kraken-cli deployment automation vs PM2 process management for production health monitoring

Deployment Comparison

To preserve KrakenJS configuration-driven isolation and folder structure loading, PM2 is the recommended approach for production. While the Kraken CLI provides rapid scaffolding, it does not offer the granular health monitoring, automatic restart policies, or resource tracking required for production stability.

Configuration Loading Behavior

KrakenJS determines which configuration files to load from the config/ directory based on the NODE_ENV environment variable. This behavior is internal to the framework and remains identical regardless of whether you use the CLI or an external manager. For example, if NODE_ENV=production, Kraken merges config/all.json with config/production.json.

Comparison Table: CLI vs. PM2

Feature	Kraken CLI (Built-in)	PM2 (External)
Config Loading	Automatic via kraken start	Manual via --env production
Health Monitoring	Basic process launch	Real-time CPU/RAM/Log metrics
Auto-Restart	Limited crash recovery	Advanced (Max restarts, delay)
Folder Guarantees	Preserved	Preserved (if env is set)

Implementation Steps for PM2

To maintain strict folder structure loading and environment isolation using PM2, follow these steps:

1. Define an Ecosystem File: Create a ecosystem.config.js to ensure environment variables are locked.
2. Configure Environment: Set env_production to include NODE_ENV: 'production'.
3. Launch the Application:

   pm2 start server.js --env production

4. Verify Loading: Check the application logs to confirm that the production configuration was merged and that the internal dependency injection is resolving production-specific services.

Assumptions and Verification

This recommendation assumes the use of a standard KrakenJS project structure where configuration is centralized in the config/ folder. To verify that PM2 is respecting your environment isolation, add a unique key to config/production.json and verify its value via a health check endpoint or the application console upon startup.

Missing Diagnostic: Are you utilizing a containerized environment (e.g., Docker) or deploying directly to a VM? If using Docker, the process management responsibilities shift toward the orchestrator (Kubernetes/ECS), making PM2 redundant.