Import & Run notebooks and web based Git repositories

Backend.AI supports executing Jupyter notebook files and importing web based Git repositories such as GitHub and GitLab on the fly. You don’t need to create or download into your local storage and re-upload it. The only thing you have to do is input a valid URL you want to execute or import, and click the button on the right side.

Import and run Jupyter notebooks

To import Jupyter notebooks and run, you need one thing, the valid URL for the notebook file. For example, if you want execute Jupyter notebook that’s in github, you can copy and paste the URL and click IMPORT & RUN button.

Note

When you trying to IMPORT & RUN Jupyter notebook file with local address, It will be regarded as invalid. You have to input URL which is not starting from localhost.

Import & Run Jupyter notebook

After clicking the button, the dialog appears. This is a session launcher dialog same as when you starts the session at Sessions page or Summary page. Difference between import notebook and starting a new session is that import notebook automatically imports jupyter notebook in the URL, but simple starting a new session doesn’t do that. the rest is same. Click LAUNCH button to the notebook after setting the environments and resource allocation as needed.

Warning

The pop-up blocker must be turned off before clicking LAUNCH button to immediately see the running notebook window. Also, if there’s not enough resources to execute the session, imported Jupyter notebook will not run.

Session Launcher In Importing Jupyter notebook

You can see the importing operation is successfully completed in Sessions page.

Session page with imported Jupyter notebook

Create executable Jupyter notebook button

You can also create HTML or Markdown button about Jupyter notebook URL, too. Input a valid Jupyter notebook URL and click CREATE button. It will show code blocks that directly links to creating a session with notebook. You can see the badge code working by inserting it in the GitHub repositories or where it supports html or markdown.

Note

your account must be logined before clicking the button. Otherwise, you have to login first.

Create a Jupyter notebook button

Importing GitHub Repositories

Importing a GitHub repository is similar to import and running jupyter notebook. All you have to do is to fill out with github repository URL and click GET TO FOLDER button. If you can access to more than one storage host, you can select one from the list.

Import GitHub repository

Warning

If there are not enough resources to start a session or folder count is at the limit, then importing repository will fail. Please check resource statistics panel and Data & Storage page before importing the repository.

You can see the repository is successfully imported as a data folder with its name.

Import GitHub repository result

Importing GitLab Repositories

From 22.03, Backend.AI supports importing from GitLab. It’s almost the same as Importing GitHub Repositories, but you need to explicitly set the branch name to import.

Import GitLab repository

Note

If there’s data folder that has the same name already, the system will append _ (underscore) and number in the imported repository folder.