机器学习服务器远程会话设置失败

Machine Learning Server remote session setup fails

我想为 R 远程会话设置机器学习服务器。不幸的是它失败了 "Unknown Error"。这是我所做的以及我得到的输出。

设置

安装 Microsoft Machine Learning Server 9.3.0 - 离线安装/使用 R,不使用 Python。

说明来自 Offline installation for Machine Learning Server for Windows

安装步骤成功

MSI (s) (50:78) [09:50:07:880]: Product: Microsoft R Server -- Installation completed successfully.

配置

R 远程会话的配置。

说明来自 Launch the administration tool/CLI to manage the operationalization configuration

这里是 az ml admin bootstrap 的输出:

D:\R-Server\R_SERVER\o16n\Microsoft.MLServer.ComputeNode>az ml admin bootstrap
Admin password:
Confirm admin password:
Exception in thread stderr-stream:
Traceback (most recent call last):
  File "C:\Program Files (x86)\Microsoft SDKs\Azure\CLI2\lib\threading.py", line 916, in _bootstrap_inner
    self.run()
  File "C:\Program Files (x86)\Microsoft SDKs\Azure\CLI2\lib\threading.py", line 864, in run
    self._target(*self._args, **self._kwargs)
  File "C:\Users\A1119231\.azure\cliextensions\azure-ml-admin-cli\azext_admin\commandline_process.py", line 272, in read_stream
    for output_line in stream:
  File "C:\Program Files (x86)\Microsoft SDKs\Azure\CLI2\lib\encodings\cp1252.py", line 23, in decode
    return codecs.charmap_decode(input,self.errors,decoding_table)[0]
UnicodeDecodeError: 'charmap' codec can't decode byte 0x81 in position 0: character maps to <undefined>


D:\R-Server\PYTHON_SERVER\python.exe is not a program. Please install Python 3.5 there if you would like to use Python Functionality.


Starting Compute Node
FAILURE! Compute Node was unable to start.
Unknown Error
Starting Web Node
SUCCESS! Web Node started. (PID: 7100, Listening on URI: http://localhost:12800/)

[{'node': 'web', 'pid': '7100', 'state': 'running'}, {'node': 'compute', 'pid': '', 'state': 'not running'}]

login to "http://localhost:12800" successful.




Preparing to run diagnostics...

***********************
DIAGNOSTIC RESULTS:
***********************


Web Node Details:
  Logs: D:\R-Server\R_SERVER\o16n\Microsoft.MLServer.WebNode\logs:
  Configured Compute Node URIs:
    Configured Values: [http://localhost:12805/]

  Available compute nodes: 1

Compute Node Details:
  Health of 'http://localhost:12805/': fail
Error processing command AttributeError
'NoneType' object has no attribute 'items'

很奇怪,即使在安装过程中没有选择,它也会尝试执行 python.exe。

问题是 Microsoft 机器学习服务器设置中的错误。 如果您在设置期间取消选择 Python,将错误地创建配置文件 ...\o16n\Microsoft.MLServer.ComputeNode\appsettings.json

你必须更换:

"Python": {},

来自

"Python": {
  "JupyterKernelGateway": {
    "Port": 8888,
    "Enabled": false,
    "KernelName": "MLO16N"
  },
  "SubProcess": {
    "Enabled": false,
    "Path": null
  }
},

那么计算节点就可以启动成功了。