mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-06-28 02:53:30 +00:00
# What does this PR do? This commit adds a new authentication system to the Llama Stack server with support for Kubernetes and custom authentication providers. Key changes include: - Implemented KubernetesAuthProvider for validating Kubernetes service account tokens - Implemented CustomAuthProvider for validating tokens against external endpoints - this is the same code that was already present. - Added test for Kubernetes - Updated server configuration to support authentication settings - Added documentation for authentication configuration and usage The authentication system supports: - Bearer token validation - Kubernetes service account token validation - Custom authentication endpoints ## Test Plan Setup a Kube cluster using Kind or Minikube. Run a server with: ``` server: port: 8321 auth: provider_type: kubernetes config: api_server_url: http://url ca_cert_path: path/to/cert (optional) ``` Run: ``` curl -s -L -H "Authorization: Bearer $(kubectl create token my-user)" http://127.0.0.1:8321/v1/providers ``` Or replace "my-user" with your service account. Signed-off-by: Sébastien Han <seb@redhat.com>
151 lines
5.1 KiB
Python
151 lines
5.1 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 os
|
|
import shutil
|
|
import tempfile
|
|
|
|
import pytest
|
|
|
|
from llama_stack.apis.models import ModelType
|
|
from llama_stack.distribution.datatypes import ModelWithACL
|
|
from llama_stack.distribution.server.auth_providers import AccessAttributes
|
|
from llama_stack.distribution.store.registry import CachedDiskDistributionRegistry
|
|
from llama_stack.providers.utils.kvstore.config import SqliteKVStoreConfig
|
|
from llama_stack.providers.utils.kvstore.sqlite import SqliteKVStoreImpl
|
|
|
|
|
|
@pytest.fixture(scope="function")
|
|
async def kvstore():
|
|
temp_dir = tempfile.mkdtemp()
|
|
db_path = os.path.join(temp_dir, "test_registry_acl.db")
|
|
kvstore_config = SqliteKVStoreConfig(db_path=db_path)
|
|
kvstore = SqliteKVStoreImpl(kvstore_config)
|
|
await kvstore.initialize()
|
|
yield kvstore
|
|
shutil.rmtree(temp_dir)
|
|
|
|
|
|
@pytest.fixture(scope="function")
|
|
async def registry(kvstore):
|
|
registry = CachedDiskDistributionRegistry(kvstore)
|
|
await registry.initialize()
|
|
return registry
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_registry_cache_with_acl(registry):
|
|
model = ModelWithACL(
|
|
identifier="model-acl",
|
|
provider_id="test-provider",
|
|
provider_resource_id="model-acl-resource",
|
|
model_type=ModelType.llm,
|
|
access_attributes=AccessAttributes(roles=["admin"], teams=["ai-team"]),
|
|
)
|
|
|
|
success = await registry.register(model)
|
|
assert success
|
|
|
|
cached_model = registry.get_cached("model", "model-acl")
|
|
assert cached_model is not None
|
|
assert cached_model.identifier == "model-acl"
|
|
assert cached_model.access_attributes.roles == ["admin"]
|
|
assert cached_model.access_attributes.teams == ["ai-team"]
|
|
|
|
fetched_model = await registry.get("model", "model-acl")
|
|
assert fetched_model is not None
|
|
assert fetched_model.identifier == "model-acl"
|
|
assert fetched_model.access_attributes.roles == ["admin"]
|
|
|
|
model.access_attributes = AccessAttributes(roles=["admin", "user"], projects=["project-x"])
|
|
await registry.update(model)
|
|
|
|
updated_cached = registry.get_cached("model", "model-acl")
|
|
assert updated_cached is not None
|
|
assert updated_cached.access_attributes.roles == ["admin", "user"]
|
|
assert updated_cached.access_attributes.projects == ["project-x"]
|
|
assert updated_cached.access_attributes.teams is None
|
|
|
|
new_registry = CachedDiskDistributionRegistry(registry.kvstore)
|
|
await new_registry.initialize()
|
|
|
|
new_model = await new_registry.get("model", "model-acl")
|
|
assert new_model is not None
|
|
assert new_model.identifier == "model-acl"
|
|
assert new_model.access_attributes.roles == ["admin", "user"]
|
|
assert new_model.access_attributes.projects == ["project-x"]
|
|
assert new_model.access_attributes.teams is None
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_registry_empty_acl(registry):
|
|
model = ModelWithACL(
|
|
identifier="model-empty-acl",
|
|
provider_id="test-provider",
|
|
provider_resource_id="model-resource",
|
|
model_type=ModelType.llm,
|
|
access_attributes=AccessAttributes(),
|
|
)
|
|
|
|
await registry.register(model)
|
|
|
|
cached_model = registry.get_cached("model", "model-empty-acl")
|
|
assert cached_model is not None
|
|
assert cached_model.access_attributes is not None
|
|
assert cached_model.access_attributes.roles is None
|
|
assert cached_model.access_attributes.teams is None
|
|
assert cached_model.access_attributes.projects is None
|
|
assert cached_model.access_attributes.namespaces is None
|
|
|
|
all_models = await registry.get_all()
|
|
assert len(all_models) == 1
|
|
|
|
model = ModelWithACL(
|
|
identifier="model-no-acl",
|
|
provider_id="test-provider",
|
|
provider_resource_id="model-resource-2",
|
|
model_type=ModelType.llm,
|
|
)
|
|
|
|
await registry.register(model)
|
|
|
|
cached_model = registry.get_cached("model", "model-no-acl")
|
|
assert cached_model is not None
|
|
assert cached_model.access_attributes is None
|
|
|
|
all_models = await registry.get_all()
|
|
assert len(all_models) == 2
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_registry_serialization(registry):
|
|
attributes = AccessAttributes(
|
|
roles=["admin", "researcher"],
|
|
teams=["ai-team", "ml-team"],
|
|
projects=["project-a", "project-b"],
|
|
namespaces=["prod", "staging"],
|
|
)
|
|
|
|
model = ModelWithACL(
|
|
identifier="model-serialize",
|
|
provider_id="test-provider",
|
|
provider_resource_id="model-resource",
|
|
model_type=ModelType.llm,
|
|
access_attributes=attributes,
|
|
)
|
|
|
|
await registry.register(model)
|
|
|
|
new_registry = CachedDiskDistributionRegistry(registry.kvstore)
|
|
await new_registry.initialize()
|
|
|
|
loaded_model = await new_registry.get("model", "model-serialize")
|
|
assert loaded_model is not None
|
|
|
|
assert loaded_model.access_attributes.roles == ["admin", "researcher"]
|
|
assert loaded_model.access_attributes.teams == ["ai-team", "ml-team"]
|
|
assert loaded_model.access_attributes.projects == ["project-a", "project-b"]
|
|
assert loaded_model.access_attributes.namespaces == ["prod", "staging"]
|