Alerts
Every time an automation rule fires, ANTS Platform records an alert. The Alerts page lists them so the people who were told can follow each one up: acknowledge it while they work on it and resolve it when it is fixed. A recovery event, such as an agent sending traces again, resolves its alert on its own.
Open Alerts in the Governance group of the sidebar (/project/<projectId>/alerts). Emails, in-app notifications and escalations link to the alert.
The Alerts page
Three counters at the top show the number of Open, Acknowledged and Resolved alerts. Select one to show only that state.
- Search with Search alerts, and filter by State and Severity. The list starts with open and acknowledged alerts. Clear goes back to that view.
- Columns: Alert (the alert title and the rule that fired it), Severity, State, Count (how many times it fired) and Last seen.
- Select an alert to open its details.
| Detail section | What it shows |
|---|---|
| Details | Rule, Event, About (the agent, key or other subject), First seen, Last seen, Acknowledged, Escalates or Escalated, Resolved (with who resolved it, or System), and Source, a link to the page about the subject. An escalated alert also has an Escalated badge. |
| Notified | The people the alert was sent to. |
| Deliveries | Each action's delivery for this alert (Sent, Acknowledge or Resolve) and its status. |
| Related audit entries | The newest 20 audit log entries for the same agent, key, credential or rule. Shown to members who can read the audit log. |
How alerts open and close
- One alert per incident. An alert belongs to one rule and one incident, for example one silent agent or one failing export. When the same incident fires again while the alert is not resolved, the alert's count and Last seen are updated instead of a new alert appearing.
- Repeats while acknowledged. An acknowledged alert stays acknowledged when it repeats; its count goes up.
- After resolving. Resolved is final. The next event for the same incident opens a new alert.
- Recovery. Events with a recovery event, such as Agent stopped sending traces, Agent error rate above limit or Model or provider failure, notify once per incident. The recovery event resolves the alert, shown as resolved by System. If such an alert has not repeated for 24 hours, the next repeat closes it and opens a new alert, which notifies again.
- Expiry. An alert for an event without a recovery event is resolved automatically after 7 days without a repeat, unless its severity is High or Critical. High and Critical alerts stay open until someone resolves them.
- Delivery outages. An alert opens even when every action of the rule fails. Recovery events, digests and Send test never open alerts.
Acknowledge and resolve
Open an alert and select Acknowledge (only from Open) or Resolve (from Open or Acknowledged). Each change is recorded with who did it and when, and written to the audit log as automationAlert with the action acknowledge or resolve.
The change is also sent to incidents that the rule's actions opened for this alert:
| Tool | Acknowledge | Resolve |
|---|---|---|
| PagerDuty | Acknowledges the incident | Resolves the incident |
| ServiceNow | Moves the incident to In Progress with a work note | Resolves the incident |
| LogicMonitor | Nothing | Sends a clear |
A resolve by a notified person who does not have automations edit access closes the alert in ANTS Platform only. Updates go one way, from ANTS Platform to the incident tool: closing the incident in PagerDuty or ServiceNow does not close the alert.
Escalation
A rule can escalate its alerts: turn on Escalate in the rule's Alert step, set After (hours) (1 to 168, default 6) and pick the recipients. If the alert is still Open after that time, they get one email and one in-app notification titled "Escalated: " and the alert title, with how long it has been open. Acknowledging or resolving the alert before then stops the escalation. Escalations are checked every minute. See Alert settings.
Who can see and act on alerts
| Who | Sees | Can acknowledge and resolve |
|---|---|---|
| Members with Automations read access (Owner, Department Admin, Security & Compliance, Cost Analyst and Auditor) | Every alert in the project | Only alerts they were notified about |
| Members with Automations write access (Owner and Security & Compliance) | Every alert in the project | Every alert |
| Other members who were notified | The alerts they were notified about | Those alerts |
Members who were neither notified nor have Automations access do not see alerts.
Retention
Resolved alerts are deleted 30 days after they were resolved, together with the rule's runs. Open and acknowledged alerts are kept.
Next steps
- Automations & Webhooks: the rules that fire alerts.
- Alerting & Notifications: what can notify your team.
- Notifications: the bell.