jupyter-serial-scheduler 0.2.0

A tool to execute Jupyter Notebooks sequentially with resource isolation.

Jupyter Notebook Scheduler

A simple Python script to execute multiple Jupyter Notebooks sequentially. This tool ensures that each notebook runs in a separate process, guaranteeing that GPU memory and other resources are fully released between runs.

Features

• Sequential Execution: Runs notebooks one after another.
• Resource Isolation: Each notebook runs in a fresh process, clearing the kernel and freeing GPU memory upon completion.
• Detailed Logging:
  • Real-time cell-level progress in the console (clean output).
  • Automatic log file generation in logs/ for each run.
  • Captures Cell Outputs: Actual output from code cells (print statements, etc.) is saved to the log file but hidden from the console to keep it tidy.
• Robust Error Handling:
  • Stops execution if a notebook fails.
  • Saves the failed notebook with the error traceback, allowing for easy debugging.

Installation

Option A: Install from PyPI (Recommended)

pip install jupyter-serial-scheduler

Option B: Install from Source

1. Clone the repository:

   git clone https://github.com/chandan2300/Serial-Notebook-Executor.git
   cd Serial-Notebook-Executor
   

2. Install the package:

   pip install .
   

Usage

Once installed, you can use the nb-scheduler command from anywhere.

1. Place Notebooks: Put your .ipynb files in a directory (e.g., notebooks/).

2. Run the Scheduler:

   Option A: Run default list (defined in script)

   nb-scheduler
   

   Option B: Run specific notebooks

   nb-scheduler notebooks/test_nb_1.ipynb notebooks/test_nb_2.ipynb
   

Configuration

You can modify scheduler.py to change the list of notebooks or the directory they are located in.