Alert Episodes
An alert episode is the period a monitor spends outside Healthy. It opens at the first crossing and closes on recovery. Healthy → Warning → Alerting → Warning → Healthy is one episode.
An episode records the opening time and value, escalation, peak state, recovery time, acknowledgements and each notification. A default monitor opens one episode per contract; its episode records the reading it was judged on and the transactions behind it.
Status
| Status | |
|---|---|
| Open | The value is outside the rule. The tier says whether it is past the warning or the alert threshold. |
| Recovered | The value returned to the healthy range. |
Episodes recover when the value does. There is no manual resolve.
Notifications
| Event | Sent |
|---|---|
| Crosses the warning threshold | Warning |
| Crosses the alert threshold | Alert. One message, whether or not a warning preceded it. |
| Alerting and unacknowledged | Reminder, hourly, up to eight messages in all. |
| Returns to healthy | Recovered, unless the monitor is snoozed or disabled at the time. |
Warnings do not remind. After a monitor pages, a new episode on it within ten minutes opens without paging. For a default monitor this covers all of its contracts.
A control change is notified once. It is never reminded, and no recovery message is sent when its episode closes.
Acknowledge
Acknowledging stops reminders. It does not change the monitor’s state; the recovery notification is still sent, and an escalation is sent even after its warning was acknowledged.
Acknowledge from History. CLI: contract.dev incidents ack <id>; unack resumes reminders.
Where episodes appear
- History under Monitoring: every episode in the window, with state, tier and start. An episode that names a transaction opens it in the explorer.
- A custom monitor’s page: episodes shaded on the chart and listed beneath it.
- A default monitor’s page: the state of each contract it covers, with the open episode’s transaction.
- The rail on Overview: open episodes, and the ones recovered today.
contract.dev incidents show <id>: the evidence at each crossing, each delivery and its status, and who acknowledged.
Delivery
Failed deliveries are retried with backoff, up to eight attempts. Each destination has a Test under Monitoring → Destinations.