Help Center/
ModelArts/
Troubleshooting/
DevEnviron/
Custom Image Faults/
Faults of Custom Images on Notebook Instances
Updated on 2025-06-06 GMT+08:00
Faults of Custom Images on Notebook Instances
If an error occurs when using a custom image, the possible causes are as follows:
- The custom image does not belong to user ma-user or user group ma-group.
- The /home/ma-user directory that stores the custom image does not belong to user ma-user or user group ma-group.
- The directory permissions of /home/ma-user must be set to 750 for the custom image.
- The created custom image cannot run properly by running the docker run command in the local host.
- The fault may be caused by the installation of JupyterLab. In this case, run commands in local JupyterLab to list related static file paths, and delete and uninstall JupyterLab.
- The user service uses ports 8888 and 8889 of the official development environment. In this case, modify the process port number.
- PYTHONPATH and sys.path are specified by the user image. In this case, the fault is caused by the service startup invoking conflict. You need to start the instance before specifying PYTHONPATH and sys.path.
- The dedicated resource pool with SUDO permissions is enabled, while SUDO is not installed or installed incorrectly.
- The used cann and CUDA environments are not compatible.
- The Docker image configuration is wrong, or there is a network or firewall block, or problems with file permissions, missing dependencies, or build commands.
- Check whether the following problems occur in the Anaconda environment:
- An invalid directory, like pythonNone, exists in the {python_env}/lib folder. It should be named python plus the version number, such as python3.7. This might happen due to wrong environment setup or accidental actions.
- You might have created an empty folder in the Anaconda environment directory {conda}/envs or created an invalid folder in the environment's lib directory. This can break the Anaconda directory structure.
- You might have deleted necessary files in an environment directory. This can stop the Anaconda environment from working correctly.
- Your custom image uses remote SSH with an OpenSSH version that is incompatible or older than 8.0.
- If you use SSH to connect to a notebook instance and have more than 10 connections at once, your connection might be limited or refused.
- The permissions on the /home/ma-user/.ssh directory were changed, so SSH will not work. You can use either of these methods to solve the problem:
- Delete the /home/ma-user/.ssh directory when creating the image. You do not need to set up SSH; it is automatically generated when the notebook starts.
- Run the following commands to change the permission on the .ssh file:
cd /home/ma-user chmod 750 .ssh chmod 644 .ssh/authorized_keys chmod 644 .ssh/config chmod 640 .ssh/environment chmod 750 .ssh/etc chmod 640 .ssh/known_hosts chmod 750 .ssh/var chmod 600 .ssh/etc/ssh_host_rsa_key chmod 640 .ssh/etc/ssh_host_rsa_key.pub chmod 750 .ssh/etc/sshd_config chmod 750 .ssh/var/run/sshd.pid
Parent topic: Custom Image Faults
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
The system is busy. Please try again later.
For any further questions, feel free to contact us through the chatbot.
Chatbot