sg!ddlZddlZddlZddlmZddlmZddlmZ ddl m Z ddl m Z gdZe eZdZdZd Z dd ed ed ed edededeefdZdZdZdejfdZy)N)closing)Optional) get_logger)barrier)create_c10d_store get_free_portget_socket_with_portzAddress already in usezSocket Timeoutz_tcp_store/num_members is_server server_addr server_port world_sizetimeoutwait_for_workers use_libuvc |tjdtjj dddk(}|dk(r|dkDrt d|d||dk7rtj d||dk(r|nd} |dk7r|} n t} tj d || |||| tj|| ||tj| || } |rt| || tj d | S#t$rS} t| tk(r6||krtjd| |||dz }ntd|d| d| Yd} ~ nd} ~ wwxYw)Nzargument use_libuv is deprecated and ignored. Set USE_LIBUV environment variable to "0" to disable libuv, or "1" to enable it. If the env var is not set, libuv will be used by default. USE_LIBUV1zCserver_port must be specified when world_size > 1, got server_port=z , world_size=z+sever_port: %s, specified, ignoring retriesziCreating c10d store on %s:%s world_size : %s is_server : %s timeout(sec): %s use_libuv : %s )seconds) host_nameportr is_masterrrr)rzSuccessfully created c10d storez)port: %s already in use, attempt: [%s/%s]zon z, port: z already in use)loggerwarningosenvironget ValueErrorinfordistTCPStoredatetime timedelta_check_full_rank RuntimeErrorstr_ADDRESS_IN_USE) r r r r rrretriesrattemptrstorees ^/var/www/html/venv/lib/python3.12/site-packages/torch/distributed/elastic/utils/distributed.pyrrs 9  {C0C7IbZ!^QR]Q^^klvkw x  b A;O%*gG  " D ?D  #       # MM%%# **7;!1#E   GD KK9 :L  1v(W$NNC  qLG&k](4&H  G s)AD E A EEc t||t|y#t$r(}t|tk(rt d|d|d}~wwxYw)N) key_prefixbarrier_timeoutztimed out waiting for all z members to join)r_TCP_STORE_INITr&r'_SOCKET_TIMEOUT TimeoutError)r+r rr,s r-r%r%vsSzowW  q6_ $,ZL8HI   s A#AAc~t}t|5|jdcdddS#1swYyxYw)a Returns an unused port on localhost. This function finds an unused port on localhost by opening to socket to bind to a port and then closing it. Returns: int: an unused port on localhost Example: >>> # xdoctest: +SKIP("Nondeterministic") >>> get_free_port() 63976 ..note: The port returned by :func:`get_free_port` is not reserved and may be taken by another process after this function returns. rN)r r getsockname)socks r-rrs9& !D %!!$%%%s3<returnctjddtjtj}|D]G}|\}}}}}tj|||} |j d|j d|cStd#t $r1}|jtjd|Yd}~d}~wwxYw) a Returns a free port on localhost that is "reserved" by binding a temporary socket on it. Close the socket before passing the port to the entity that requires it. Usage example :: sock = _get_socket_with_port() with closing(sock): port = sock.getsockname()[1] sock.close() # there is still a race-condition that some other process # may grab this port before func() runs func(port) localhostN)hostrfamilytype)r9rrzSocket creation attempt failed.)exc_infozFailed to create a socket) socket getaddrinfo AF_UNSPEC SOCK_STREAMbindlistenOSErrorcloserrr&)addrsaddrr;r<proto_sr,s r-r r s"    tF,<,<6CUCU E J$(!eQ MM&$ . J FF# $ HHQKH J 2 33 J GGI NN contextlibrtypingrtorch.distributed distributedr!'torch.distributed.elastic.utils.loggingr%torch.distributed.elastic.utils.storer__all____name__rr(r2r1boolr'intfloatrr%rr r-rYs  >9 I H *"* ! $UUUU U  U  U~Up %04fmm4rX