a h# @sddlZddlZddlZddlmZddlmZddlmZ ddl m Z ddl m Z gdZe eZdZdZd ZdeeeeeeeedddZddZddZejdddZdS)N)closing)Optional) get_logger)barrier)create_c10d_store get_free_portget_socket_with_portzAddress already in usezSocket Timeoutz_tcp_store/num_membersXT) is_server server_addr server_port world_sizetimeoutwait_for_workers use_libuvc CsT|durtdtjdddk}|dkrH|dkrHtd|d||dkr\td||dkrh|nd}|dkrz|} nt} td || ||||zBtj || ||t j |d ||d } |rt | ||d td | WSt yL} zZt| tkr6||krtd| |||d7}nt d|d| d| nWYd} ~ qld} ~ 00qldS)Nzargument use_libuv is deprecated and ignored. Set USE_LIBUV environment variable to "0" to disable libuv, or "1" to enable it. If the env var is not set, libuv will be used by default.Z USE_LIBUV1r r zCserver_port must be specified when world_size > 1, got server_port=z , world_size=z+sever_port: %s, specified, ignoring retriesziCreating c10d store on %s:%s world_size : %s is_server : %s timeout(sec): %s use_libuv : %s )seconds)Z host_nameportrZ is_masterrrr)rzSuccessfully created c10d storez)port: %s already in use, attempt: [%s/%s]zon z, port: z already in use)loggerwarningosenvironget ValueErrorinfordistZTCPStoredatetime timedelta_check_full_rank RuntimeErrorstr_ADDRESS_IN_USE) r rrrrrretriesrattemptrstoreer)Y/var/www/auris/lib/python3.9/site-packages/torch/distributed/elastic/utils/distributed.pyrsl       rc Cs`zt||t|dWnFtyZ}z.t|tkrDtd|d|nWYd}~n d}~00dS)N)Z key_prefixZbarrier_timeoutztimed out waiting for all z members to join)r_TCP_STORE_INITr"r#_SOCKET_TIMEOUT TimeoutError)r'rrr(r)r)r*r!vs  r!cCs>t}t||dWdS1s00YdS)a Returns an unused port on localhost. This function finds an unused port on localhost by opening to socket to bind to a port and then closing it. Returns: int: an unused port on localhost Example: >>> # xdoctest: +SKIP("Nondeterministic") >>> get_free_port() 63976 .. note:: The port returned by :func:`get_free_port` is not reserved and may be taken by another process after this function returns. r N)rr getsockname)sockr)r)r*rs r)returnc Cstjddtjtjd}|D]z}|\}}}}}t|||}z|d|d|WSty}z"|tj d|dWYd}~qd}~00qt ddS) a Returns a free port on localhost that is "reserved" by binding a temporary socket on it. Close the socket before passing the port to the entity that requires it. Usage example :: sock = _get_socket_with_port() with closing(sock): port = sock.getsockname()[1] sock.close() # there is still a race-condition that some other process # may grab this port before func() runs func(port) localhostN)hostrfamilytype)r1rrzSocket creation attempt failed.)exc_infozFailed to create a socket) socket getaddrinfo AF_UNSPEC SOCK_STREAMbindlistenOSErrorcloserrr")addrsaddrr3r4proto_sr(r)r)r*rs    &r)r r r Tr N)rrr6 contextlibrtypingrZtorch.distributedZ distributedrZ'torch.distributed.elastic.utils.loggingrZ%torch.distributed.elastic.utils.storer__all____name__rr$r,r+boolr#intfloatrr!rrr)r)r)r* s<      X