Skip to main content
Alerts are used to monitor business events and operational metrics on the platform and send alert messages to specified recipients. The alert module consists of Notification Management, Alarm Subscription, and Alarm Overview.

Alert Delivery Chain

For initial configuration, typically create notification targets first, then create alert subscriptions. After an alert is triggered, you can view delivery records in the alarm overview.

Alert Types

Alert subscriptions support two types of alert sources:
  • Event: Triggered by platform business events, suitable for monitoring status changes in training jobs, development environments, inference services, images, orders, quotas, storage, and other objects.
  • Metric: Triggered by Prometheus metric thresholds, suitable for monitoring GPU, CPU, memory, network, or custom business metrics.