From 5221a1172ae68515f812b5a88a954f0e946e45a6 Mon Sep 17 00:00:00 2001 From: zhengchenyu Date: Fri, 5 Dec 2025 11:15:56 +0800 Subject: [PATCH] Use fully qualified domain name instead of hostname. --- torchft/checkpointing/http_transport.py | 2 +- torchft/manager.py | 2 +- torchft/parameter_server.py | 4 ++-- 3 files changed, 4 insertions(+), 4 deletions(-) diff --git a/torchft/checkpointing/http_transport.py b/torchft/checkpointing/http_transport.py index 5613b2d5..da5426ae 100644 --- a/torchft/checkpointing/http_transport.py +++ b/torchft/checkpointing/http_transport.py @@ -171,7 +171,7 @@ def address(self) -> str: an HTTP address """ port = self._server.socket.getsockname()[1] - return f"http://{socket.gethostname()}:{port}/checkpoint/" + return f"http://{socket.getfqdn()}:{port}/checkpoint/" def _serve(self) -> None: try: diff --git a/torchft/manager.py b/torchft/manager.py index 7e785846..286282dd 100644 --- a/torchft/manager.py +++ b/torchft/manager.py @@ -179,7 +179,7 @@ def __init__( lighthouse_addr: Optional[str] = None, replica_id: Optional[str] = None, port: Optional[int] = None, - hostname: str = socket.gethostname(), + hostname: str = socket.getfqdn(), heartbeat_interval: timedelta = timedelta(milliseconds=100), checkpoint_transport: Optional[CheckpointTransport[Dict[str, T]]] = None, init_sync: bool = True, diff --git a/torchft/parameter_server.py b/torchft/parameter_server.py index 533c923d..06f7abd8 100644 --- a/torchft/parameter_server.py +++ b/torchft/parameter_server.py @@ -70,7 +70,7 @@ def do_GET(self): session_id = str(uuid.uuid4()) store_addr = ( - f"{socket.gethostname()}:{ps.store.port}/session/{session_id}" + f"{socket.getfqdn()}:{ps.store.port}/session/{session_id}" ) logger.info(f"creating new session {session_id}") @@ -123,7 +123,7 @@ def address(self) -> str: an HTTP address """ port = self._server.socket.getsockname()[1] - return f"http://{socket.gethostname()}:{port}/new_session" + return f"http://{socket.getfqdn()}:{port}/new_session" def _serve(self) -> None: try: