Skip to main content
GPU Cloud provides alerting capabilities to help you identify resource or service anomalies and send anomaly messages to specified channels through the Notifier. You can use Preset Rules to monitor common anomalies or use PromQL to create Custom Rules. After accessing the Alerts page, you can view custom rules, preset rules, notification channels, and the number of alert events in the past 15 days. You can also configure rules, query events, and manage notifications on the Custom Rules, Preset Rules, History, and Notifier tabs. You can also access the OBSERVABILITY > Alerts page from the CKS cluster console to view or configure alerts related to cluster O&M.

​
Usage workflow

When using alerts for the first time, we recommend following the steps below for configuration and troubleshooting:
  1. Create a notifier to receive alert messages.
  2. Select an alert rule based on the monitoring scenario.
  3. After enabling the rule, confirm that it is working as expected and is associated with the correct Notifier.
  4. After an alert is triggered, view the event status, trigger value, and tag information on the History tab, then return to the relevant resource or monitoring metric to troubleshoot the issue. Alerts homepage

​
Alert Rule Types

GPU Cloud supports the following two types of alert rules:

​
Alert Event Status

On the History tab, you can view the current status of alerts. Alert statuses include:

​
Alarm Severity

Custom Rules and historical events can use alert severity to distinguish event severity. Alert severity levels include CRITICAL, WARNING, and INFO.

​
Alarm Handling Recommendations

After receiving an alert, it is recommended that you first view the rule name, trigger time, duration, trigger value, and label information on the History tab. Labels may include information related to the product, region, cluster, resource, or rule, which can help you return to the relevant resource or monitoring metric for further troubleshooting.