Alerts

Stay ahead of stalls, spikes, and misbehaving queues with actionable alerts.

Alert types

• Failed jobs above a threshold
• Queue depth/backlog
• Slow processing time
• Error rate spikes

Channels

• Email
• Slack
• Webhook
• PagerDuty (Enterprise)

Incident lifecycle

• Alerts show whether they are healthy, currently firing, or have not been checked recently
• Each incident sends one trigger notification instead of repeating on every evaluation
• A recovery notification is sent when the queue returns to normal
• Cooldowns can delay a rapidly re-firing incident without pausing health checks

Create an alert

1. Go to Alerts in your project
2. Click Create alert
3. Pick a type and threshold (e.g., “>5 failed in 10 min”)
4. Choose channels and save

Recommendations

• Start with failure + backlog alerts on production queues
• Send low-severity alerts to Slack; reserve Enterprise PagerDuty escalation for sustained failures
• Review thresholds weekly after seeing real traffic