phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-10-22 16:23:08 +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

151
llama_stack/providers/utils/kvstore/config.py
View file

@ -4,143 +4,20 @@
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
import re
from enum import Enum
from typing import Annotated, Literal
from typing import Annotated
from pydantic import BaseModel, Field, field_validator
from llama_stack.core.utils.config_dirs import RUNTIME_BASE_DIR
class KVStoreType(Enum):
redis = "redis"
sqlite = "sqlite"
postgres = "postgres"
mongodb = "mongodb"
class CommonConfig(BaseModel):
namespace: str | None = Field(
default=None,
description="All keys will be prefixed with this namespace",
)
class RedisKVStoreConfig(CommonConfig):
type: Literal["redis"] = KVStoreType.redis.value
host: str = "localhost"
port: int = 6379
@property
def url(self) -> str:
return f"redis://{self.host}:{self.port}"
@classmethod
def pip_packages(cls) -> list[str]:
return ["redis"]
@classmethod
def sample_run_config(cls):
return {
"type": "redis",
"host": "${env.REDIS_HOST:=localhost}",
"port": "${env.REDIS_PORT:=6379}",
}
class SqliteKVStoreConfig(CommonConfig):
type: Literal["sqlite"] = KVStoreType.sqlite.value
db_path: str = Field(
default=(RUNTIME_BASE_DIR / "kvstore.db").as_posix(),
description="File path for the sqlite database",
)
@classmethod
def pip_packages(cls) -> list[str]:
return ["aiosqlite"]
@classmethod
def sample_run_config(cls, __distro_dir__: str, db_name: str = "kvstore.db"):
return {
"type": "sqlite",
"db_path": "${env.SQLITE_STORE_DIR:=" + __distro_dir__ + "}/" + db_name,
}
class PostgresKVStoreConfig(CommonConfig):
type: Literal["postgres"] = KVStoreType.postgres.value
host: str = "localhost"
port: int = 5432
db: str = "llamastack"
user: str
password: str | None = None
ssl_mode: str | None = None
ca_cert_path: str | None = None
table_name: str = "llamastack_kvstore"
@classmethod
def sample_run_config(cls, table_name: str = "llamastack_kvstore", **kwargs):
return {
"type": "postgres",
"host": "${env.POSTGRES_HOST:=localhost}",
"port": "${env.POSTGRES_PORT:=5432}",
"db": "${env.POSTGRES_DB:=llamastack}",
"user": "${env.POSTGRES_USER:=llamastack}",
"password": "${env.POSTGRES_PASSWORD:=llamastack}",
"table_name": "${env.POSTGRES_TABLE_NAME:=" + table_name + "}",
}
@classmethod
@field_validator("table_name")
def validate_table_name(cls, v: str) -> str:
# PostgreSQL identifiers rules:
# - Must start with a letter or underscore
# - Can contain letters, numbers, and underscores
# - Maximum length is 63 bytes
pattern = r"^[a-zA-Z_][a-zA-Z0-9_]*$"
if not re.match(pattern, v):
raise ValueError(
"Invalid table name. Must start with letter or underscore and contain only letters, numbers, and underscores"
)
if len(v) > 63:
raise ValueError("Table name must be less than 63 characters")
return v
@classmethod
def pip_packages(cls) -> list[str]:
return ["psycopg2-binary"]
class MongoDBKVStoreConfig(CommonConfig):
type: Literal["mongodb"] = KVStoreType.mongodb.value
host: str = "localhost"
port: int = 27017
db: str = "llamastack"
user: str | None = None
password: str | None = None
collection_name: str = "llamastack_kvstore"
@classmethod
def pip_packages(cls) -> list[str]:
return ["pymongo"]
@classmethod
def sample_run_config(cls, collection_name: str = "llamastack_kvstore"):
return {
"type": "mongodb",
"host": "${env.MONGODB_HOST:=localhost}",
"port": "${env.MONGODB_PORT:=5432}",
"db": "${env.MONGODB_DB}",
"user": "${env.MONGODB_USER}",
"password": "${env.MONGODB_PASSWORD}",
"collection_name": "${env.MONGODB_COLLECTION_NAME:=" + collection_name + "}",
}
from pydantic import Field
from llama_stack.core.storage.datatypes import (
MongoDBKVStoreConfig,
PostgresKVStoreConfig,
RedisKVStoreConfig,
SqliteKVStoreConfig,
StorageBackendType,
)
KVStoreConfig = Annotated[
RedisKVStoreConfig | SqliteKVStoreConfig | PostgresKVStoreConfig | MongoDBKVStoreConfig,
Field(discriminator="type", default=KVStoreType.sqlite.value),
RedisKVStoreConfig | SqliteKVStoreConfig | PostgresKVStoreConfig | MongoDBKVStoreConfig, Field(discriminator="type")
]
@ -148,13 +25,13 @@ def get_pip_packages(store_config: dict | KVStoreConfig) -> list[str]:
"""Get pip packages for KV 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.KV_SQLITE.value:
return SqliteKVStoreConfig.pip_packages()
elif store_type == "postgres":
elif store_type == StorageBackendType.KV_POSTGRES.value:
return PostgresKVStoreConfig.pip_packages()
elif store_type == "redis":
elif store_type == StorageBackendType.KV_REDIS.value:
return RedisKVStoreConfig.pip_packages()
elif store_type == "mongodb":
elif store_type == StorageBackendType.KV_MONGODB.value:
return MongoDBKVStoreConfig.pip_packages()
else:
raise ValueError(f"Unknown KV store type: {store_type}")

41
llama_stack/providers/utils/kvstore/kvstore.py
View file

@ -4,9 +4,17 @@
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
# 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.
from __future__ import annotations
from llama_stack.core.storage.datatypes import KVStoreReference, StorageBackendConfig, StorageBackendType
from .api import KVStore
from .config import KVStoreConfig, KVStoreType
from .config import KVStoreConfig
def kvstore_dependencies():
@ -44,20 +52,41 @@ class InmemoryKVStoreImpl(KVStore):
del self._store[key]
async def kvstore_impl(config: KVStoreConfig) -> KVStore:
if config.type == KVStoreType.redis.value:
_KVSTORE_BACKENDS: dict[str, KVStoreConfig] = {}
def register_kvstore_backends(backends: dict[str, StorageBackendConfig]) -> None:
"""Register the set of available KV store backends for reference resolution."""
global _KVSTORE_BACKENDS
_KVSTORE_BACKENDS.clear()
for name, cfg in backends.items():
_KVSTORE_BACKENDS[name] = cfg
async def kvstore_impl(reference: KVStoreReference) -> KVStore:
backend_name = reference.backend
backend_config = _KVSTORE_BACKENDS.get(backend_name)
if backend_config is None:
raise ValueError(f"Unknown KVStore backend '{backend_name}'. Registered backends: {sorted(_KVSTORE_BACKENDS)}")
config = backend_config.model_copy()
config.namespace = reference.namespace
if config.type == StorageBackendType.KV_REDIS.value:
from .redis import RedisKVStoreImpl
impl = RedisKVStoreImpl(config)
elif config.type == KVStoreType.sqlite.value:
elif config.type == StorageBackendType.KV_SQLITE.value:
from .sqlite import SqliteKVStoreImpl
impl = SqliteKVStoreImpl(config)
elif config.type == KVStoreType.postgres.value:
elif config.type == StorageBackendType.KV_POSTGRES.value:
from .postgres import PostgresKVStoreImpl
impl = PostgresKVStoreImpl(config)
elif config.type == KVStoreType.mongodb.value:
elif config.type == StorageBackendType.KV_MONGODB.value:
from .mongodb import MongoDBKVStoreImpl
impl = MongoDBKVStoreImpl(config)