如果 osmnx 预先下载数据，Dask 会抛出 DNS 类型错误

Question

我在使用 OSMNX 下载数据后尝试运行 dask；

Dask 抛出一个类型错误，如下所示；我尝试在运行 OSMNX 函数之前实现 Client()，它运行良好；在本地运行 Dask 之前下载文件会导致问题

我在 Ubuntu WSL 上运行 VSC；

study_area = study_area.geometry.unary_union

osm_graph= ox.graph.graph_from_polygon(study_area, network_type='drive')
osm_graph = ox.projection.project_graph(osm_graph, to_crs=local_crs)

streets = ox.consolidate_intersections(osm_graph, rebuild_graph=True, tolerance=15, dead_ends=False)

streets = ox.graph_to_gdfs(
    osm_graph,
    nodes=False,
    edges=True,
    node_geometry=False,
    fill_edge_geometry=True
)
streets.head()

client = Client()

TypeError                                 Traceback (most recent call last)

File ~/miniconda3/lib/python3.10/site-packages/tornado/tcpclient.py:265, in TCPClient.connect(self, host, port, af, ssl_options, max_buffer_size, source_ip, source_port, timeout)
    261     addrinfo = await gen.with_timeout(
    262         timeout, self.resolver.resolve(host, port, af)
    263     )
    264 else:
--> 265     addrinfo = await self.resolver.resolve(host, port, af)
    266 connector = _Connector(
    267     addrinfo,
    268     functools.partial(
   (...)
    273     ),
    274 )
    275 af, addr, stream = await connector.start(connect_timeout=timeout)

File ~/miniconda3/lib/python3.10/site-packages/distributed/comm/tcp.py:476, in _DefaultLoopResolver.resolve(self, host, port, family)
    466 async def resolve(
    467     self, host: str, port: int, family: socket.AddressFamily = socket.AF_UNSPEC
    468 ) -> list[tuple[int, Any]]:
   (...)
    472     # matter (we discard the one we get back in the results),
    473     # so the addresses we return should still be usable with SOCK_DGRAM.
    474     return [
    475         (fam, address)
--> 476         for fam, _, _, _, address in await _getaddrinfo(
    477             host, port, family=family, type=socket.SOCK_STREAM
    478         )
    479     ]

File ~/miniconda3/lib/python3.10/site-packages/distributed/comm/tcp.py:436, in _getaddrinfo(host, port, family, type)
    429 if sys.version_info >= (3, 9):
    430     # If host and port are numeric, then getaddrinfo doesn't block and we
    431     # can skip get_running_loop().getaddrinfo which is implemented by
    432     # running in a ThreadPoolExecutor. So we try first with the
    433     # _NUMERIC_ONLY flags set, and then only use the threadpool if that
    434     # fails with EAI_NONAME:
    435     try:
--> 436         return socket.getaddrinfo(
    437             host,
    438             port,
    439             family=family,
    440             type=type,
    441             flags=_NUMERIC_ONLY,
    442         )
    443     except socket.gaierror as e:
    444         if e.errno != socket.EAI_NONAME:

TypeError: _config_dns.._getaddrinfo() got an unexpected keyword argument 'family'

Answer 1

这看起来像是 _getaddrinfo() 中 family 关键字参数的错误

你应该尝试更新 dask 和 tornado 到他们的最新版本

pip install --upgrade dask tornado

您也可以尝试通过替换 IP

 来指定调度程序的 IP 地址

client = Client('tcp://<IP>:8786')