RuntimeError: The server socket has failed to listen on any local network address. The server socket

RuntimeError: The server socket has failed to listen on any local network address. The server socket has failed to bind to [::]:6021 (errno: 98 - Address already in use). The server socket has failed to bind to 0.0.0.0:6021 (errno: 98 - Address already in use).

报错是说我的服务器无法监听到我的本地网络地址，6021是我的本地端口号。

出现这个错误的原因：

我通过

 oarsub -p "host in ('byta6000i0')" -l "cpu=1/gpuset=1/host=1",walltime=168:00:00 "bash exp.sh"

这句指令在服务器的passive mode 上执行了我的train.py任务，我发现结果不太对，于是我就又重新修改了网络结构，想再跑一下，但是我不想停止我之前的训练，于是又执行了上述指令，而没有修改代码里指定的端口号6021

解决办法：

parser.add_argument('--port_num', type=str, default='6021',help='port selection for code')

把这句代码里的 --port_num 修改为6022

看了网上有相似的错误，也有一些解决办法，但是感觉还是有点麻烦。

如果我的不能帮助到你，可以参考：

RuntimeError: The server socket has failed to listen on any local network address. The server socket has failed to bind to [::]:29500 - distributed - PyTorch Forums

https://github.com/microsoft/DeepSpeed/issues/3916

torchrun (Elastic Launch) — PyTorch 2.1 documentation