1

我正在尝试让 mlflow mlproject 工作。

当我使用 repo 名称运行 mlflow 时

mlflow run  git@gitlabe2.xx.yy.zz:name/mlflow-example.git

执行失败并出现以下错误

File "/home/example/miniconda/envs/mlflow/lib/python3.7/site-packages/mlflow/projects/__init__.py", line 265, in run
use_conda=use_conda, storage_dir=storage_dir, synchronous=synchronous, run_id=run_id)
File "/home/example/miniconda/envs/mlflow/lib/python3.7/site-packages/mlflow/projects/__init__.py", line 144, in _run
conda_env_name = _get_or_create_conda_env(project.conda_env_path)
File "/home/example/miniconda/envs/mlflow/lib/python3.7/site-packages/mlflow/projects/__init__.py", line 144, in _run
conda_env_name = _get_or_create_conda_env(project.conda_env_path)
File "/home/example/miniconda/envs/mlflow/lib/python3.7/site-packages/mlflow/projects/__init__.py", line 498, in _get_or_create_conda_env
conda_env_path], stream_output=True)
File "/home/example/miniconda/envs/mlflow/lib/python3.7/site-packages/mlflow/utils/process.py", line 38, in exec_cmd
raise ShellCommandException("Non-zero exitcode: %s" % (exit_code))

关于我应该在哪里看的任何指示？

怀疑 conda.yaml 文件有一些问题，尤其是 conda env 名称。对于创建项目的环境和运行项目的环境，我有不同的名称。有关系吗 ？

谢谢

4

3 回答 3

0

我遇到了这个问题，这是由于我从环境中创建 conda.yaml 时对其进行了编码。以下是基于我在 vscode 中对 windows 10 的观察。

这不起作用：

conda env export > conda.yamlUTF-16 LE这会导致在创建环境时导致 conda 失败 的编码。

改用这个：

conda env export -f conda.yaml 这导致UTF-8conda 可以用来创建环境的编码。

于 2021-01-31T01:05:55.287 回答
0

One of the reasons for getting this error may be low disk space. If you are out of disk space, consider freeing some memory and then go to folder ~/anaconda/envs and delete the folder under this directory. The folder name starts with conda. This folder contains some libraries used during the conda environment creation, but because of the low memory it stopped and did not install all libraries needed. After freeing some memory you can run it again.

于 2021-10-18T07:55:57.453 回答
0

怀疑的问题与 conda yaml 文件有关。使用 --no-builds 选项创建项目的 yaml 文件解决了该问题

conda env export --no-builds > conda.yaml

谢谢

于 2019-07-28T03:43:54.000 回答