FedVision
FedVision copied to clipboard
部署fedVision报错
使用的是fedvision-deploy deploy deploy --config standalone_template.yaml 命令 报错信息如下 Traceback (most recent call last):
File "/root/fedvision/fedvision/bin/fedvision-deploy", line 8, in
File "/root/fedvision/fedvision/lib/python3.6/site-packages/fedvision_deploy_toolkit/_deploy.py", line 44, in deploy _maybe_create_python_venv(machine)
File "/root/fedvision/fedvision/lib/python3.6/site-packages/fedvision_deploy_toolkit/_deploy.py", line 69, in _maybe_create_python_venv raise RuntimeError(f"python executable {machine['python']} not valid")
KeyError: 'python'
standalone_template.yaml 文件 ` machines:
- name: machine1 ip: 127.0.0.1 ssh_string: 127.0.0.1:22 base_dir: /data/projects/fedvision python_for_venv_create: python3 # use to create venv, python3.7+ required
coordinator start/stop only if machine provided
coordinator: name: coordinator1 machine: machine1 port: 10000
clusters:
- name: cluster1
manager:
machine: machine1
port: 10001
workers:
- name: worker1 machine: machine1 ports: 12000-12099 max_tasks: 10
masters:
-
name: master1 machine: machine1 submit_port: 10002 coordinator: coordinator1 cluster: cluster1
-
name: master2 machine: machine1 submit_port: 10003 coordinator: coordinator1 cluster: cluster1
-
name: master3 machine: machine1 submit_port: 10004 coordinator: coordinator1 cluster: cluster1
-
name: master4 machine: machine1 submit_port: 10005 coordinator: coordinator1 cluster: cluster1 `
他启动时是这样的,会ssh到每台机器上然后 验证python版本 -> 创建虚拟环境 -> 拷贝代码 -> 在虚拟环境安装依赖 你报这个错的原因是在验证python版本就错了。看看部署的机器上 python3 能否执行、版本符不符合要求。
报错的内容是配置里没'python'这个key,这个是作者代码的问题,正常的话不会走到这个分支。跟实际的问题没有关系、