Files
LocalAI/backend/python/common/model_identity.py
walcz-de 6d1bbb74c4 fix(backend/python): don't await sync servicer behaviors in AsyncModelIdentityInterceptor (#10980)
* fix(backend/python): don't await sync servicer behaviors in AsyncModelIdentityInterceptor

The model-identity interceptor (added for #10952) is installed on every Python
backend's gRPC server. Its grpc.aio variant invokes the wrapped servicer
behavior itself and awaits the result unconditionally:

    result = await original(request, context)                 # LoadModel
    return await original_unary(request, context)             # guarded RPCs
    async for response in original_stream(request, context):  # streaming

But a backend's servicer methods may be plain sync functions. The transformers
backend, for one, defines `def LoadModel` and `def Embedding` (not `async def`).
grpc.aio's own dispatch adapts both shapes, but this interceptor calls the
behavior directly and bypasses that. For a sync method `original(...)` returns a
message object, not a coroutine, so the `await` raises:

    TypeError: object Result can't be used in 'await' expression

The model loads, then the LoadModel RPC dies on return; the guarded sync
Embedding fails the same way. It happens on every platform, not just one backend
build. CI never caught it because AsyncModelIdentityInterceptor had no
behavioral test -- only an "is it installed" assertion.

Fix: await only when the behavior actually returned an awaitable
(inspect.isawaitable), mirroring grpc.aio's own sync/async adaptation. The
streaming guard iterates a sync generator with `for` and an async one with
`async for`.

Adds async-path coverage to model_identity_test.py exercising both sync and
async LoadModel / guarded-unary / streaming behaviors. The sync cases fail on
the current code with the TypeError above and pass with this fix.

Signed-off-by: stefanwalcz <stefan.walcz@walcz.de>

* fix(backend/python): dispatch sync servicer behaviors off the event loop

Addresses review feedback: awaiting only awaitable results removed the
TypeError, but still ran a sync LoadModel/Embedding -- and stepped a sync stream
via next() -- on the asyncio event-loop thread, so a slow load/inference/stream
could freeze all aio RPC handling.

Route sync behavior through run_in_executor (a worker thread) while awaiting
native async behavior directly. A callable wrapper that returns an awaitable is
run in the thread and its awaitable awaited back on the loop. Sync streaming
pulls each item via the executor with a done sentinel, so StopIteration cannot
escape through a Future.

Adds regression tests that record the handler thread id and assert it differs
from the event-loop thread, for LoadModel, a guarded unary RPC and a sync stream.

Signed-off-by: stefanwalcz <stefan.walcz@walcz.de>

---------

Signed-off-by: stefanwalcz <stefan.walcz@walcz.de>
2026-07-22 16:03:46 +02:00

284 lines
11 KiB
Python

"""Model-identity enforcement for LocalAI Python backends.
In distributed mode the controller caches a routing row naming a backend's
host:port. A worker can recycle a stopped backend's gRPC port for a different
model's backend, and the controller's health probe checks liveness rather than
identity, so the request is dispatched to whatever now occupies the port and
the caller gets a silent wrong-model answer (#10952).
Every request message that reaches a backend through the distributed router
carries a ModelIdentity field naming the model the request is for, so the
backend can reject it at the point of use. This module enforces that for every
Python backend at once: all 36 of them build their server through
grpc_auth.get_auth_interceptors(), so wiring it there needs no per-backend
change. There is no shared BackendServicer base class to hook instead, and the
backends store their loaded model in wildly different attributes, so an
interceptor is the only single point that sees both the LoadModel request and
the inference requests.
Enforcement is deliberately narrow: it compares two strings and never inspects
the model itself.
"""
import asyncio
import inspect
import threading
import grpc
# Must match grpcerrors.ModelMismatchSentinel in pkg/grpc/grpcerrors/errors.go.
# The router requires this substring AND the NOT_FOUND code before it treats a
# reply as a mismatch, because NOT_FOUND alone is not exclusively ours on these
# RPCs (insightface's Embedding returns it for "no face detected").
MODEL_MISMATCH_SENTINEL = "model identity mismatch"
_LOAD_METHOD = "/backend.Backend/LoadModel"
# Every RPC whose request message carries a ModelIdentity field. This set IS
# the enforcement surface for all 36 Python backends: an RPC missing here is
# silently unprotected, so model_identity_test.py pins the full list.
#
# The guard reads request.ModelIdentity generically, so nothing here is
# modality-specific — a backend that does not implement an RPC simply never
# sees it.
#
# TTS and SoundGeneration are guarded on ModelIdentity, NOT on their `model`
# field: the controller's FileStagingClient rewrites `model` to a worker-local
# absolute path, so comparing that would reject valid requests in distributed
# mode. ModelIdentity is a separate, untranslated field for exactly that reason.
#
# AudioEncode/AudioDecode are absent deliberately: the opus codec backend they
# target is loaded from a literal rather than a ModelConfig, so no value carries
# the load-time/request-time equality guarantee this comparison depends on.
_GUARDED_METHODS = frozenset(
(
# PredictOptions RPCs (#10970)
"/backend.Backend/Predict",
"/backend.Backend/PredictStream",
"/backend.Backend/Embedding",
"/backend.Backend/TokenizeString",
# Remaining modalities
"/backend.Backend/GenerateImage",
"/backend.Backend/GenerateVideo",
"/backend.Backend/TTS",
"/backend.Backend/TTSStream",
"/backend.Backend/SoundGeneration",
"/backend.Backend/AudioTranscription",
"/backend.Backend/AudioTranscriptionStream",
"/backend.Backend/Detect",
"/backend.Backend/Depth",
"/backend.Backend/FaceVerify",
"/backend.Backend/FaceAnalyze",
"/backend.Backend/VoiceVerify",
"/backend.Backend/VoiceAnalyze",
"/backend.Backend/VoiceEmbed",
"/backend.Backend/Rerank",
"/backend.Backend/TokenClassify",
"/backend.Backend/Score",
"/backend.Backend/VAD",
"/backend.Backend/Diarize",
"/backend.Backend/SoundDetection",
"/backend.Backend/AudioTransform",
)
)
class ModelIdentityState:
"""The identity this process loaded, and the rule for judging a request.
A backend process serves exactly one model (worker process keys are
model+backend+replica), so a single value is enough.
"""
def __init__(self):
self._lock = threading.Lock()
self._loaded = ""
def record(self, model: str) -> None:
with self._lock:
self._loaded = model or ""
@property
def loaded(self) -> str:
with self._lock:
return self._loaded
def mismatch(self, requested: str):
"""Return an error message when `requested` names another model.
Either side being empty means "skip": the request side is empty for a
controller that predates the field and for internally synthesized
requests, and the loaded side is empty when such a controller performed
the load. Neither can judge the other, and a false rejection is worse
than the miss it prevents.
"""
if not requested:
return None
loaded = self.loaded
if not loaded or loaded == requested:
return None
return "{}: loaded {!r}, requested {!r}".format(
MODEL_MISMATCH_SENTINEL, loaded, requested
)
def _rebuild(handler, behavior):
"""Return a copy of `handler` with its behavior replaced.
Only unary-request handlers are ever passed here: LoadModel and every
entry in _GUARDED_METHODS take a single request message. The bidirectional
streams (AudioTranscriptionLive, AudioTransformStream, AudioToAudioStream,
Forward) are not guarded and never reach this function.
"""
if handler.response_streaming:
return grpc.unary_stream_rpc_method_handler(
behavior,
request_deserializer=handler.request_deserializer,
response_serializer=handler.response_serializer,
)
return grpc.unary_unary_rpc_method_handler(
behavior,
request_deserializer=handler.request_deserializer,
response_serializer=handler.response_serializer,
)
class ModelIdentityInterceptor(grpc.ServerInterceptor):
"""Sync interceptor that records the loaded model and guards inference."""
def __init__(self, state: ModelIdentityState = None):
self.state = state or ModelIdentityState()
def intercept_service(self, continuation, handler_call_details):
method = handler_call_details.method
if method != _LOAD_METHOD and method not in _GUARDED_METHODS:
return continuation(handler_call_details)
handler = continuation(handler_call_details)
if handler is None:
return handler
if method == _LOAD_METHOD:
original = handler.unary_unary
def record(request, context):
result = original(request, context)
# Only a successful load owns the identity; a failed one leaves
# no model, which the model-not-loaded signal already covers.
if getattr(result, "success", True):
self.state.record(getattr(request, "Model", ""))
return result
return _rebuild(handler, record)
original = handler.unary_stream if handler.response_streaming else handler.unary_unary
def guard(request, context):
message = self.state.mismatch(getattr(request, "ModelIdentity", ""))
if message is not None:
# abort() raises, so the request never reaches the model.
context.abort(grpc.StatusCode.NOT_FOUND, message)
return original(request, context)
return _rebuild(handler, guard)
_STREAM_DONE = object()
def _next_or_done(iterator):
"""next(iterator), returning the _STREAM_DONE sentinel at exhaustion.
StopIteration must not propagate out of a function run via run_in_executor:
it cannot travel through a Future and would surface as an opaque error.
"""
try:
return next(iterator)
except StopIteration:
return _STREAM_DONE
async def _call_behavior(behavior, request, context):
"""Invoke a unary servicer behavior without blocking the event loop.
Native async behavior is awaited directly. A sync behavior -- many backends
define `def LoadModel` / `def Embedding`, not `async def` -- is dispatched to
a worker thread so a slow load/inference cannot freeze all aio RPC handling,
mirroring grpc.aio's own sync-handler adaptation. A callable wrapper that
returns an awaitable is supported too: the (cheap) call runs in the thread,
then the awaitable is awaited back on the loop.
"""
if inspect.iscoroutinefunction(behavior):
return await behavior(request, context)
loop = asyncio.get_running_loop()
result = await loop.run_in_executor(None, behavior, request, context)
if inspect.isawaitable(result):
result = await result
return result
class AsyncModelIdentityInterceptor(grpc.aio.ServerInterceptor):
"""Async counterpart for backends running grpc.aio servers."""
def __init__(self, state: ModelIdentityState = None):
self.state = state or ModelIdentityState()
async def intercept_service(self, continuation, handler_call_details):
method = handler_call_details.method
if method != _LOAD_METHOD and method not in _GUARDED_METHODS:
return await continuation(handler_call_details)
handler = await continuation(handler_call_details)
if handler is None:
return handler
if method == _LOAD_METHOD:
original = handler.unary_unary
async def record(request, context):
# A backend's LoadModel may be a plain sync method (many define
# `def LoadModel`, not `async def`). Dispatch it so it neither
# crashes with "object <T> can't be used in 'await'" nor runs its
# (potentially slow) body on the event loop thread.
result = await _call_behavior(original, request, context)
if getattr(result, "success", True):
self.state.record(getattr(request, "Model", ""))
return result
return _rebuild(handler, record)
if handler.response_streaming:
original_stream = handler.unary_stream
async def guard_stream(request, context):
message = self.state.mismatch(getattr(request, "ModelIdentity", ""))
if message is not None:
await context.abort(grpc.StatusCode.NOT_FOUND, message)
# A sync backend yields a plain generator, an async one an async
# generator. Async: iterate directly. Sync: pull each item via a
# worker thread so a slow producer doesn't block the event loop
# (and so StopIteration can't escape through a Future).
stream = original_stream(request, context)
if hasattr(stream, "__aiter__"):
async for response in stream:
yield response
else:
loop = asyncio.get_running_loop()
while True:
item = await loop.run_in_executor(None, _next_or_done, stream)
if item is _STREAM_DONE:
break
yield item
return _rebuild(handler, guard_stream)
original_unary = handler.unary_unary
async def guard(request, context):
message = self.state.mismatch(getattr(request, "ModelIdentity", ""))
if message is not None:
await context.abort(grpc.StatusCode.NOT_FOUND, message)
return await _call_behavior(original_unary, request, context)
return _rebuild(handler, guard)