Heartbeat Service (Node.js)

Tracks process heartbeats published on the message broker and terminates those that fail to report within a timeout. Backed by MongoDB for storage. Intended for detecting and cleaning up hung or orphaned worker processes. Also enforces the instance limits defined in a PM2 ecosystem file, rejecting registrations that exceed the configured count for a given app name. Each heartbeat updates CPU, memory, and network byte counts for the process based on its PID. Heartbeats are tagged with a service-instance session ID so that restarts do not conflict with stale database entries. On shutdown the service marks all heartbeats from its current session as killed to allow clean restarts.

API

• Broker topic heartbeat
  • Services publish { pid: number, name: string } messages to this topic.
  • Heartbeats exceeding instance limits are ignored.

Environment

• MONGO_URL (default mongodb://127.0.0.1:27017)
• DB_NAME (default heartbeat_db)
• HEARTBEAT_TIMEOUT milliseconds before a process is considered stale (default 10000)
• CHECK_INTERVAL monitor interval in milliseconds (default 5000)
• ECOSYSTEM_CONFIG path to a PM2 ecosystem config file; defaults to ../../system/daemons/ecosystem.config.js
• BROKER_URL WebSocket URL of the message broker (default ws://127.0.0.1:7000)

Development

pnpm install
pnpm test

⚠️ Killing processes requires appropriate permissions. Use with care.