auralis-worker 0.2.5

Auralis Worker Node - Distributed AI Compute Platform

🌌 Auralis Worker Node

The worker node polls for pending jobs, executes them in Docker containers, and streams logs back to the dashboard.

Prerequisites

• Python 3.10+
• Docker installed and running
• AWS credentials configured
• Supabase credentials configured

Setup

1. Install dependencies:

   pip install -r requirements.txt
   

2. Configure environment variables: Create a .env file in the worker directory:

   # Supabase
   SUPABASE_URL=your-supabase-url
   SUPABASE_SERVICE_ROLE_KEY=your-service-role-key
   
   # AWS S3
   AWS_ACCESS_KEY_ID=your-aws-key
   AWS_SECRET_ACCESS_KEY=your-aws-secret
   AWS_REGION=us-east-1
   S3_BUCKET_NAME=auralis-job
   
   # Worker Config
   WORKER_ID=my-local-worker
   COMPUTE_TYPE=cpu
   POLL_INTERVAL=10
   WORK_DIR=/tmp/auralis-worker
   

Usage

Auto-Poll Mode (Default)

Continuously polls for pending jobs and executes them:

python main.py

Run Specific Job

Execute a specific job by ID:

python main.py --job-id abc123-def456

Options

-j, --job-id    Run a specific job by ID
-w, --worker-id Unique worker identifier (default: worker-<pid>)

How It Works

1. Poll: Worker checks Supabase for PENDING jobs
2. Claim: Worker atomically claims a job (PENDING → CLAIMED)
3. Download: Downloads project zip from S3
4. Build: Builds Docker image from Dockerfile
5. Run: Executes container with output volume mounted
6. Stream: Logs are streamed to Supabase in real-time
7. Upload: Output files are uploaded to S3
8. Complete: Job marked as COMPLETED or FAILED

Architecture

worker/
├── main.py           # Entry point, polling loop
├── job_runner.py     # Docker build & run logic
├── supabase_client.py # Job queries & log streaming
├── s3_client.py      # File upload/download
└── requirements.txt  # Dependencies