llama-stack-mirror/llama_stack/distribution/providers.py
Charlie Doern 436f8ade9e feat: implement provider updating
add `v1/providers/` which uses PUT to allow users to change their provider configuration

this is a follow up to #1429 and related to #1359

a user can call something like:

`llama_stack_client.providers.update(api="inference", provider_id="ollama", provider_type="remote::ollama", config={'url': 'http:/localhost:12345'})`

or

`llama-stack-client providers update inference ollama remote::ollama "{'url': 'http://localhost:12345'}"`

this API works by adding a `RequestMiddleware` to the server which checks requests, and if the user is using PUT /v1/providers, the routes are re-registered with the re-initialized provider configurations/methods

for the client, `self.impls` is updated to hold the proper methods+configurations

this depends on a client PR, the CI will fail until then but succeeded locally

Signed-off-by: Charlie Doern <cdoern@redhat.com>
2025-07-01 10:04:10 -04:00

215 lines
8.2 KiB
Python

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
import asyncio
import copy
from typing import Any
from pydantic import BaseModel
from llama_stack.apis.providers import ListProvidersResponse, ProviderInfo, Providers
from llama_stack.log import get_logger
from llama_stack.providers.datatypes import HealthResponse, HealthStatus
from .datatypes import Provider, StackRunConfig
from .utils.config import redact_sensitive_fields
logger = get_logger(name=__name__, category="core")
class ProviderImplConfig(BaseModel):
run_config: StackRunConfig
async def get_provider_impl(config, deps):
impl = ProviderImpl(config, deps)
await impl.initialize()
return impl
class ProviderImpl(Providers):
def __init__(self, config, deps):
self.config = config
self.deps = deps
async def initialize(self) -> None:
pass
async def shutdown(self) -> None:
logger.debug("ProviderImpl.shutdown")
pass
async def list_providers(self) -> ListProvidersResponse:
run_config = self.config.run_config
safe_config = StackRunConfig(**redact_sensitive_fields(run_config.model_dump()))
providers_health = await self.get_providers_health()
ret = []
for api, providers in safe_config.providers.items():
for p in providers:
# Skip providers that are not enabled
if p.provider_id is None:
continue
ret.append(
ProviderInfo(
api=api,
provider_id=p.provider_id,
provider_type=p.provider_type,
config=p.config,
health=providers_health.get(api, {}).get(
p.provider_id,
HealthResponse(
status=HealthStatus.NOT_IMPLEMENTED, message="Provider does not implement health check"
),
),
)
)
return ListProvidersResponse(data=ret)
async def inspect_provider(self, provider_id: str) -> ProviderInfo:
all_providers = await self.list_providers()
for p in all_providers.data:
if p.provider_id == provider_id:
return p
raise ValueError(f"Provider {provider_id} not found")
async def get_providers_health(self) -> dict[str, dict[str, HealthResponse]]:
"""Get health status for all providers.
Returns:
Dict[str, Dict[str, HealthResponse]]: A dictionary mapping API names to provider health statuses.
Each API maps to a dictionary of provider IDs to their health responses.
"""
providers_health: dict[str, dict[str, HealthResponse]] = {}
timeout = 1.0
async def check_provider_health(impl: Any) -> tuple[str, HealthResponse] | None:
# Skip special implementations (inspect/providers) that don't have provider specs
if not hasattr(impl, "__provider_spec__"):
return None
api_name = impl.__provider_spec__.api.name
if not hasattr(impl, "health"):
return (
api_name,
HealthResponse(
status=HealthStatus.NOT_IMPLEMENTED, message="Provider does not implement health check"
),
)
try:
health = await asyncio.wait_for(impl.health(), timeout=timeout)
return api_name, health
except TimeoutError:
return (
api_name,
HealthResponse(
status=HealthStatus.ERROR, message=f"Health check timed out after {timeout} seconds"
),
)
except Exception as e:
return (
api_name,
HealthResponse(status=HealthStatus.ERROR, message=f"Health check failed: {str(e)}"),
)
# Create tasks for all providers
tasks = [check_provider_health(impl) for impl in self.deps.values()]
# Wait for all health checks to complete
results = await asyncio.gather(*tasks)
# Organize results by API and provider ID
for result in results:
if result is None: # Skip special implementations
continue
api_name, health_response = result
providers_health[api_name] = health_response
return providers_health
async def update_provider(
self, api: str, provider_id: str, provider_type: str, config: dict[str, Any]
) -> ProviderInfo:
# config = ast.literal_eval(provider_request.config)
prov = Provider(
provider_id=provider_id,
provider_type=provider_type,
config=config,
)
assert prov.provider_id is not None
existing_provider = None
# if the provider isn't there or the API is invalid, we should not continue
for prov_api, providers in self.config.run_config.providers.items():
if prov_api != api:
continue
for p in providers:
if p.provider_id == provider_id:
existing_provider = p
break
if existing_provider is not None:
break
if existing_provider is None:
raise ValueError(f"Provider {provider_id} not found, you can only update already registered providers.")
new_config = self.merge_providers(existing_provider, prov)
existing_provider.config = new_config
providers_health = await self.get_providers_health()
# takes a single provider, validates its in the registry
# if it is, merge the provider config with the existing one
ret = ProviderInfo(
api=api,
provider_id=prov.provider_id,
provider_type=prov.provider_type,
config=new_config,
health=providers_health.get(api, {}).get(
p.provider_id,
HealthResponse(status=HealthStatus.NOT_IMPLEMENTED, message="Provider does not implement health check"),
),
)
return ret
def merge_dicts(self, base: dict[str, Any], overrides: dict[str, Any]) -> dict[str, Any]:
"""Recursively merges `overrides` into `base`, replacing only specified keys."""
merged = copy.deepcopy(base) # Preserve original dict
for key, value in overrides.items():
if isinstance(value, dict) and isinstance(merged.get(key), dict):
# Recursively merge if both are dictionaries
merged[key] = self.merge_dicts(merged[key], value)
else:
# Otherwise, directly override
merged[key] = value
return merged
def merge_configs(
self, global_config: dict[str, list[Provider]], new_config: dict[str, list[Provider]]
) -> dict[str, list[Provider]]:
merged_config = copy.deepcopy(global_config) # Preserve original structure
for key, new_providers in new_config.items():
if key in merged_config:
existing_providers = {p.provider_id: p for p in merged_config[key]}
for new_provider in new_providers:
if new_provider.provider_id in existing_providers:
# Override settings of existing provider
existing = existing_providers[new_provider.provider_id]
existing.config = self.merge_dicts(existing.config, new_provider.config)
else:
# Append new provider
merged_config[key].append(new_provider)
else:
# Add new category entirely
merged_config[key] = new_providers
return merged_config
def merge_providers(self, current_provider: Provider, new_provider: Provider) -> dict[str, Any]:
return self.merge_dicts(current_provider.config, new_provider.config)