- Update Makefile with native build targets (preparing for C++) - Add profiler and performance regression detector commands - Update CI/testing scripts - Add additional unit tests for API, jupyter, queue, manifest
2 KiB
2 KiB
Scripts Directory
This directory contains setup and utility scripts for FetchML.
Production Scripts
setup-prod.sh
Purpose: Automated production setup for Rocky Linux bare metal deployment
Usage: sudo ./scripts/setup-prod.sh [base_path] [user] [group]
What it does:
- Creates system user and groups
- Sets up directory structure (
/data/ml-experiments/*) - Installs dependencies (Go, Podman, Redis)
- Configures GPU support for Podman
- Creates systemd service files
- Sets up log rotation
Example:
sudo ./scripts/setup-prod.sh /data/ml-experiments ml-user ml-group
Configuration validation
Validate configs using the built-in config lint targets:
make configlint
make worker-configlint
Cleanup Recommendation
These legacy scripts can be removed or archived. The current production setup only needs:
setup-prod.sh
Usage Workflow
First-Time Production Setup
# 1. Run production setup
sudo ./scripts/setup-prod.sh
# 2. Copy and configure
sudo cp configs/api/prod.yaml /etc/fetch_ml/config.yaml
sudo cp configs/workers/worker-prod.toml /etc/fetch_ml/worker.toml
sudo vim /etc/fetch_ml/config.yaml # Update API keys, etc.
# 3. Build and install
make prod
sudo make install
# 4. Validate
./bin/configlint --schema configs/schema/api_server_config.yaml /etc/fetch_ml/config.yaml
./bin/configlint --schema configs/schema/worker_config_schema.yaml /etc/fetch_ml/worker.toml
# 5. Start services
sudo systemctl start fetchml-api fetchml-worker
sudo systemctl enable fetchml-api fetchml-worker
Development Setup (macOS)
# Use docker-compose for local development
docker-compose up -d
# Or run components directly
make dev
./bin/api-server -config configs/api/dev.yaml
Script Maintenance
When adding new scripts:
- Add executable permission:
chmod +x scripts/new-script.sh - Add header comment with purpose and usage
- Update this README
- Use consistent error handling and logging