[Test] Initialize tensor-parallel groups in native SANA accuracy fixtures

Signed-off-by: tzhouam <tzhouam@connect.ust.hk>

tests/e2e/accuracy/sana_video/conftest.py

@@ -0,0 +1,26 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
+
+import pytest
+
+
+@pytest.fixture
+def sana_model_parallel(tmp_path):
+    """Construct native parallel layers with real single-rank GPU groups."""
+    from vllm_omni.diffusion.distributed.parallel_state import (
+        destroy_distributed_env,
+        init_distributed_environment,
+        initialize_model_parallel,
+    )
+
+    init_distributed_environment(
+        world_size=1,
+        rank=0,
+        local_rank=0,
+        distributed_init_method=f"file://{tmp_path / 'sana-distributed'}",
+    )
+    try:
+        initialize_model_parallel()
+        yield
+    finally:
+        destroy_distributed_env()

tests/e2e/accuracy/sana_video/test_sana_video_golden.py

@@ -202,6 +202,7 @@ def test_sana_video_transformer_matches_frozen_golden(
     variant: str,
     _hardware,
     accuracy_artifact_root: Path,
+    sana_model_parallel,
 ) -> None:
     del _hardware
     from vllm.config import LoadConfig

tests/e2e/accuracy/sana_video/test_sana_video_pipeline_alignment.py

@@ -189,6 +189,7 @@ def _cfg_prediction(
 def test_sana_video_quick_pipeline_stages_match_diffusers(
     variant: str,
     _hardware,
+    sana_model_parallel,
 ) -> None:
     """Compare the first denoising step without replacing native attention.
 

tests/engine/test_omni_request.py

@@ -13,15 +13,13 @@ substitutable with the base class — including base-style calls that pass
 
 import inspect
 
-import msgspec
 import numpy as np
 import pytest
 import torch
 from vllm.sampling_params import SamplingParams
-from vllm.v1.engine import EngineCoreRequest
 from vllm.v1.request import Request
 
-from vllm_omni.engine import OmniEngineCoreRequest, PromptEmbedsPayload
+from vllm_omni.engine import PromptEmbedsPayload
 from vllm_omni.request import OmniRequest, OmniStreamingUpdate
 
 pytestmark = [pytest.mark.core_model, pytest.mark.cpu]
@@ -113,24 +111,3 @@ def test_model_intermediate_buffer_round_trips_to_streaming_update():
     assert update is not None
     assert request.model_intermediate_buffer is buffer
     assert update.model_intermediate_buffer is buffer
-
-
-@pytest.mark.parametrize("session_id", [None, "session"])
-def test_session_survives_omni_upgrade_wire_transport_and_scheduler_conversion(session_id):
-    core = EngineCoreRequest(
-        request_id="session-request",
-        prompt_token_ids=[1, 2],
-        mm_features=None,
-        sampling_params=SamplingParams(max_tokens=2),
-        pooling_params=None,
-        arrival_time=0.0,
-        lora_request=None,
-        cache_salt=None,
-        data_parallel_rank=None,
-        session_id=session_id,
-    )
-    upgraded = OmniEngineCoreRequest.from_request(core)
-    decoded = msgspec.msgpack.decode(msgspec.msgpack.encode(upgraded), type=OmniEngineCoreRequest)
-    scheduled = OmniRequest.from_engine_core_request(decoded, block_hasher=None)
-    assert decoded.session_id == session_id
-    assert scheduled.session_id == session_id

vllm_omni/engine/__init__.py

@@ -123,7 +123,6 @@ class OmniEngineCoreRequest(EngineCoreRequest):
             trace_headers=request.trace_headers,
             resumable=request.resumable,
             external_req_id=request.external_req_id,
-            session_id=request.session_id,
             reasoning_ended=request.reasoning_ended,
             reasoning_parser_kwargs=request.reasoning_parser_kwargs,
             abort_immediately=request.abort_immediately,