如何解决MLFlow UI出现的FileNotFoundError: [WinError 2]错误？

问题背景 #

我通过以下代码创建MLFlow跟踪文件夹并记录指标、参数：

dir_parent = os.getcwd()
dir_mlflow = 'file:' + os.sep + os.path.join(dir_parent, 'mlflow_test')
mlflow.set_tracking_uri(dir_mlflow)

# 记录参数到mlflow
mlflow.log_param('batch_size', args.batch_size)
mlflow.log_param('epochs', args.epochs)
mlflow.log_param('learning_rate', args.learning_rate)

# 定义回调函数跟踪训练过程指标
class LogMetricsCallback(keras.callbacks.Callback):
    def on_epoch_end(self, epoch, logs={}):
        mlflow.log_metric('training_loss', logs['loss'], epoch)
        mlflow.log_metric('training_accuracy', logs['accuracy'], epoch)

# 获取测试集指标
test_loss, test_acc = model.evaluate(x_test, y_test, verbose=2)
mlflow.log_metric('test_loss', test_loss)
mlflow.log_metric('test_accuracy', test_acc)

# 将模型保存到mlflow
mlflow.keras.log_model(model, artifact_path = 'keras-model')

执行后生成mlflow_test文件夹，内含.trash、0、mlruns三个子文件夹。我进入该文件夹后执行mlflow ui或mlflow ui --host 0.0.0.0命令尝试查看运行记录，触发以下错误：

Traceback (most recent call last):
  File "C:\Program Files\Python39\lib\runpy.py", line 197, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "C:\Program Files\Python39\lib\runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\mlflow\__main__.py", line 3, in <module>
    cli.main()
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\click\core.py", line 1055, in main
    rv = self.invoke(ctx)
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\click\core.py", line 1657, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\click\core.py", line 1404, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\click\core.py", line 760, in invoke
    return __callback(*args, **kwargs)
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\mlflow\cli.py", line 399, in server
    _run_server(
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\mlflow\server\__init__.py", line 161, in _run_server
    _exec_cmd(full_command, extra_env=env_map, capture_output=False)
  File "C:\Users\PJ\AppData\Roaming\Python\Python39\site-packages\mlflow\utils\process.py", line 95, in _exec_cmd
    process = subprocess.Popen(
  File "C:\Program Files\Python39\lib\subprocess.py", line 951, in __init__
    self._execute_child(args, executable, preexec_fn, close_fds,
  File "C:\Program Files\Python39\lib\subprocess.py", line 1420, in _execute_child
    hp, ht, pid, tid = _winapi.CreateProcess(executable, args,
FileNotFoundError: [WinError 2] The system cannot find the file specified

环境信息 #

• MLFlow版本：2.1.1
• Python版本：3.9.5 64位
• 操作系统：Windows 10

解决方法 #

1. 对齐Python环境与MLFlow路径 #

• 执行where python查看当前命令行使用的Python路径，再执行pip show mlflow确认MLFlow安装的site-packages路径是否匹配，确保使用的是同一环境。
• 如果用了虚拟环境，先激活对应环境再运行mlflow ui。

2. 显式指定跟踪URI启动UI #

不用进入文件夹，直接指定运行记录的存储路径启动：

mlflow ui --backend-store-uri ./mlflow_test/mlruns

或使用完整绝对路径：

mlflow ui --backend-store-uri "C:\你的完整路径\mlflow_test\mlruns"

3. 替换兼容Windows的服务器 #

MLFlow默认用的gunicorn不支持Windows，改用waitress：

• 先安装waitress：

pip install waitress

• 启动UI时指定使用waitress：

mlflow ui --backend-store-uri ./mlflow_test/mlruns --serve-backend waitress

4. 升级MLFlow版本 #

MLFlow 2.1.1存在Windows下的已知兼容性问题，升级到2.3.0及以上版本可修复：

pip install --upgrade mlflow

内容的提问来源于stack exchange，提问作者PJ_