[V0 Deprecation] Remove VLLM_USE_V1 from tests (#26341)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

tests/v1/engine/test_async_llm.py

@@ -95,17 +95,11 @@ async def generate(
 )
 @pytest.mark.asyncio
 async def test_load(
-    monkeypatch: pytest.MonkeyPatch,
     output_kind: RequestOutputKind,
     engine_args: AsyncEngineArgs,
     prompt: PromptType,
 ):
-    # TODO(rickyx): Remove monkeypatch once we have a better way to test V1
-    # so that in the future when we switch, we don't have to change all the
-    # tests.
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(engine_args)
         after.callback(engine.shutdown)
@@ -149,14 +143,11 @@ async def test_load(
 )
 @pytest.mark.asyncio
 async def test_abort(
-    monkeypatch: pytest.MonkeyPatch,
     output_kind: RequestOutputKind,
     engine_args: AsyncEngineArgs,
     prompt: PromptType,
 ):
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(engine_args)
         after.callback(engine.shutdown)
@@ -222,13 +213,8 @@ async def test_abort(
     "output_kind", [RequestOutputKind.DELTA, RequestOutputKind.FINAL_ONLY]
 )
 @pytest.mark.asyncio
-async def test_multi_abort(
-    monkeypatch: pytest.MonkeyPatch,
-    output_kind: RequestOutputKind,
-):
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+async def test_multi_abort(output_kind: RequestOutputKind):
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
         after.callback(engine.shutdown)
@@ -304,14 +290,11 @@ async def test_multi_abort(
 )
 @pytest.mark.asyncio
 async def test_finished_flag(
-    monkeypatch: pytest.MonkeyPatch,
     n: int,
     engine_args: AsyncEngineArgs,
     prompt: PromptType,
 ):
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(engine_args)
         after.callback(engine.shutdown)
@@ -341,12 +324,10 @@ async def test_finished_flag(
 )
 @pytest.mark.asyncio
 async def test_mid_stream_cancellation(
-    monkeypatch: pytest.MonkeyPatch, engine_args: AsyncEngineArgs, prompt: PromptType
+    engine_args: AsyncEngineArgs, prompt: PromptType
 ):
     """Test that requests can be cancelled mid-stream."""
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(engine_args)
         after.callback(engine.shutdown)
@@ -411,9 +392,7 @@ async def test_customize_loggers(monkeypatch):
     be added to the default loggers.
     """
 
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(
                 TEXT_ENGINE_ARGS,
@@ -430,10 +409,8 @@ async def test_customize_loggers(monkeypatch):
 
 
 @pytest.mark.asyncio(scope="module")
-async def test_dp_rank_argument(monkeypatch: pytest.MonkeyPatch):
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+async def test_dp_rank_argument():
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
         after.callback(engine.shutdown)
@@ -466,7 +443,7 @@ async def test_dp_rank_argument(monkeypatch: pytest.MonkeyPatch):
 
 
 @pytest.mark.asyncio
-async def test_check_health(monkeypatch: pytest.MonkeyPatch):
+async def test_check_health():
     """Test that check_health returns normally for healthy engine
     and raises EngineDeadError when the engine is dead.
     """
@@ -474,9 +451,7 @@ async def test_check_health(monkeypatch: pytest.MonkeyPatch):
 
     from vllm.v1.engine.exceptions import EngineDeadError
 
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
         after.callback(engine.shutdown)
@@ -503,15 +478,10 @@ async def test_check_health(monkeypatch: pytest.MonkeyPatch):
     "output_kind", [RequestOutputKind.DELTA, RequestOutputKind.FINAL_ONLY]
 )
 @pytest.mark.asyncio
-async def test_abort_final_output(
-    monkeypatch: pytest.MonkeyPatch,
-    output_kind: RequestOutputKind,
-):
+async def test_abort_final_output(output_kind: RequestOutputKind):
     """Test that abort() returns a final output with correct information."""
 
-    with monkeypatch.context() as m, ExitStack() as after:
-        m.setenv("VLLM_USE_V1", "1")
-
+    with ExitStack() as after:
         with set_default_torch_num_threads(1):
             engine = AsyncLLM.from_engine_args(TEXT_ENGINE_ARGS)
         after.callback(engine.shutdown)