runpod-scheduler / README.md

RunPod Scheduler

Automated RunPod scheduler that manages GPU pods for Siraaj instances.

Last updated: 4/16/2026GitHub

RunPod Scheduler

Automated scheduler for RunPod GPU instances with Siraaj API integration. Creates/terminates pods on schedule and updates Siraaj instances with dynamic RunPod URLs.

Problem Solved

RunPod assigns new IPs/ports daily (e.g., 213.173.105.7:30205). This service automatically updates your Siraaj instances via REST API when pods get new addresses, eliminating manual configuration updates.

Key Features

  • ✅ Automated Pod Management: Schedule daily start/stop to save costs
  • ✅ Dynamic IP/Port Handling: Automatically extracts new RunPod endpoints using GraphQL
  • ✅ Siraaj API Integration: Updates LLM providers and AI Meet configs via REST API
  • ✅ Safe Pod Termination: Only terminates your configured pods
  • ✅ Configuration Persistence: --stop preserves Siraaj configs for next --start

Quick Start

# Install dependencies
uv sync

# Configure environment
cp .env.example .env
# Edit .env with RunPod API key and Siraaj instance credentials

# Configure your models
# Edit config/pods_config.json with your pod specifications

# Test manually
python -m src.main --start    # Create pods and update Siraaj
python -m src.main --stop     # Terminate pods (keeps Siraaj configs)
python -m src.main --status   # Check status and current configurations

# Run scheduled daemon (starts pods at 6 AM, stops at 6 PM daily)
python -m src.main --daemon

Architecture

RunPod API ←→ RunPod Manager ←→ Scheduler ←→ Config Manager ←→ Siraaj API
     ↓              ↓              ↓              ↓              ↓
  Create/Stop    IP/Port      Scheduling     REST API      LLM Providers
    Pods        Extraction    (Cron Jobs)    Updates      AI Meet Configs

Configuration Files

  • .env - API keys, scheduling times, and Siraaj instance credentials
  • config/pods_config.json - Pod specifications with Siraaj integration settings

Docker Deployment

# Build and run with docker-compose
docker-compose up -d

# Or build manually
docker build -t runpod-scheduler .
docker run -d --env-file .env runpod-scheduler --daemon

Usage Examples

# Development workflow
python -m src.main --start     # Start pods for development
python -m src.main --status    # Check if pods are ready
# ... do your ML work ...
python -m src.main --stop      # Stop pods to save money

# Production deployment
python -m src.main --daemon    # Run continuously with scheduled start/stop

For detailed setup, configuration, and troubleshooting, see ML_ENGINEER_GUIDE.md.