remove changes

Author: deependujha

src/lightning/fabric/plugins/environments/lightning.py

@@ -15,17 +15,11 @@
 import os
 import socket
 
-from filelock import FileLock
 from typing_extensions import override
 
 from lightning.fabric.plugins.environments.cluster_environment import ClusterEnvironment
 from lightning.fabric.utilities.rank_zero import rank_zero_only
 
-BASE_PORT = 10000
-MAX_PORT = 65000
-STEP = 20
-LOCK_FILE = "lightning_ports.lock"
-
 
 class LightningEnvironment(ClusterEnvironment):
     """The default environment used by Lightning for a single node or free cluster (not managed).
@@ -111,57 +105,15 @@ def teardown(self) -> None:
             del os.environ["WORLD_SIZE"]
 
 
-def find_free_network_port(base: int = BASE_PORT, step: int = STEP) -> int:
+def find_free_network_port() -> int:
     """Finds a free port on localhost.
 
     It is useful in single-node training when we don't want to connect to a real main node but have to set the
     `MASTER_PORT` environment variable.
 
     """
-    PL_FORCE_DETERMINISTIC_PORTS = os.environ.get("PL_FORCE_DETERMINISTIC_PORTS", "0")
-
-    if PL_FORCE_DETERMINISTIC_PORTS == "0":
-        s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
-        s.bind(("", 0))
-        port = s.getsockname()[1]
-        s.close()
-        return port
-
-    # use the last assigned port + step strategy with a file lock to avoid race conditions
-    lock_path = os.path.join(os.getcwd(), LOCK_FILE)
-    os.makedirs(os.path.dirname(lock_path), exist_ok=True)
-
-    with FileLock(lock_path + ".lock"):
-        # read used ports
-        if os.path.exists(lock_path):
-            with open(lock_path) as f:
-                used = [int(x.strip()) for x in f if x.strip()]
-        else:
-            used = []
-
-        candidate = base if not used else used[-1] + step
-        if candidate > MAX_PORT:
-            candidate = base
-
-        tries = 0
-        max_tries = (MAX_PORT - base) // step
-        while (not is_port_available(candidate) or candidate in used) and tries < max_tries:
-            candidate += step
-            if candidate > MAX_PORT:
-                candidate = base
-            tries += 1
-
-        if tries >= max_tries:
-            raise RuntimeError("No free port found in range")
-
-        # write the new port to the file
-        with open(lock_path, "a") as f:
-            f.write(f"{candidate}\n")
-
-    return candidate
-
-
-def is_port_available(port: int) -> bool:
-    with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
-        s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
-        return s.connect_ex(("localhost", port)) != 0
+    s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+    s.bind(("", 0))
+    port = s.getsockname()[1]
+    s.close()
+    return port

tests/tests_fabric/conftest.py

@@ -66,7 +66,6 @@ def restore_env_variables():
         "CUDA_MODULE_LOADING",  # leaked by PyTorch
         "CRC32C_SW_MODE",  # set by tensorboardX
         "OMP_NUM_THREADS",  # set by our launchers
-        "PL_FORCE_DETERMINISTIC_PORTS",  # to force deterministic behavior in tests
         # set by torchdynamo
         "TRITON_CACHE_DIR",
         "TORCHINDUCTOR_CACHE_DIR",
@@ -137,28 +136,6 @@ def reset_deterministic_algorithm():
     torch.use_deterministic_algorithms(False)
 
 
-@pytest.fixture(scope="session", autouse=True)
-def cleanup_lightning_ports():
-    """Delete Lightning deterministic port files after all tests."""
-    from lightning.fabric.plugins.environments.lightning import LOCK_FILE
-
-    # Ensure deterministic ports in tests
-    os.environ["PL_FORCE_DETERMINISTIC_PORTS"] = "1"
-
-    yield
-
-    files_to_remove = [
-        LOCK_FILE,  # the port storage file
-        LOCK_FILE + ".lock",  # the filelock lock file
-    ]
-    for f in files_to_remove:
-        try:
-            if os.path.exists(f):
-                os.remove(f)
-        except Exception as e:
-            print(f"Warning: failed to remove {f}: {e}")
-
-
 @pytest.fixture
 def reset_cudnn_benchmark():
     """Ensures that the `torch.backends.cudnn.benchmark` setting gets reset before the next test runs."""
@@ -221,8 +198,6 @@ def leave_no_artifacts_behind():
     yield
     files_after = {p for p in tests_root.rglob("*") if "__pycache__" not in p.parts}
     difference = files_after - files_before
-    # ignore lightning port files
-    difference = {f for f in difference if not f.name.startswith("lightning_ports.lock")}
     difference = {str(f.relative_to(tests_root)) for f in difference}
     # ignore the .coverage files
     difference = {f for f in difference if not f.endswith(".coverage")}

tests/tests_pytorch/conftest.py

@@ -94,7 +94,6 @@ def restore_env_variables():
         "TF_CPP_MIN_LOG_LEVEL",
         "TF_GRPC_DEFAULT_OPTIONS",
         "XLA_FLAGS",
-        "PL_FORCE_DETERMINISTIC_PORTS",  # to force deterministic behavior in tests
         "TORCHINDUCTOR_CACHE_DIR",  # leaked by torch.compile
         # TensorFlow and TPU related variables
         "TF2_BEHAVIOR",
@@ -139,28 +138,6 @@ def reset_deterministic_algorithm():
     torch.use_deterministic_algorithms(False)
 
 
-@pytest.fixture(scope="session", autouse=True)
-def cleanup_lightning_ports():
-    """Delete Lightning deterministic port files after all tests."""
-    from lightning.fabric.plugins.environments.lightning import LOCK_FILE
-
-    # Ensure deterministic ports in tests
-    os.environ["PL_FORCE_DETERMINISTIC_PORTS"] = "1"
-
-    yield
-
-    files_to_remove = [
-        LOCK_FILE,  # the port storage file
-        LOCK_FILE + ".lock",  # the filelock lock file
-    ]
-    for f in files_to_remove:
-        try:
-            if os.path.exists(f):
-                os.remove(f)
-        except Exception as e:
-            print(f"Warning: failed to remove {f}: {e}")
-
-
 @pytest.fixture(autouse=True)
 def thread_police_duuu_daaa_duuu_daaa():
     """Attempts to stop left-over threads to avoid test interactions."""
@@ -349,8 +326,6 @@ def leave_no_artifacts_behind():
     yield
     files_after = {p for p in tests_root.rglob("*") if "__pycache__" not in p.parts}
     difference = files_after - files_before
-    # ignore lightning port files
-    difference = {f for f in difference if not f.name.startswith("lightning_ports.lock")}
     difference = {str(f.relative_to(tests_root)) for f in difference}
     # ignore the .coverage files
     difference = {f for f in difference if not f.endswith(".coverage")}