Horizon: Metrics and Dashboards

Make queues observable and actionable via Horizon.

Naming & Tags

• Use named queues (e.g., high, default, low); route jobs intentionally
• Tag jobs with domain identifiers (user ID, aggregate ID) for tracing

Workers & Concurrency

• Right‑size --max-time, --tries, --backoff
• Separate critical queues into dedicated supervisors

Failures

• Use failed() method/logging with structured context
• Idempotency—ensure safe retries (unique keys, upserts, guards)