wait for tgi server to start before starting server

scripts/docker/tgi/compose.yaml

@@ -15,11 +15,6 @@ services:
       - HF_MODULES_CACHE=/data
       - HF_HUB_CACHE=/data
     command: ["--dtype", "bfloat16", "--usage-stats", "on", "--sharded", "false", "--model-id", "meta-llama/Llama-3.1-8B-Instruct", "--port", "5009", "--cuda-memory-fraction", "0.3"]
-    healthcheck:
-      test: ["CMD", "curl", "-f", "http://localhost:5009/health"]
-      interval: 5s
-      timeout: 5s
-      retries: 10
     deploy:
       resources:
         reservations:
@@ -34,11 +29,15 @@ services:
             # reservation.
             capabilities: [gpu]
     runtime: nvidia
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://text-generation-inference:5009/health"]
+      interval: 5s
+      timeout: 5s
+      retries: 30
   llamastack-local-cpu:
     depends_on:
       text-generation-inference:
         condition: service_healthy
-    restart: on-failure
     image: llamastack-local-cpu
     network_mode: "host"
     volumes:
@@ -47,9 +46,10 @@ services:
       - ./tgi-run.yaml:/root/llamastack-run-tgi.yaml
     ports:
       - "5000:5000"
-    command: ["--yaml_config", "/root/llamastack-run-tgi.yaml"]
+    # Hack: wait for TGI server to start before starting docker
-    healthcheck:
+    entrypoint: bash -c "sleep 60; python -m llama_stack.distribution.server.server --yaml_config /root/llamastack-run-tgi.yaml"
-      test: ["CMD", "curl", "-f", "http://localhost:5009/health"]
+    restart_policy:
-      interval: 5s
+      condition: on-failure
-      timeout: 5s
+      delay: 3s
-      retries: 10
+      max_attempts: 5
+      window: 60s