Alerts and history
Create threshold alerts, route them, and audit acknowledge or silence events.
Alerts evaluate namespace metrics on a schedule and notify configured channels.
Create a rule
- Open Alerts in the console.
- Choose a metric, for example
system.cpu.usage_percent. - Set comparison (
>,<,>=), threshold, duration, and severity (critical,warning,info). - Optionally filter by host or other labels.
- Attach notification channels: email, Slack, or webhook. See Notifications.
- Save. The rule is evaluated only inside the current namespace.
Lifecycle
An open alert can be:
- Acknowledged — someone is working it
- Silenced — muted for a period
- Resolved — condition cleared or closed by a user
These actions are available in the console and in the mobile app.
History
Alerts → History is the audit log for a rule or the whole namespace. Each event records:
- timestamp
- action (fired, acknowledged, silenced, resolved)
- actor email when a user performed the action
- previous and new state
Use history when you need to prove who silenced a production page or how long a disk alert stayed open.
Tips
- Start with host CPU, memory, and disk. Add application counters after custom metrics exist.
- Keep one rule per symptom. Combine related symptoms on a dashboard instead of a single noisy expression.
- Plan limits cap the number of rules. Check Billing if create is disabled.