runqy-worker

Stateless task processor with server-driven bootstrap.
Part of the runqy distributed task queue system.
Documentation · Website

Features

• Server-driven bootstrap — Workers receive all config (Redis, code repo, queue routing) from the central server
• Automatic code deployment — Git clone, virtualenv creation, pip install on startup
• Two execution modes — Long-running process for ML inference, or one-shot per task
• Multi-queue support — Process tasks from multiple queues with priority weighting
• Health monitoring — Heartbeat and process health tracked in Redis

Zero-Touch Deployment

Push code to GitHub. Workers deploy themselves. No SSH. No Docker builds. No CI pipelines to maintain.

Installation

Docker

docker pull ghcr.io/publikey/runqy-worker:latest

For GPU/ML workloads:

docker pull ghcr.io/publikey/runqy-worker:inference

Binary Download

Download from GitHub Releases:

curl -LO https://github.com/publikey/runqy-worker/releases/latest/download/runqy-worker_latest_linux_amd64.tar.gz
tar -xzf runqy-worker_latest_linux_amd64.tar.gz

See Installation Guide for all platforms and options.

Quick Start

1. Create config.yml:

server:
  url: "http://localhost:3000"
  api_key: "your-api-key"

worker:
  queue: "inference"

2. Run the worker:

./runqy-worker -config config.yml

The worker connects to the server, pulls your code from Git, installs dependencies, and starts processing tasks.

See Configuration Reference for all options.

See Also

• runqy — Central server with CLI and monitoring dashboard
• runqy-python — Python SDK (runqy-task)
• Documentation — Full documentation

License

MIT License — see LICENSE for details.