LCORE-536 Add error handling for bad llama-stack URLs; no more silent fails

src/app/main.py

@@ -8,6 +8,7 @@
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import JSONResponse
 from starlette.routing import Mount, Route, WebSocketRoute
+from llama_stack_client import APIConnectionError
 
 from authorization.azure_token_manager import AzureEntraIDManager
 import metrics
@@ -50,10 +51,23 @@ async def lifespan(_app: FastAPI) -> AsyncIterator[None]:
                 "Token refresh will be retried on next Azure request."
             )
 
-    await AsyncLlamaStackClientHolder().load(configuration.configuration.llama_stack)
+    llama_stack_config = configuration.configuration.llama_stack
+    await AsyncLlamaStackClientHolder().load(llama_stack_config)
     client = AsyncLlamaStackClientHolder().get_client()
     # check if the Llama Stack version is supported by the service
-    await check_llama_stack_version(client)
+    try:
+        await check_llama_stack_version(client)
+    except APIConnectionError as e:
+        llama_stack_url = llama_stack_config.url
+        logger.error(
+            "Failed to connect to Llama Stack at '%s'. "
+            "Please verify that the 'llama_stack.url' configuration is correct "
+            "and that the Llama Stack service is running and accessible. "
+            "Original error: %s",
+            llama_stack_url,
+            e,
+        )
+        raise
 
     logger.info("Registering MCP servers")
     await register_mcp_servers_async(logger, configuration.configuration)

src/client.py

@@ -70,8 +70,10 @@ def _load_service_client(self, config: LlamaStackConfiguration) -> None:
             "Using timeout of %d seconds for Llama Stack requests", config.timeout
         )
         api_key = config.api_key.get_secret_value() if config.api_key else None
+        # Convert AnyHttpUrl to string for the client
+        base_url = str(config.url) if config.url else None
         self._lsc = AsyncLlamaStackClient(
-            base_url=config.url, api_key=api_key, timeout=config.timeout
+            base_url=base_url, api_key=api_key, timeout=config.timeout
         )
 
     def _enrich_library_config(

src/models/config.py

@@ -533,10 +533,11 @@ class LlamaStackConfiguration(ConfigurationBase):
       - [Build AI Applications with Llama Stack](https://llamastack.github.io/)
     """
 
-    url: Optional[str] = Field(
+    url: Optional[AnyHttpUrl] = Field(
         None,
         title="Llama Stack URL",
-        description="URL to Llama Stack service; used when library mode is disabled",
+        description="URL to Llama Stack service; used when library mode is disabled. "
+        "Must be a valid HTTP or HTTPS URL.",
     )
 
     api_key: Optional[SecretStr] = Field(

tests/integration/test_configuration.py

@@ -69,7 +69,7 @@ def test_loading_proper_configuration(configuration_filename: str) -> None:
     # check 'llama_stack' section
     ls_config = cfg.llama_stack_configuration
     assert ls_config.use_as_library_client is False
-    assert ls_config.url == "http://localhost:8321"
+    assert str(ls_config.url) == "http://localhost:8321/"
     assert ls_config.api_key is not None
     assert ls_config.api_key.get_secret_value() == "xyzzy"
 

tests/unit/models/config/test_authentication_configuration.py

@@ -307,7 +307,7 @@ def test_authentication_configuration_in_config_noop() -> None:
         llama_stack=LlamaStackConfiguration(
             use_as_library_client=True,
             library_client_config_path="tests/configuration/run.yaml",
-            url="localhost",
+            url="http://localhost",
             api_key=SecretStr(""),
             timeout=60,
         ),
@@ -346,7 +346,7 @@ def test_authentication_configuration_skip_readiness_probe() -> None:
         llama_stack=LlamaStackConfiguration(
             use_as_library_client=True,
             library_client_config_path="tests/configuration/run.yaml",
-            url="localhost",
+            url="http://localhost",
             api_key=SecretStr(""),
             timeout=60,
         ),
@@ -393,7 +393,7 @@ def test_authentication_configuration_in_config_k8s() -> None:
         llama_stack=LlamaStackConfiguration(
             use_as_library_client=True,
             library_client_config_path="tests/configuration/run.yaml",
-            url="localhost",
+            url="http://localhost",
             api_key=SecretStr(""),
             timeout=60,
         ),
@@ -450,7 +450,7 @@ def test_authentication_configuration_in_config_rh_identity() -> None:
         llama_stack=LlamaStackConfiguration(
             use_as_library_client=True,
             library_client_config_path="tests/configuration/run.yaml",
-            url="localhost",
+            url="http://localhost",
             api_key=SecretStr(""),
             timeout=60,
         ),
@@ -497,7 +497,7 @@ def test_authentication_configuration_in_config_jwktoken() -> None:
         llama_stack=LlamaStackConfiguration(
             use_as_library_client=True,
             library_client_config_path="tests/configuration/run.yaml",
-            url="localhost",
+            url="http://localhost",
             api_key=SecretStr(""),
             timeout=60,
         ),

tests/unit/models/config/test_llama_stack_configuration.py

@@ -1,6 +1,7 @@
 """Unit tests for LlamaStackConfiguration model."""
 
 import pytest
+from pydantic import ValidationError
 
 from utils.checks import InvalidConfigurationError
 
@@ -89,3 +90,37 @@ def test_llama_stack_wrong_configuration_no_config_file() -> None:
         LlamaStackConfiguration(
             use_as_library_client=True
         )  # pyright: ignore[reportCallIssue]
+
+
+def test_llama_stack_configuration_valid_http_url() -> None:
+    """Test that valid HTTP URLs are accepted."""
+    config = LlamaStackConfiguration(
+        url="http://localhost:8321"
+    )  # pyright: ignore[reportCallIssue]
+    assert config is not None
+    assert str(config.url) == "http://localhost:8321/"
+
+
+def test_llama_stack_configuration_valid_https_url() -> None:
+    """Test that valid HTTPS URLs are accepted."""
+    config = LlamaStackConfiguration(
+        url="https://llama-stack.example.com:8321"
+    )  # pyright: ignore[reportCallIssue]
+    assert config is not None
+    assert str(config.url) == "https://llama-stack.example.com:8321/"
+
+
+def test_llama_stack_configuration_malformed_url_rejected() -> None:
+    """Test that malformed URLs are rejected with a ValidationError."""
+    with pytest.raises(ValidationError, match="Input should be a valid URL"):
+        LlamaStackConfiguration(
+            url="not-a-valid-url"
+        )  # pyright: ignore[reportCallIssue]
+
+
+def test_llama_stack_configuration_invalid_scheme_rejected() -> None:
+    """Test that URLs without http/https scheme are rejected."""
+    with pytest.raises(ValidationError, match="URL scheme should be 'http' or 'https'"):
+        LlamaStackConfiguration(
+            url="ftp://localhost:8321"
+        )  # pyright: ignore[reportCallIssue]

tests/unit/test_configuration.py

@@ -142,7 +142,7 @@ def test_init_from_dict() -> None:
     # check for llama_stack_configuration subsection
     assert cfg.llama_stack_configuration.api_key is not None
     assert cfg.llama_stack_configuration.api_key.get_secret_value() == "xyzzy"
-    assert cfg.llama_stack_configuration.url == "http://x.y.com:1234"
+    assert str(cfg.llama_stack_configuration.url) == "http://x.y.com:1234/"
     assert cfg.llama_stack_configuration.use_as_library_client is False
 
     # check for service_configuration subsection