test(realtime): pin the relay failure exit with a client frame the provider rejects

This commit is contained in:
mateo-berri 2026-11-02 22:03:51 -08:00
parent cedad9ddf9
commit 23515cf5b4

View file

@ -28,6 +28,8 @@ from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLogging
from litellm.litellm_core_utils.realtime_errors import close_after_upstream_handshake_refusal
from litellm.litellm_core_utils.realtime_streaming import RealTimeStreaming
from litellm.litellm_core_utils.streaming_handler import CustomStreamWrapper
from litellm.llms.base_llm.realtime.transformation import BaseRealtimeConfig
from litellm.llms.vertex_ai.audio_transcription.realtime_transformation import VertexChirpRealtimeConfig
from litellm.router_strategy.adaptive_router.hooks import AdaptiveRouterPostCallHook
from litellm.router_utils.router_callbacks.track_deployment_metrics import get_deployment_successes_for_current_minute
from litellm.types.litellm_params import RoutingStrategyName
@ -582,18 +584,19 @@ async def _relay_upstream(
litellm_logging_obj: LiteLLMLogging,
*upstream_frames: bytes | Exception,
client_ws: MagicMock | None = None,
backend_send_error: Exception | None = None,
provider_config: BaseRealtimeConfig | None = None,
upstream_stays_open: bool = False,
) -> None:
backend_ws: Final = MagicMock()
backend_ws.recv = AsyncMock(
side_effect=_frames_then_silence(*upstream_frames) if upstream_stays_open else list(upstream_frames)
)
backend_ws.send = AsyncMock(side_effect=backend_send_error)
backend_ws.send = AsyncMock()
relay: Final = RealTimeStreaming(
client_ws or _realtime_client_ws(),
backend_ws,
litellm_logging_obj,
provider_config=provider_config,
model="gpt-realtime-2.1",
logging_worker=_DiscardingLoggingWorker(),
)
@ -640,20 +643,25 @@ def _client_ws_sending(*client_frames: str) -> MagicMock:
return client_ws
_CLIENT_TEXT_ITEM: Final = json.dumps(
{
"type": "conversation.item.create",
"item": {"type": "message", "role": "user", "content": [{"type": "input_text", "text": "hello"}]},
}
)
async def _vertex_access_token() -> str:
return "token"
async def _relay_write_to_upstream_fails(litellm_logging_obj: LiteLLMLogging) -> None:
async def _client_frame_the_provider_rejects(litellm_logging_obj: LiteLLMLogging) -> None:
deferred_setup_provider: Final = VertexChirpRealtimeConfig(
resolve_access_token=_vertex_access_token, project="proj-1", location="us"
)
await _relay_upstream(
litellm_logging_obj,
_REALTIME_SESSION_CREATED,
client_ws=_client_ws_sending(_CLIENT_TEXT_ITEM),
backend_send_error=OSError("broken pipe"),
client_ws=_client_ws_sending(
json.dumps(
{
"type": "session.update",
"session": {"type": "transcription", "audio": {"input": {"format": {"type": "audio/pcmu"}}}},
}
)
),
provider_config=deferred_setup_provider,
upstream_stays_open=True,
)
@ -677,7 +685,7 @@ async def _refused_handshake(litellm_logging_obj: LiteLLMLogging) -> None:
(_refused_handshake, 0),
(_upstream_internal_error_after_a_frame, 0),
(_upstream_drop_without_close_frame_after_a_frame, 0),
(_relay_write_to_upstream_fails, 0),
(_client_frame_the_provider_rejects, 0),
(_client_hangs_up_after_a_frame, 1),
],
ids=[
@ -686,7 +694,7 @@ async def _refused_handshake(litellm_logging_obj: LiteLLMLogging) -> None:
"refused-handshake",
"upstream-1011-after-a-frame",
"upstream-1006-after-a-frame",
"relay-write-to-upstream-fails",
"client-frame-the-provider-rejects",
"client-hangs-up-after-a-frame",
],
)