yottactl 0.1.4

This is a lightweight library that works as a connector to Yottalabs public API.

Yottalabs Public API Connector (Python)

A lightweight Python library for connecting to the Yottalabs public API.
Easily interact with /openapi/v1/* endpoints, with support for custom base URLs, request timeouts, HTTP proxies, and more.

Requirements

• Python 3.8 or higher

Installation

pip install yottactl

Quickstart

from yotta.pod import PodApi

# API key is required for user data endpoints
client = PodApi(api_key='<api_key>')

params = {
    "image": "yottalabsai/pytorch:2.8.0-py3.11-cuda12.8.1-cudnn-devel-ubuntu22.04-2025050802",
    "gpu_type": "NVIDIA_L4_24G",
    "gpu_count": 1,
    "expose": [
        {"port": 22, "protocol": "SSH"}
    ]
}

response = client.new_pod(**params)
print(response)

See the examples/ folder for more usage examples.

Configuration

To use the example scripts, create a examples/config.ini file with your API key:

[keys]
api_key=YOUR_API_KEY_HERE

Customization

• Base URL:
  If not provided, defaults to https://api.yottalabs.ai.

• Timeout:
  Set the timeout parameter (in seconds) to control how long to wait for a server response.
  By default, requests do not time out.

  client = PodApi(timeout=1)
  

• Logging:
  Set the log level to DEBUG to log request URLs, payloads, and responses.

  client = PodApi(debug=True)
  

Error Handling

Two types of errors are raised:

• yotta.error.ClientError
  Raised for 4XX client errors. Properties:

  • status_code: HTTP status code
  • error_code: Server error code (if available)
  • error_message: Server error message
  • header: Full response headers
  • error_data: Additional data (if provided)

• yotta.error.ServerError
  Raised for 5XX server errors.

Contributing

Contributions are welcome! Please open issues or pull requests.

License

MIT