feat: introduce "enabled" field for providers

Closes: #2622
Signed-off-by: Sébastien Han <seb@redhat.com>

llama_stack/distribution/datatypes.py

@@ -146,11 +146,10 @@ in the runtime configuration to help route to the correct provider.""",
 
 
 class Provider(BaseModel):
-    # provider_id of None means that the provider is not enabled - this happens
-    # when the provider is enabled via a conditional environment variable
-    provider_id: str | None
+    provider_id: str
     provider_type: str
     config: dict[str, Any]
+    enabled: bool = Field(default=True, description="Whether the provider is enabled")
 
 
 class LoggingConfig(BaseModel):

llama_stack/distribution/resolver.py

@@ -199,7 +199,7 @@ def validate_and_prepare_providers(
 
         specs = {}
         for provider in providers:
-            if not provider.provider_id or provider.provider_id == "__disabled__":
+            if not provider.provider_id or not provider.enabled:
                 logger.warning(f"Provider `{provider.provider_type}` for API `{api}` is disabled")
                 continue
 

llama_stack/distribution/stack.py

@@ -99,19 +99,10 @@ async def register_resources(run_config: StackRunConfig, impls: dict[Api, Any]):
         method = getattr(impls[api], register_method)
         for obj in objects:
             # Do not register models on disabled providers
-            if hasattr(obj, "provider_id") and obj.provider_id is not None and obj.provider_id == "__disabled__":
-                logger.debug(f"Skipping {rsrc.capitalize()} registration for disabled provider.")
-                continue
-            # In complex templates, like our starter template, we may have dynamic model ids
-            # given by environment variables. This allows those environment variables to have
-            # a default value of __disabled__ to skip registration of the model if not set.
-            if (
-                hasattr(obj, "provider_model_id")
-                and obj.provider_model_id is not None
-                and "__disabled__" in obj.provider_model_id
-            ):
+            if hasattr(obj, "provider_model_id") and obj.provider_model_id is not None and not obj.enabled:
                 logger.debug(f"Skipping {rsrc.capitalize()} registration for disabled model.")
                 continue
+
             # we want to maintain the type information in arguments to method.
             # instead of method(**obj.model_dump()), which may convert a typed attr to a dict,
             # we use model_dump() to find all the attrs and then getattr to get the still typed value.
@@ -155,17 +146,20 @@ def replace_env_vars(config: Any, path: str = "") -> Any:
         for i, v in enumerate(config):
             try:
                 # Special handling for providers: first resolve the provider_id to check if provider
-                # is disabled so that we can skip config env variable expansion and avoid validation errors
+                # is disabled so that we can skip config env variable expansion and avoid validation
+                # errors
                 if isinstance(v, dict) and "provider_id" in v:
                     try:
-                        resolved_provider_id = replace_env_vars(v["provider_id"], f"{path}[{i}].provider_id")
-                        if resolved_provider_id == "__disabled__":
+                        # We have to set a default to True because we use Pydantic
+                        # exclude_defaults=True from the serializer so the loaded config only has
+                        # 'enabled' field when it is set to False explicitly.
+                        if not v.get("enabled", True):
                             logger.debug(
-                                f"Skipping config env variable expansion for disabled provider: {v.get('provider_id', '')}"
+                                f"Skipping config env variable expansion for disabled provider: {v.get('provider_type', '') + '/' if v.get('provider_type', '') else ''}{v.get('provider_id', '')}"
                             )
                             # Create a copy with resolved provider_id but original config
                             disabled_provider = v.copy()
-                            disabled_provider["provider_id"] = resolved_provider_id
+                            disabled_provider["provider_id"] = v["provider_id"]
                             result.append(disabled_provider)
                             continue
                     except EnvVarError:

llama_stack/providers/inline/post_training/huggingface/config.py

@@ -68,5 +68,5 @@ class HuggingFacePostTrainingConfig(BaseModel):
     dataloader_pin_memory: bool = True
 
     @classmethod
-    def sample_run_config(cls, __distro_dir__: str, **kwargs: Any) -> dict[str, Any]:
+    def sample_run_config(cls, **kwargs: Any) -> dict[str, Any]:
         return {"checkpoint_format": "huggingface", "distributed_backend": None, "device": "cpu"}

llama_stack/providers/utils/inference/model_registry.py

@@ -25,6 +25,7 @@ class ProviderModelEntry(BaseModel):
     llama_model: str | None = None
     model_type: ModelType = ModelType.llm
     metadata: dict[str, Any] = Field(default_factory=dict)
+    enabled: bool = Field(default=True, description="Whether the model is enabled")
 
 
 def get_huggingface_repo(model_descriptor: str) -> str | None:

llama_stack/templates/starter/starter.py

@@ -78,22 +78,25 @@ def _get_model_entries_for_provider(provider_type: str) -> list[ProviderModelEnt
     if provider_type == "ollama":
         return [
             ProviderModelEntry(
-                provider_model_id="${env.OLLAMA_INFERENCE_MODEL:=__disabled__}",
+                provider_model_id="${env.OLLAMA_INFERENCE_MODEL:=}",
                 model_type=ModelType.llm,
+                enabled=False,
             ),
             ProviderModelEntry(
-                provider_model_id="${env.OLLAMA_EMBEDDING_MODEL:=__disabled__}",
+                provider_model_id="${env.OLLAMA_EMBEDDING_MODEL:=}",
                 model_type=ModelType.embedding,
                 metadata={
                     "embedding_dimension": "${env.OLLAMA_EMBEDDING_DIMENSION:=384}",
                 },
+                enabled=False,
             ),
         ]
     elif provider_type == "vllm":
         return [
             ProviderModelEntry(
-                provider_model_id="${env.VLLM_INFERENCE_MODEL:=__disabled__}",
+                provider_model_id="${env.VLLM_INFERENCE_MODEL:=}",
                 model_type=ModelType.llm,
+                enabled=False,
             ),
         ]
 
@@ -129,29 +132,29 @@ def get_remote_inference_providers() -> tuple[list[Provider], dict[str, list[Pro
         provider_type = provider_spec.adapter.adapter_type
 
         # Build the environment variable name for enabling this provider
-        env_var = f"ENABLE_{provider_type.upper().replace('-', '_').replace('::', '_')}"
+        # only get the provider type after the ::
         model_entries = _get_model_entries_for_provider(provider_type)
         config = _get_config_for_provider(provider_spec)
         providers.append(
             (
-                f"${{env.{env_var}:=__disabled__}}",
                 provider_type,
                 model_entries,
                 config,
             )
         )
-        available_models[f"${{env.{env_var}:=__disabled__}}"] = model_entries
+        available_models[provider_type] = model_entries
 
     inference_providers = []
-    for provider_id, provider_type, model_entries, config in providers:
+    for provider_type, model_entries, config in providers:
         inference_providers.append(
             Provider(
-                provider_id=provider_id,
+                provider_id=provider_type,
                 provider_type=f"remote::{provider_type}",
                 config=config,
+                enabled=False,
             )
         )
-        available_models[provider_id] = model_entries
+        available_models[provider_type] = model_entries
     return inference_providers, available_models
 
 
@@ -162,33 +165,33 @@ def get_distribution_template() -> DistributionTemplate:
 
     vector_io_providers = [
         Provider(
-            provider_id="${env.ENABLE_FAISS:=faiss}",
+            provider_id="faiss",
             provider_type="inline::faiss",
             config=FaissVectorIOConfig.sample_run_config(f"~/.llama/distributions/{name}"),
         ),
         Provider(
-            provider_id="${env.ENABLE_SQLITE_VEC:=__disabled__}",
+            provider_id="sqlite-vec",
             provider_type="inline::sqlite-vec",
             config=SQLiteVectorIOConfig.sample_run_config(f"~/.llama/distributions/{name}"),
+            enabled=False,
         ),
         Provider(
-            provider_id="${env.ENABLE_MILVUS:=__disabled__}",
+            provider_id="milvus",
             provider_type="inline::milvus",
             config=MilvusVectorIOConfig.sample_run_config(f"~/.llama/distributions/{name}"),
+            enabled=False,
         ),
         Provider(
-            provider_id="${env.ENABLE_CHROMADB:=__disabled__}",
+            provider_id="chromadb",
             provider_type="remote::chromadb",
-            config=ChromaVectorIOConfig.sample_run_config(url="${env.CHROMADB_URL:=}"),
+            config=ChromaVectorIOConfig.sample_run_config(),
+            enabled=False,
         ),
         Provider(
-            provider_id="${env.ENABLE_PGVECTOR:=__disabled__}",
+            provider_id="pgvector",
             provider_type="remote::pgvector",
-            config=PGVectorVectorIOConfig.sample_run_config(
-                db="${env.PGVECTOR_DB:=}",
-                user="${env.PGVECTOR_USER:=}",
-                password="${env.PGVECTOR_PASSWORD:=}",
-            ),
+            config=PGVectorVectorIOConfig.sample_run_config(),
+            enabled=False,
         ),
     ]
 
@@ -216,14 +219,14 @@ def get_distribution_template() -> DistributionTemplate:
         config=LocalfsFilesImplConfig.sample_run_config(f"~/.llama/distributions/{name}"),
     )
     embedding_provider = Provider(
-        provider_id="${env.ENABLE_SENTENCE_TRANSFORMERS:=sentence-transformers}",
+        provider_id="sentence-transformers",
         provider_type="inline::sentence-transformers",
         config=SentenceTransformersInferenceConfig.sample_run_config(),
     )
     post_training_provider = Provider(
         provider_id="huggingface",
         provider_type="inline::huggingface",
-        config=HuggingFacePostTrainingConfig.sample_run_config(f"~/.llama/distributions/{name}"),
+        config=HuggingFacePostTrainingConfig.sample_run_config(),
     )
     default_tool_groups = [
         ToolGroupInput(

llama_stack/templates/template.py

@@ -105,7 +105,8 @@ class RunConfigSettings(BaseModel):
             if api_providers := self.provider_overrides.get(api_str):
                 # Convert Provider objects to dicts for YAML serialization
                 provider_configs[api_str] = [
-                    p.model_dump(exclude_none=True) if isinstance(p, Provider) else p for p in api_providers
+                    p.model_dump(exclude_defaults=True, exclude_none=True) if isinstance(p, Provider) else p
+                    for p in api_providers
                 ]
                 continue
 
@@ -133,7 +134,7 @@ class RunConfigSettings(BaseModel):
                         provider_id=provider_id,
                         provider_type=provider_type,
                         config=config,
-                    ).model_dump(exclude_none=True)
+                    ).model_dump(exclude_defaults=True, exclude_none=True)
                 )
 
         # Get unique set of APIs from providers