Open jeff-shepherd opened 1 year ago
We are working to update the Compute Instance image to avoid this issue. It can be resolved by running the following command on the Compute Instance:
conda activate azureml_py38
pip install certifi==2022.9.24
sudo chmod 755 /anaconda/envs/azureml_py38/lib/python3.8/site-packages/nimbusml/internal/libs/pybridge.so
@jeff-shepherd Thanks for this. Was having the same issue and it worked.
AzureML V1-SDK v1.49.0 has been released on Feb 13. Please make sure your environment is upgraded to latest version (v1.49.0 or above) to avoid "Operation not permitted" mentioned above. https://pypi.org/project/azureml-sdk/
Describe the bug Local and many model automl V1 sample notebooks can fail on the latest Compute Instance. This includes:
To Reproduce Steps to reproduce the behavior:
Expected behavior The notebook should execute successfully but if fails with: Install the required versions of packages using the requirements file. Requirements file location: /anaconda/envs/azureml_py38/lib/python3.8/site-packages/azureml/automl/core/validated_linux_requirements.txt. Alternatively, use remote target to avoid dependency management. Required version/Installed version certifi<=2022.9.24/certifi 2022.12.7
The auto-ml-forecasting-backtest-many-models.ipynb notebook fails at "from azureml.contrib.automl.pipeline.steps import AutoMLPipelineBuilder" with: File /anaconda/envs/azureml_py38/lib/python3.8/site-packages/azureml/training/tabular/models/forecasting_pipeline_wrapper.py:10 7 import uuid 8 from typing import Any, Dict, List, Optional, Tuple, Union, cast ---> 10 import nimbusml 11 import numpy as np 12 import pandas as pd File /anaconda/envs/azureml_py38/lib/python3.8/site-packages/nimbusml/init.py:33 31 mode = oct(os.stat(dotso).st_mode & 0o777) 32 if mode != "0o755": ---> 33 os.chmod(dotso, 0o755) 35 # clean up the package namespace 36 del os, sys PermissionError: [Errno 1] Operation not permitted: '/anaconda/envs/azureml_py38/lib/python3.8/site-packages/nimbusml/internal/libs/pybridge.so'