Skip to main content

Monitoring and Alerts

Monitoring should cover infrastructure, runtime quality, and business-level behavior.

Infrastructure Metrics​

  • CPU / memory / disk
  • Redis latency and errors
  • Postgres connections and slow queries
  • Container restart counts

Runtime Metrics​

  • Signals per interval
  • Order execution attempts vs successful placements
  • Runtime error rate by strategy
  • API latency for market endpoints

AI and ML Metrics​

  • AI approval rate
  • AI quality distribution
  • ML inference latency and failure rate
  • ML score distribution drift

Alerting Levels​

  • Warning: degradation trend (latency up, small error rise)
  • Critical: hard dependency down, sustained order failures, or data pipeline broken
  • Route warnings to on-call channel.
  • Escalate critical alerts with immediate rollback options.
  • Track each incident with postmortem actions.