@ services:
     ipc: host
     volumes:
       - ${GEMINI_HF_CACHE:-${HOME}/.cache/huggingface}:/root/.cache/huggingface
+    healthcheck:
+      test: ["CMD", "curl", "--fail", "--silent", "--show-error", "--max-time", "5", "http://localhost:8000/health"]
+      interval: 10s
+      timeout: 6s
+      retries: 120
+      start_period: 30s
     deploy:
       mode: global
-      # resources:
-      #   reservations:
-      #     devices:
-      #       - driver: nvidia
-      #         device_ids: ['0']
-      #         capabilities: [gpu]
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              device_ids: ['0']
+              capabilities: [gpu]
 
   fasttext-language-detection:
     <<: *component-base