phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-23 16:37:28 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

feat(stores)!: use backend storage references instead of configs (#3697)

Browse source 
**This PR changes configurations in a backward incompatible way.**

Run configs today repeat full SQLite/Postgres snippets everywhere a
store is needed, which means duplicated credentials, extra connection
pools, and lots of drift between files. This PR introduces named storage
backends so the stack and providers can share a single catalog and
reference those backends by name.

## Key Changes

- Add `storage.backends` to `StackRunConfig`, register each KV/SQL
backend once at startup, and validate that references point to the right
family.
- Move server stores under `storage.stores` with lightweight references
(backend + namespace/table) instead of full configs.
- Update every provider/config/doc to use the new reference style;
docs/codegen now surface the simplified YAML.

## Migration

Before:
```yaml
metadata_store:
  type: sqlite
  db_path: ~/.llama/distributions/foo/registry.db
inference_store:
  type: postgres
  host: ${env.POSTGRES_HOST}
  port: ${env.POSTGRES_PORT}
  db: ${env.POSTGRES_DB}
  user: ${env.POSTGRES_USER}
  password: ${env.POSTGRES_PASSWORD}
conversations_store:
  type: postgres
  host: ${env.POSTGRES_HOST}
  port: ${env.POSTGRES_PORT}
  db: ${env.POSTGRES_DB}
  user: ${env.POSTGRES_USER}
  password: ${env.POSTGRES_PASSWORD}
```

After:
```yaml
storage:
  backends:
    kv_default:
      type: kv_sqlite
      db_path: ~/.llama/distributions/foo/kvstore.db
    sql_default:
      type: sql_postgres
      host: ${env.POSTGRES_HOST}
      port: ${env.POSTGRES_PORT}
      db: ${env.POSTGRES_DB}
      user: ${env.POSTGRES_USER}
      password: ${env.POSTGRES_PASSWORD}
  stores:
    metadata:
      backend: kv_default
      namespace: registry
    inference:
      backend: sql_default
      table_name: inference_store
      max_write_queue_size: 10000
      num_writers: 4
    conversations:
      backend: sql_default
      table_name: openai_conversations
```

Provider configs follow the same pattern—for example, a Chroma vector
adapter switches from:

```yaml
providers:
  vector_io:
  - provider_id: chromadb
    provider_type: remote::chromadb
    config:
      url: ${env.CHROMADB_URL}
      kvstore:
        type: sqlite
        db_path: ~/.llama/distributions/foo/chroma.db
```

to:

```yaml
providers:
  vector_io:
  - provider_id: chromadb
    provider_type: remote::chromadb
    config:
      url: ${env.CHROMADB_URL}
      persistence:
        backend: kv_default
        namespace: vector_io::chroma_remote
```

Once the backends are declared, everything else just points at them, so
rotating credentials or swapping to Postgres happens in one place and
the stack reuses a single connection pool.
This commit is contained in:
Ashwin Bharambe 2025-10-20 13:20:09 -07:00 • committed by GitHub
parent add64e8e2a
commit 2c43285e22
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
105 changed files with 2290 additions and 1292 deletions
Download patch file Download diff file Expand all files Collapse all files

114
llama_stack/providers/utils/sqlstore/sqlstore.py
View file

@ -4,90 +4,28 @@
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from abc import abstractmethod
from enum import StrEnum
from pathlib import Path
from typing import Annotated, Literal
from typing import Annotated, cast
from pydantic import BaseModel, Field
from pydantic import Field
from llama_stack.core.utils.config_dirs import RUNTIME_BASE_DIR
from llama_stack.core.storage.datatypes import (
PostgresSqlStoreConfig,
SqliteSqlStoreConfig,
SqlStoreReference,
StorageBackendConfig,
StorageBackendType,
)
from .api import SqlStore
sql_store_pip_packages = ["sqlalchemy[asyncio]", "aiosqlite", "asyncpg"]
class SqlStoreType(StrEnum):
sqlite = "sqlite"
postgres = "postgres"
class SqlAlchemySqlStoreConfig(BaseModel):
@property
@abstractmethod
def engine_str(self) -> str: ...
# TODO: move this when we have a better way to specify dependencies with internal APIs
@classmethod
def pip_packages(cls) -> list[str]:
return ["sqlalchemy[asyncio]"]
class SqliteSqlStoreConfig(SqlAlchemySqlStoreConfig):
type: Literal[SqlStoreType.sqlite] = SqlStoreType.sqlite
db_path: str = Field(
default=(RUNTIME_BASE_DIR / "sqlstore.db").as_posix(),
description="Database path, e.g. ~/.llama/distributions/ollama/sqlstore.db",
)
@property
def engine_str(self) -> str:
return "sqlite+aiosqlite:///" + Path(self.db_path).expanduser().as_posix()
@classmethod
def sample_run_config(cls, __distro_dir__: str, db_name: str = "sqlstore.db"):
return {
"type": "sqlite",
"db_path": "${env.SQLITE_STORE_DIR:=" + __distro_dir__ + "}/" + db_name,
}
@classmethod
def pip_packages(cls) -> list[str]:
return super().pip_packages() + ["aiosqlite"]
class PostgresSqlStoreConfig(SqlAlchemySqlStoreConfig):
type: Literal[SqlStoreType.postgres] = SqlStoreType.postgres
host: str = "localhost"
port: int = 5432
db: str = "llamastack"
user: str
password: str | None = None
@property
def engine_str(self) -> str:
return f"postgresql+asyncpg://{self.user}:{self.password}@{self.host}:{self.port}/{self.db}"
@classmethod
def pip_packages(cls) -> list[str]:
return super().pip_packages() + ["asyncpg"]
@classmethod
def sample_run_config(cls, **kwargs):
return {
"type": "postgres",
"host": "${env.POSTGRES_HOST:=localhost}",
"port": "${env.POSTGRES_PORT:=5432}",
"db": "${env.POSTGRES_DB:=llamastack}",
"user": "${env.POSTGRES_USER:=llamastack}",
"password": "${env.POSTGRES_PASSWORD:=llamastack}",
}
_SQLSTORE_BACKENDS: dict[str, StorageBackendConfig] = {}
SqlStoreConfig = Annotated[
SqliteSqlStoreConfig | PostgresSqlStoreConfig,
Field(discriminator="type", default=SqlStoreType.sqlite.value),
Field(discriminator="type"),
]
@ -95,9 +33,9 @@ def get_pip_packages(store_config: dict | SqlStoreConfig) -> list[str]:
"""Get pip packages for SQL store config, handling both dict and object cases."""
if isinstance(store_config, dict):
store_type = store_config.get("type")
if store_type == "sqlite":
if store_type == StorageBackendType.SQL_SQLITE.value:
return SqliteSqlStoreConfig.pip_packages()
elif store_type == "postgres":
elif store_type == StorageBackendType.SQL_POSTGRES.value:
return PostgresSqlStoreConfig.pip_packages()
else:
raise ValueError(f"Unknown SQL store type: {store_type}")
@ -105,12 +43,28 @@ def get_pip_packages(store_config: dict | SqlStoreConfig) -> list[str]:
return store_config.pip_packages()
def sqlstore_impl(config: SqlStoreConfig) -> SqlStore:
if config.type in [SqlStoreType.sqlite, SqlStoreType.postgres]:
def sqlstore_impl(reference: SqlStoreReference) -> SqlStore:
backend_name = reference.backend
backend_config = _SQLSTORE_BACKENDS.get(backend_name)
if backend_config is None:
raise ValueError(
f"Unknown SQL store backend '{backend_name}'. Registered backends: {sorted(_SQLSTORE_BACKENDS)}"
)
if isinstance(backend_config, SqliteSqlStoreConfig | PostgresSqlStoreConfig):
from .sqlalchemy_sqlstore import SqlAlchemySqlStoreImpl
impl = SqlAlchemySqlStoreImpl(config)
config = cast(SqliteSqlStoreConfig | PostgresSqlStoreConfig, backend_config).model_copy()
return SqlAlchemySqlStoreImpl(config)
else:
raise ValueError(f"Unknown sqlstore type {config.type}")
raise ValueError(f"Unknown sqlstore type {backend_config.type}")
return impl
def register_sqlstore_backends(backends: dict[str, StorageBackendConfig]) -> None:
"""Register the set of available SQL store backends for reference resolution."""
global _SQLSTORE_BACKENDS
_SQLSTORE_BACKENDS.clear()
for name, cfg in backends.items():
_SQLSTORE_BACKENDS[name] = cfg